USF Libraries
USF Digital Collections

A roof runoff strategy and model for augmenting public water supply

MISSING IMAGE

Material Information

Title:
A roof runoff strategy and model for augmenting public water supply
Physical Description:
Book
Language:
English
Creator:
Carnahan, Robert
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
: water resources
Risks
Roofing material
Water quality
Metals
Dissertations, Academic -- Civil & Environmental Engineering -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Water is the essential resource that is becoming extremely scarce worldwide. The 21st century will further stress all available water resources through the growth and expansion of developing nations. It is not only the quantity of cheap water that is being depleted, but the quality of these waters is being endangered. Florida is an example where rapid development and an exploding population are competing for shrinking groundwater resources. Current water use does not address the use of alternative supplies and reuses in the United States. The objective of this research was to determine a strategy for augmenting existing water supplies with alternative sources that could be developed economically. Having reviewed numerous alternative sources, it was determined that runoff from roofs potentially provides a source that might meet the augmentation requirement for a small community of a population of 30,000 or less. This research has shown that the quality of water collected from five different roof surfaces meets the drinking water standards and will not degrade the current quality of the main source of water supply. This work not only required the collection of hydrological data from the roof systems, but chemically and biological analyzes samples for contaminants. Since rainfall events vary periodically and in duration, 100,000 meteorological events were analyzed for wind speed, relative humidity, rainfall intensity, and the rainwater runoff across five roofing surfaces to analyze variables that contribute to the effects on the water quality of the source. The model establishes the economics and the public health value of this water. The research assesses the local regulatory aspects of using the water with the outcome of a working objective and rational decision matrix that will permit agencies to select an optimal and safe utilization of the water sources.
Thesis:
Dissertation (PHD)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Robert Carnahan.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0004622
usfldc handle - e14.4622
System ID:
SFS0027937:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004622
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Carnahan, Robert.
0 245
A roof runoff strategy and model for augmenting public water supply
h [electronic resource] /
by Robert Carnahan.
260
[Tampa, Fla] :
b University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
502
Dissertation (PHD)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: Water is the essential resource that is becoming extremely scarce worldwide. The 21st century will further stress all available water resources through the growth and expansion of developing nations. It is not only the quantity of cheap water that is being depleted, but the quality of these waters is being endangered. Florida is an example where rapid development and an exploding population are competing for shrinking groundwater resources. Current water use does not address the use of alternative supplies and reuses in the United States. The objective of this research was to determine a strategy for augmenting existing water supplies with alternative sources that could be developed economically. Having reviewed numerous alternative sources, it was determined that runoff from roofs potentially provides a source that might meet the augmentation requirement for a small community of a population of 30,000 or less. This research has shown that the quality of water collected from five different roof surfaces meets the drinking water standards and will not degrade the current quality of the main source of water supply. This work not only required the collection of hydrological data from the roof systems, but chemically and biological analyzes samples for contaminants. Since rainfall events vary periodically and in duration, 100,000 meteorological events were analyzed for wind speed, relative humidity, rainfall intensity, and the rainwater runoff across five roofing surfaces to analyze variables that contribute to the effects on the water quality of the source. The model establishes the economics and the public health value of this water. The research assesses the local regulatory aspects of using the water with the outcome of a working objective and rational decision matrix that will permit agencies to select an optimal and safe utilization of the water sources.
590
Advisor: Mark Ross, Ph.D.
653
: water resources
Risks
Roofing material
Water quality
Metals
690
Dissertations, Academic
z USF
x Civil & Environmental Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4622



PAGE 1

A Roof Runoff Strategy and Model fo r Augmenting Public Water Supply by Robert P. Carnahan, Jr. A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Civil and Environmental Engineering College of Engineering University of South Florida Co-Major Professor: Mark Ross, Ph.D. Co-Major Professor: Thomas Mason, Ph.D. Mark Amen, Ph.D. William Carpenter, Ph.D. Lisa Robbins, Ph.D. Dewey Rundus, Ph.D. Date of Approval: April 28, 2010 Keywords: water resources, risks, roof ing material, water quality, metals Copyright 2010, Robert P. Carnahan, Jr.

PAGE 2

To my father and mother, to whom I owe everything.

PAGE 3

ACKNOWLEDGEMENTS I would like to thank my committee me mbers for their support throughout this research, in particular: to Dr. Robbins for her extensive review and recommendations for strengthening this research; to Dr. Thomas Mason for his inspirational comments, encouragement, insights, and continuous support; to Dr. Mark Ross for his numerous meetings and guidance with the engineering sc ience; and to Dr. Heid i Kay for the use of her laboratory, equipment, supplies, and guidance in the laboratory. This project could not have been comple ted without financial support from an inkind donation from Nick Nicholson and Dona ld Ferguson, Jr., management of Gold Sealed Roofing, for the materials used in cons truction of the roofing panels in accordance with code. Special thanks goes to Bob Sullivan for his time, guidance, expertise in the laboratory, and for championing my research. De ep gratitude and special thanks to Dale and Katharine Dixon for their large in-k ind donation of 5,364 individual chemical analyses by Benchmark Enviroanalytical La boratories, an environmental reference laboratory. I would like to acknowledge a ll those at the City of Te mple Terrace who assisted me on many occasions, especially the Planning and Public Works department. Finally, I would like to thank my family and friends for their continued support and encouragement, and special thanks to my co lleagues at Eckerd College, especially Patty Cooksey Fisher and director Margret Skaftado ttir, for their support, encouragement, and for assigning additional courses for me to teach.

PAGE 4

i TABLE OF CONTENTS LIST OF TABLES ............................................................................................................. iv LIST OF FIGURES .......................................................................................................... xii LIST OF SYMBOLS ....................................................................................................... xiv ABSTRACT .................................................................................................................... x vii CHAPTER I: INTRODUCTION .........................................................................................1 Background ..............................................................................................................1 Statement of the Problem .........................................................................................4 Problem Identification .............................................................................................5 Purpose of the Study ................................................................................................5 Methodology of the Dissertation .............................................................................8 CHAPTER II: BACKGROUND .......................................................................................11 Overview ................................................................................................................11 Meteorological Effects on Rainwater ....................................................................12 Thunderstorms ...........................................................................................12 Scavenging Effects of Rain ........................................................................14 Biogeochemical Processes .....................................................................................18 Microbial Effects on Water Quality .......................................................................19 Roof Surface ..............................................................................................19 Meteorological Influences on Micr obial Concentrations on Roofs ...........20 Cisterns ..................................................................................................................21 Cisterns Founded in the South Pacific .......................................................22 Cistern Founded in the U.S. and British Virgin Islands ............................23 Cisterns Founded in the Greece Islands .....................................................25 Process That Changes Rainwater pH .....................................................................26 Public Policy for Drinking Water ..........................................................................30 Federal Regulations for Drinking Water in the United States ...................30 State of Texas Regulations for Rainwater Harvesting ...................30 State of Florida Regulations for Rainwater Harvesting .................31 Regulations for Drinking Water in the Western United States ......32 Regulations for Drinking Wa ter in Other Countries ..................................33 Stormwater Compared to Roof Runoff ..................................................................34 CHAPTER III: METHODOLOGY ...................................................................................36 Material Selection and Description ........................................................................36

PAGE 5

ii Roof Panels ................................................................................................36 Control or Reference Sample .................................................................................37 Instrumentation ......................................................................................................37 Field Sample Instrumentation ....................................................................37 Biological Instrumentation.........................................................................38 Meteorological Instrumentation .................................................................40 Analytical Instrumentation.........................................................................42 Data Collection Procedures ....................................................................................44 Preliminary Testing ....................................................................................44 Primary Meteorological Data Collection ...................................................44 Primary Water Data Collection ..................................................................45 Data Analysis .........................................................................................................47 Meteorological Data Analysis....................................................................47 Antecedent Historical Analysis..................................................................48 Rainfall Data Records for the Model .........................................................48 Experimental Design ..................................................................................49 Sample Size ................................................................................................49 Data Exclusion Criteria ..............................................................................50 CHAPTER IV: RESULTS .................................................................................................52 Regression Model and Transf ormation of the Data ...............................................68 Correlation Matrix .................................................................................................68 Descriptive Statistical Analysis .............................................................................70 Paired T-Test ..............................................................................................70 Wilcoxon Signed Rank Test ..................................................................................75 CHAPTER V: WATER QUALITY RESULTS AND DISCUSSION ..............................76 Summary of Chemical Anal ysis of Roof Runoff ...................................................88 Item Analysis .............................................................................................89 Wind Direction Analysis............................................................................91 Limitations of the Study and the Data ...................................................................99 CHAPTER VI: MODEL DEVELO PMENT AND DISCUSSION .................................102 The Design and Processes Used in the Model Development ..............................104 Water Quality Parameter in the Model Development ..............................105 Meteorological Parameters in the Model Development ..........................105 Demographic Parameters of the Model Development .............................107 Geographic and Demographics Conditions on the Model ...................................107 Sector Concept for the Feasibility and Viability of the Model ................108 Demographics Conditions ............................................................108 Geographic Conditions ................................................................109 The Configurations and Hydraulic Cond itions Incorporated into the Model ......110 Conceptual Description of the Configuration Design ..............................110 Hydraulic Conditions ...............................................................................111 The Model Data Modules for the Input and Output Screens ...............................112

PAGE 6

iii Water Data Module ..................................................................................112 Meteorological Data Module ...................................................................112 Geographical and Demographic Module .................................................113 The Model Elements for the Variable s Sheet Input and Output Screens ............113 User Input Variables ................................................................................115 User Input Constraints .............................................................................115 Variable Input Screen Output ..................................................................117 The Model Inputs Elements for the Sector Collection and Piping System ..........117 Model Generated Results as Tables and Graphs ..................................................120 Variable Outcome Output ........................................................................122 Output of the Utilization Resu lts as Tables and Graphs ..........................124 Output of Frequencies of Rain .................................................................124 The Model as a Feasibility T ool for Alternative Sources ....................................127 The System Advantages Contra sted to Individual Units .........................129 Regulatory and Policy Issues ...................................................................131 The Augmentation System Compar ed to Aquifer Storage and Recovery ............................................................................................131 CHAPTER VII: CONCLUSIONS ...................................................................................134 CHAPTER VIII: RECOMMENDATIONS .....................................................................135 REFERENCES ................................................................................................................136 APPENDICES .................................................................................................................146 Appendix I: Standards and Analysis ....................................................................147 Appendix II: Model Development .......................................................................207 ABOUT THE AUTHOR ....................................................................................... End Page

PAGE 7

iv LIST OF TABLES Table 1-1: Seasonal Temperature and Precipitation for Tampa Florida ............................2 Table 2-1: Global Mobilization Fact ors Based on Annual Emission Rates. ...................15 Table 2-2: Concentrations of Metals Ranges Found in Wet Deposition .........................16 Table 2-3: pH Effects on Solubilities of Metals in Rainwater Between Locations in Turkey and Mexico ....................................................................................29 Table 2-4: Comparisons of Concentrat ions from Rainfall and Roof Runoff ...................35 Table 4-1: Preliminary Data Results for Samples at the Site Using the AA ...................54 Table 4-2: Roof Runo ff Concentrations Summary Results from the Site .......................55 Table 4-3: Roof Runoff Analyzed for pH Analyses from the Site ..................................56 Table 4-4: Roof Runoff Analyzed for Alkalinity from the Site .......................................57 Table 4-5: Roof Runoff Analyzed for Total Dissolved Solids at the Site .......................58 Table 4-6: Roof Runoff Analy zed for Zinc from the Site ...............................................59 Table 4-7: Roof Runoff Analy zed for Lead from the Site ...............................................60 Table 4-8: Roof Runoff Analy zed for Cadmium from the Site .......................................61 Table 4-9: Roof Runoff Analy zed for Nickel from the Site ............................................62 Table 4-10: Roof Runoff Analy zed for Iron from the Site ................................................63 Table 4-11: Roof Runoff Analyzed for Manganese from the Site .....................................64 Table 4-12: Roof Runoff Analyzed for Chromium from the Site .....................................65 Table 4-13: Roof Runoff Analyzed for Copper from the Site ...........................................66 Table 4-14: Roof Runoff Analyzed for Magnesium from the Site ....................................67

PAGE 8

v Table 4-15: Roof Surfaces and Variables' Effect on the Copper Correlation ....................69 Table 4-16: Copper Paired T-Test S1 and Control ............................................................74 Table 4-17: Copper Wilcoxon S2-Control .........................................................................75 Table 5-1: The Zinc Concentratio ns Analyses of the Roof Runoff mg -1 at the Site ..................................................................................................................77 Table 5-2: A Comparison of pH Levels at the Site ..........................................................78 Table 5-3: Comparisons of Total Dissolved Solids mg -1, Levels at the Site .................79 Table 5-4: Comparisons of Chromium mg -1, Levels at the Site ....................................81 Table 5-5: Comp arisons of Copper mg -1, Levels at the Site ..........................................81 Table 5-6: Compar isons of Magnesium mg -1, Levels at the Site ...................................82 Table 5-7: Roof Surfaces and Vari ables' Effect on the pH Correlation ...........................82 Table 5-8: Roof Surfaces and Variab les' Effect on the Total Dissolved Solids Correlation ......................................................................................................83 Table 5-9: Roof Surfaces and Vari ables' Effect on the Zinc Correlation ........................83 Table 5-10: Roof Surfaces and Variab les' Effect on the Lead Correlation........................84 Table 5-11: Roof Surfaces and Variables' Effect on the Cadmium Correlation ................84 Table 5-12: Roof Surfaces and Variables' Effect on the Nickel Correlation .....................85 Table 5-13: Roof Surfaces and Variables' Effect on the Manganese Correlation .............85 Table 5-14: Roof Surfaces and Variables' Effect on the Chromium Correlation ..............86 Table 5-15: Roof Surfaces and Variables' Effect on the Copper Correlation ....................86 Table 5-16: Roof Surfaces and Variables' Effect on the Magnesium Correlation .............87 Table 5-17: Roof Surfaces and Variab les' Effect on the Iron Correlation .........................87 Table 5-18: Summary of Statis tical Analyses of the Surf ace Runoff Data for S1-S5 ..............................................................................................................90

PAGE 9

vi Table 5-19: The Southern Wind Effect s on Concentration of the Control ........................93 Table 5-20: The Northern Wind Effects on Concentrations of the Control ......................94 Table 5-21: Effects of Changes in West by Southwestern Wind Direction on Plate Counts ............................................................................................................95 Table 5-22: Effects of Cha nges in East by Southeastern Wind Direction on Plate Counts .............................................................................................................95 Table 5-23: Effects of Cha nges in Southern Wind Direction on Plate Counts ..................96 Table 5-24: Effects of Changes in North Wind Direction on Plate Counts .......................97 Table 5-25: Effects of Cha nges in Southwestern Wind Direction on Plate Counts ..........98 Table 6-1: Model Output Frequenc y of the Piping and Rain Events .............................125 Table 6-2: Summary of th e Sector Piping Constraints ..................................................126 Table 6-3: Data Rain Records Annual Rainfall .............................................................126 Table A-1: pH Descriptive Statistic al Analysis Material Surfaces S1-S5 .....................147 Table A-2: pH Paired T-Te st Analysis S1 and Control .................................................149 Table A-3: pH Paired T-Te st Analysis S2 and Control .................................................149 Table A-4: pH Paired T-Te st Analysis S3 and Control .................................................150 Table A-5: pH Paired T-Te st Analysis S4 and Control .................................................150 Table A-6: pH Paired T-Te st Analysis S5 and Control .................................................151 Table A-7: pH Wilcoxon Analysis S1-Control ..............................................................151 Table A-8: pH Wilcoxon Analysis S2-Control ..............................................................152 Table A-9: pH Wilcoxon Analysis S3-Control ..............................................................152 Table A-10: pH Wilcoxon Analysis S4-Control ..............................................................152 Table A-11: pH Wilcoxon Analysis S5-Control ..............................................................153 Table A-12: TDS Descriptive Statistical Analysis Material Surfaces S1-S5 ..................153

PAGE 10

vii Table A-13: TDS Paired T-Test Analysis S1 and Control ..............................................155 Table A-14: TDS Paired T-Test Analysis S2 and Control ..............................................155 Table A-15: TDS Paired T-Test Analysis S3 and Control ..............................................156 Table A-16: TDS Paired T-Test Analysis S4 and Control ..............................................156 Table A-17: TDS Paired T-Test Analysis S5 and Control ..............................................157 Table A-18: TDS Wilcoxon Analysis S1-Control ...........................................................157 Table A-19: TDS Wilcoxon Analysis S2-Control ...........................................................158 Table A-20: TDS Wilcoxon Analysis S3-Control ...........................................................158 Table A-21: TDS Wilcoxon Analysis S4-Control ...........................................................158 Table A-22: TDS Wilcoxon Analysis S5-Control ...........................................................159 Table A-23: Zinc Descriptive Statistical Analysis Material Surfaces S1-S5 ..................159 Table A-24: Zinc Paired T-Test Analysis S1 and Control ...............................................161 Table A-25: Zinc Paired T-Test Analysis S2 and Control ...............................................161 Table A-26: Zinc Paired T-Test Analysis S3 and Control ...............................................162 Table A-27: Zinc Paired T-Test Analysis S4 and Control ...............................................162 Table A-28: Zinc Paired T-Test Analysis S5 and Control ...............................................163 Table A-29: Zinc Wilcoxon Analysis S1-Control ...........................................................163 Table A-30: Zinc Wilcoxon Analysis S2-Control ...........................................................163 Table A-31: Zinc Wilcoxon Analysis S3-Control ...........................................................164 Table A-32: Zinc Wilcoxon Analysis S4-Control ...........................................................164 Table A-33: Zinc Wilcoxon Analysis S5-Control ...........................................................164 Table A-34: Lead Descriptive Statistical Analysis Material Surfaces S1-S5 ..................165 Table A-35: Lead T-Test S1 Analysis and Control .........................................................167

PAGE 11

viii Table A-36: Lead Paired T-Test Analysis S2 and Control ..............................................167 Table A-37: Lead Paired T-Test Analysis S3 and Control ..............................................168 Table A-38: Lead Paired T-Test Analysis S4 and Control ..............................................168 Table A-39: Lead Paired T-Test Analysis S5 and Control ..............................................169 Table A-40: Lead Wilcoxon Analysis S1-Control ...........................................................169 Table A-41: Lead Wilcoxon Analysis S2-Control ...........................................................169 Table A-42: Lead Wilcoxon Analysis S3-Control ...........................................................170 Table A-43: Lead Wilcoxon Analysis S4-Control ...........................................................170 Table A-44: Lead Wilcoxon Analysis S5-Control ...........................................................170 Table A-45: Cadmium Descriptive Statistic al Analysis Material Surfaces S1-S5 ..........171 Table A-46: Cadmium Paired T-Te st Analysis S1 and Control ......................................173 Table A-47: Cadmium Paired T-Te st Analysis S2 and Control ......................................173 Table A-48: Cadmium Paired T-Te st Analysis S3 and Control ......................................174 Table A-49: Cadmium Paired T-Te st Analysis S4 and Control ......................................174 Table A-50: Cadmium Paired T-Te st Analysis S5 and Control ......................................175 Table A-51: Cadmium Wilc oxon Analysis S1-Control ...................................................175 Table A-52: Cadmium Wilc oxon Analysis S2-Control ...................................................175 Table A-53: Cadmium Wilc oxon Analysis S3-Control ...................................................176 Table A-54: Cadmium Wilc oxon Analysis S4-Control ...................................................176 Table A-55: Cadmium Wilc oxon Analysis S5-Control ...................................................176 Table A-56: Nickel Descript ive Statistical Analysis Material Surfaces S1-S5 ...............177 Table A-57: Nickel Paired T-Te st Analysis S1 and Control ...........................................179 Table A-58: Nickel Paired T-Te st Analysis S2 and Control ...........................................179

PAGE 12

ix Table A-59: Nickel Paired T-Te st Analysis S3 and Control ...........................................180 Table A-60: Nickel Paired T-Te st Analysis S4 and Control ...........................................180 Table A-61: Nickel Paired T-Te st Analysis S5 and Control ...........................................181 Table A-62: Nickel Wilc oxon Analysis S1-Control ........................................................181 Table A-63: Nickel Wilc oxon Analysis S2-Control ........................................................181 Table A-64: Nickel Wilc oxon Analysis S3-Control ........................................................182 Table A-65: Nickel Wilc oxon Analysis S4-Control ........................................................182 Table A-66: Nickel Wilc oxon Analysis S5-Control ........................................................182 Table A-67: Iron Descriptive Statistica l Analysis Material Surfaces S1-S5 ...................183 Table A-68: Iron Paired T-Test Analysis S1 and Control ...............................................185 Table A-69: Iron Paired T-Test Analysis S2 and Control ...............................................185 Table A-70: Iron Paired T-Test Analysis S3 and Control ...............................................186 Table A-71: Iron Paired T-Test Analysis S4 and Control ...............................................186 Table A-72: Iron Paired T-Test Analysis S5 and Control ...............................................187 Table A-73: Iron Wilcoxon Analysis S1-Control ............................................................187 Table A-74: Iron Wilcoxon Analysis S2-Control ............................................................187 Table A-75: Iron Wilcoxon Analysis S3-Control ............................................................188 Table A-76: Iron Wilcoxon Analysis S4-Control ............................................................188 Table A-77: Iron Wilcoxon Analysis S5-Control ............................................................188 Table A-78: Manganese Descriptive Statisti cal Analysis Material Surfaces S1-S5 ........189 Table A-79: Manganese Paired T-Te st Analysis S1 and Control ....................................191 Table A-80: Manganese Paired T-Te st Analysis S2 and Control ....................................191 Table A-81: Manganese Paired T-Te st Analysis S3 and Control ....................................192

PAGE 13

x Table A-82: Manganese Paired T-Te st Analysis S4 and Control ....................................192 Table A-83: Manganese Paired T-Te st Analysis S5 and Control ....................................193 Table A-84: Manganese Wilc oxon Analysis S1-Control ................................................193 Table A-85: Manganese Wilc oxon Analysis S2-Control ................................................193 Table A-86: Manganese Wilc oxon Analysis S3-Control ................................................194 Table A-87: Manganese Wilc oxon Analysis S4-Control ................................................194 Table A-88: Manganese Wilc oxon Analysis S5-Control ................................................194 Table A-89: Chromium Descriptive Statisti cal Analysis Material Surfaces S1-S5 ........195 Table A-90: Chromium Paired T-Te st Analysis S1 and Control .....................................197 Table A-91: Chromium Paired T-Te st Analysis S2 and Control .....................................197 Table A-92: Chromium Paired T-Te st Analysis S3 and Control .....................................198 Table A-93: Chromium Paired T-Te st Analysis S4 and Control .....................................198 Table A-94: Chromium Paired T-Te st Analysis S5 and Control .....................................199 Table A-95: Chromium Wilc oxon Analysis S1-Control .................................................199 Table A-96: Chromium Wilc oxon Analysis S2-Control .................................................199 Table A-97: Chromium Wilc oxon Analysis S3-Control .................................................200 Table A-98: Chromium Wilc oxon Analysis S4-Control .................................................200 Table A-99: Chromium Wilc oxon Analysis S5-Control .................................................200 Table A-100: Copper Descriptive Statistica l Analysis Material Surfaces S1-S5 ............201 Table A-101: Copper Paired T-Test Analysis S1 and Control ........................................203 Table A-102: Copper Paired T-Test Analysis S2 and Control ........................................203 Table A-103: Copper Paired T-Test Analysis S3 and Control ........................................204 Table A-104: Copper Paired T-Test Analysis S4 and Control ........................................204

PAGE 14

xi Table A-105: Copper Paired T-Test Analysis S5 and Control ........................................205 Table A-106: Copper Wilcoxon Analysis S1-Control .....................................................205 Table A-107: Copper Wilcoxon Analysis S2-Control .....................................................205 Table A-108: Copper TDS Analysis Wilcoxon S3-Control ............................................206 Table A-109: Copper TDS Analysis Wilcoxon S4-Control ............................................206 Table A-110: Copper TDS Analysis Wilcoxon S5-Control ............................................206 Table B-1: Model Variables for the Hydraulic Calculations ...........................................212 Table B-2: Model Cost Schedule Analysis ......................................................................214 Table B-3: Model Estimated Cost and Payback Period Analysis ....................................215 Table B-4: Model Loan Schedule Analysis .....................................................................216

PAGE 15

xii LIST OF FIGURES Figure 3-1: Photo of the A pparatus and Monitoring Station ...........................................42 Figure 4-1: Outliers’ Effect on the Lead Concentration Data ..........................................51 Figure 4-2: The Results of Removal of Outlier from the Lead Concentrations ..............51 Figure 5-1: The Prevailing Wind Directi on at the Research Site Over the Study Period ............................................................................................................92 Figure 6-1: Conceptual Inputs for Creating the Au gmentation Model Matrix ..............103 Figure 6-2: The Flow Chart of Proce sses for the Development of the Model ...............106 Figure 6-3: Temple Terrace Map Secti on of the Houses Used in the Model ................108 Figure 6-4: Variab le Input Screen ..................................................................................114 Figure 6-5: Piping Rout ing Input Page Screen ..............................................................119 Figure 6-6: The Variable Input Sheet Output Table for the Model ...............................121 Figure 6-7: The Model Graphi cal Output of the Utilization ..........................................123 Figure A-1: pH Fre quency Plot Analysis of Material Surface ......................................148 Figure A-2: TDS Frequency Pl ot Analysis of Material .................................................154 Figure A-3: Zinc Fr equency Plot Analysis of Material Surface ....................................160 Figure A-4: Lead Frequency Plot Analysis of Material Surface ...................................166 Figure A-5: Cadmium Frequency Plot Analysis of Material Surface ............................172 Figure A-6: Nickel Frequency Pl ot Analysis of Material Surface .................................178 Figure A-7: Iron Frequency Plot Analysis of Material Surface .....................................184 Figure A-8: Manganese Frequency Plot Analysis of Material Surface .........................190

PAGE 16

xiii Figure A-9: Chromium Frequency Pl ot Analysis of Material Surface ..........................196 Figure A-10: Copper Freque ncy Plot Analysis ................................................................202 Figure B-1: Rain Mode l Analyses for Demand .............................................................213

PAGE 17

xiv LIST OF SYMBOLS AAS Atomic Absorption Spectrophotometry Ancedent abbreviated form of Antecedent (aq.) Aqueous Solutions As Arsenic Avg Average Cd Cadmium Co Cobalt coef coefficient Con Control Cr Chromium Cu Copper cf cubic feet c.f.s cubic-feet-per-second (also cfs) DRRH Domestic Roof Rainwater Harvesting eff efficiency EPA (United States) Environmental Protection Agency F degrees Fahrenheit Fe Iron F.A.C Florida Administrative Code FDEP Florida Department of Environmental Protection

PAGE 18

xv FDOT Florida Department of Transportation F.S. Florida Statute g gram (g) gaseous GIS Geographic Information Systems gpm gallons per minute g/cap -day gallons capacity per day HDPE high-density polyethylene HNO3 nitric acid ICP Inductively Coupled Plasma ICP-MS Inductive Coupled Plasma-Mass Spectrometry Max maximum MCLs maximum contaminant levels MGD million gallons per day Min minimum ml millimeter Mg Magnesium mg/L miligram per liter Mn Manganese nm nanometer Ni Nickel %RSD relative standa rd deviation express as a percentage Pcpn precipitation

PAGE 19

xvi Pb Lead QA Quality Assurance Ref Reference S1 Clay Barrel tile S2 Glazed tile S3 Shaker tile S4 Painted galvanized tin S5 Galvanized tin 16” sixteen inch diameter pipe SDWA Safe Drinking Water Act SFWMD South Florida Wate r Management District SWFWMD Southwest Florida Water Management District sf/home square feet per home TDS Total Dissolved Solids USDA United States De partment of Agriculture USGS United States Geological Survey Vol Volume WMD Water Management District yr year Zn Zinc Statistically significant at .01 ** Statistically significant at .05*

PAGE 20

xvii A Roof Runoff Strategy and Model fo r Augmenting Public Water Supply Robert P. Carnahan, Jr. ABSTRACT Water is the essential resource that is becoming extremely scarce worldwide. The 21st century will further stress all available water resour ces through the growth and expansion of developing nations It is not only the quantity of cheap water that is being depleted, but the quality of these waters is being endange red. Florida is an example where rapid development and an exploding populati on are competing for shrinking groundwater resources. Current water use does not address th e use of alternative s upplies and reuses in the United States. The objective of this research was to de termine a strategy for augmenting existing water supplies with alternative sources th at could be develope d economically. Having reviewed numerous alternative sources, it was determined that runoff from roofs potentially provides a source that might meet the augmentation requirement for a small community of a population of 30,000 or less. This research has shown that the quality of water collected from five different roof surfaces meets the drinking water standa rds and will not degrade the current quality of the main source of water supply. This work not only required the collection of hydrological data from the roof systems, but chemically and biological analyzes samples for contaminants. Since rainfall events vary periodically and in duration, 100,000 meteorological events were analyzed for wind speed, relative humidity, rainfall intensity,

PAGE 21

xviii and the rainwater runoff across five roofing su rfaces to analyze vari ables that contribute to the effects on the water quality of the s ource. The model establishes the economics and the public health value of this water. The rese arch assesses the local regulatory aspects of using the water with the outcome of a worki ng objective and rational decision matrix that will permit agencies to select an optimal and safe utilization of the water sources.

PAGE 22

1 CHAPTER I: INTRODUCTION Background Water is the essence of life—the most precious resource of the 21st century. Florida is an example of a region where ra pid development and an exploding population are competing for shrinking groundwater resour ces. The Floridian peninsula is unique in that it is on the same latitude as some of the world’s major deserts, yet its average yearly rainfall is 53 inches. The most common climate classification is the Kppen, which divides the state of Florida into two clim ate types. Most of Florida has a humid subtropical climate, as at the study site, with the southern portion of th e state as a tropical savanna from approximately Ft. Pierce to Miam i to the Keys (Ferna ld & Purdem, 1998). The study site was located in a suburban ne ighborhood in West-Centr al Florida area in the City of Temple Terrace. This area is considered a humid mesothermal climate using the Thornthwaite classification system whic h uses evapotranspiration and rainfall to determine boundaries which divides the state in to three climate type s and is most often used by water resource professionals (Ferna ld & Purdem, 1998). Florida exhibits a bimodal annual rainfall pattern: the dry seas on from December through May, which has an average seasonal total pr ecipitation of 14.73 inches and average temperature of 67.3 F; and the rainy season from June through November, which has an average seasonal total precipitation of 30.04 inches and average temperature of 78.9 F (NOAA, 2005).

PAGE 23

2 Table 1-1: Seasonal Temperature and Precipitation for Tampa Florida. Pcpn inches SeasonMaxMinAvgHeating Cooling Total Winter (Dec/Jan/Feb) 71.2 53.6 62.4 4671927.24 Spring (Mar/Apr/ May) 81.163.372.2767217.49 Summer (Jun/Jul/Aug) 89.574.982.201600 19.59 Autumn (Sep/Oct/Nov) 83.767.575.64896910.45 Degree Days Temperature F Ref: National Weather Service Ruskin, Florida Table 1-1 illustrates the seasonal temp eratures for Tampa, especially the maximum temperature during the seasons and the importance of the inland heating and evaporation effect that drive the weather pa tterns of Florida. Th e degree-days provides the ability to compare different years’ seasons to each other; for example, degree-days cooling is the average daily temperature de grees F minus 65 F degrees equal the cooling days. The degree-days are accumulated each da y over the course of a heating/cooling season and can be compared to a long-term (multi-year) average, or normal, to see if that season was warmer or cooler than usual. Th e precipitation in total inches is 44.77, which is deficit from the norm of 53 inches. The highest evaporation period is duri ng the rainy season, when it ranges from 46 to 50 inches in central Florida (Fernald & Pu rdem, 1998). In the dry winter months, there is a dramatic increase in demand for water by agriculture and indus try. This seasonality of rainfall and water demand affects the water budget of local communities. In Florida, a bimodal annual rainfall pattern provided ex tremes during this investigation. Other locations throughout the country, such as As heville, North Carolina, with an annual

PAGE 24

3 rainfall of 47.7 inches, are consistent each month in the average rainfall. Likewise, Lexington, Kentucky has an average rainfall of 45.68 inches. Both locations usually have consistent precipitation with th ree (3) to four (4) inches pe r month based upon a standard 30 year period recorded between 1951 to 1980 (Leeden, et al., 1990). Florida’s significant drainage systems move the water in the rainy season. Florida’s gulf coastal lowlands are flat, wi th productive agricultural land interspersed with wetlands. The same drainage system ca rries urban runoff in a highly populated area. When the soil becomes saturated, the precipita tion exceeds the infiltra tion capacity of the soil and the soil can no longer absorb water, re ducing the amount of infiltrated water that reaches the aquifer. Instead, the overland wa ter flows commence as surface runoff, thus bypassing aquifer recharge to the system that local communities rely on for their water supply. According to the United States Geological Survey (USGS) in 2000, approximately 85 percent of th e population of the United States receives their water from a public supplier; 63 percent is from surface water sources. California and Florida public suppliers have the largest gr oundwater withdrawals (Hutson, et al., 2004). Approximately 80 percent of the water used in the Tampa Ba y region is groundwater, with coastal areas experiencing saltwater intrus ion due to over-pumping of the Floridian aquifer system (Hydrologic Evaluation Section, 2002). Florida's population grew by more than 3 million between 1990 and 2000, more than any other st ate except California and Texas. This represented a 23.5 percent increase, the sevent h largest growth rate of any state, and Florida is expected to surpass New York by around 2010 to become the nation’s third largest state. If the projec tions are correct, Florida’s popul ation is expected to reach

PAGE 25

4 almost 26 million by 2030 (Smith, 2005). Florid a is a microcosm of the scarcity of freshwater and the demands of an incr easing population grow th around the globe. To meet the ever-growing population’s water demands, local municipalities and counties in the Tampa Bay area have create d a regional water supplier, which supplies approximately 172 million-gallons-per-day to the region. Florida's population is approximately 13 million persons. Previous estim ates had Florida growing at a rate of 487 persons-per-day. This does not incl ude the influx of approximately 100,000 temporary residents to the Tampa Bay area du ring the winter months, in the same period as decreased rainfall. According to the 2003 United States Geological Survey’s “Water Use Report,” the average Florida resident us es 174 gallons-per-day for household use (USGS, 2004). The growth in population a nd high demand for water requires that new alternative water sources be utilized. Statement of the Problem Fresh water worldwide is a limited resour ce. Florida is an example where rapid development and an exploding population ar e competing for finite or shrinking groundwater resources. Current water use does not address the use of alternative supplies such as roof runoff and reuse in the Unite d States. There is a need for planning and development of alternative fresh water source s that are economically available and viable to develop, while assessing publ ic health aspects and govern ment policies towards this proposed alternative.

PAGE 26

5 Problem Identification The current rate of housing development in Florida has increased by 23.5 percent, which exceeds the capacity and the ability of governmental agencies to supply adequate water to consumers (Smith, 2005). One exam ple of increased demand in Florida for potable water was the formation of a regi onal water supplier: Tampa Bay Water. The focus changed from local counties and cities to regional planning to create projects to supply regional needs for water to approxima tely two million persons in three counties and three metropolitan cities. The current strategy is to continue groundwater and surface water withdrawals and examine the feasibility of creating large re servoirs for impounding the water. There are other poten tial alternative sources for potable water, such as capture of rainfall from roofs, brackish water sources, and seawater desalination. The use of Domestic Roof Rainwater Harvesting (DRRH) represents a feasible supplemental water supply. It appears to be the most economical alternative, because of the low capital investment of implementa tion, but there are seasonal limitations. Both brackish and seawater sources present high capital, operation and maintenance costs for the amount of water production, but the source is sustainable. There is a need for a rational model that provides a method for se lecting the appropriate use of roof runoff water. Purpose of the Study The objective of this investigation was to determine if roof runoff from five common roof surfaces could be a viable potable source considering regional treatment. The scope of this research was to assess th e quantity, quality and economics of recovered

PAGE 27

6 water from roofs delivered to a regional tr eatment facility. The concentrations of the metal elements were used as water quality i ndices because these would be the most costly public health risks to assess and treat. Th e economics of using the roof runoff for a smaller community with a population of approximately 30,000 was examined. The research briefly investigated the local and state regulatory is sues of using the recovered water, but social acceptance issues were not examined. The results and outcomes of the investigations is an objective and rational de cision matrix that will permit agencies to determine if this is an operational altern ative for safe, economical, and optimal use of water sources for their comm unity and their consumers. The factors that contribute to the quan tity and quality of the roof runoff were identified, such as the physiographic elemen ts of the roofing ma terial composition and climatic and atmospheric deposition factors th at are essential to the development of the model. The types of climatic factors and variab les are precipitation type, convective, orographic, and cyclonic type precipitation, the direction a nd trajectory of the storm, temperature, and humidity, which can all affect the constituents a nd particle deposition within the water. In addition, the time and duration, intensity, and the antecedent period between rain events can affect the concentra tion of the constituents in the sample. These variables provided insight to the variation and the larger weather system factors that need to be included in the development of a m odel. Physiographic variables of significant concern consist of the composition of the r oof surface materials, slope, and roughness of the material. These characteristics are important variables in high-intensity and short duration storms when determining the capaci ty of the roof for runoff production.

PAGE 28

7 Roofing material has a potential for l eaching metals along with potential of collecting atmospheric deposition of suspe nded metals, which are major factors in evaluating the quality of the water. The physical and chemical processes of the interaction of roofing material with the atmospheric constituents has a potential to pose an increase in concentration of a complex in the water runoff and, hence, an increase in exposure. Precipitation is geographically and spatia lly temporal and random. Because of these properties, this experiment requires that samples be taken and tested to assess the variability in concentration. The chemical analysis of the water from each event provided guidance as to the required treatment level n eeded to remediate water quality to a safe level for potable use that m eet the requirements of the EPA and/or local regulatory agencies. The combination of on-site climatic data field sampling, and laboratory analysis of the control and five different roofing surfaces' water samples for each event provided a mechanism to compare the water quality. Us ing statistical methods and analyses, a comparison of the water quality and quantit y between the control sample and roofing samples allowed for the selection of the roofing material that is preferred from a water quality perspective. The development of th e model incorporated these nuances of the experimental research findings for selecti ng appropriate use of the water resource.

PAGE 29

8 Methodology of the Dissertation The significance and uniqueness of this res earch is the integration of engineering and public health risks in asse ssing the use of DRRH. The expe rimental research phase of this study with the utilization of hydrological and chemical da ta adds to the knowledge of the effects of metal leaching from differe nt roofing material s and provides basic understanding of the water quality discharged from the roofs located in the Southeastern United States. One research outcome was to develop a ma nagement model for use of roof runoff as a potable water source. Th e purpose of this work was to accurately calculate the quality and quantity of the water that is rec overed from roofs. Meteorological data and roof surfaces were used to analyze the effect s of the variables on th e water quality of the roof runoff. Next, the economics and the publ ic health value, th e potential risk of concentration levels found in this water, wa s established. The research briefly examined the regulatory issue of using the water. The analyses of th e data collected allowed the creation and the development that resulted in a formulated decision matrix model that permits agencies to select an optimal and safe utilization of this alternative water source. This dissertation was structur ed as compilations of inde pendent research chapters with specific study objectives. Chapter II: Background consists of a review of current literature to identify gaps in the body of knowledge of the use of potable DRRH for regi onal treatment. Despite the years of using rooftop harvesting, there are not many references in l iterature as to the constituents and elements found in the water runoff. The review did not find specific roof runoff information on the materials used in th is study in Florida, and a study in Texas on

PAGE 30

9 roof runoff had only one similar material of th e five that were test ed in this study. The majority of the literature inve stigates the labor savings and quality of life improvement in water access and the role it plays in poverty-stricken regions and countries. Other research investigators concentrated on the st orage of water in cist erns and based their investigations upon the types of construction materials a nd how the structures and materials affected the overall water qual ity. The area of concern in investigating alternative water source s is the public health safety i ssues and their long-term exposure effects. Chapter III: Methodology presents th e means in which the selection and description of the experimental apparatus was determined. In order to provide more accurate estimates of the temporal quantity a nd quality variation of the water from the various types of roofing mate rial, an experimental design was initiated. Five different roofing materials were selected based upon th e frequency and popularity of the customers in the region according to a local roofi ng and construction company. This chapter presents the instrumentation and standard methods necessary to conduct the experiments and the standards in addition to the data co llection procedures. The chapter includes the data analysis of the variables a nd limitations with a final summary. Chapter IV: Results, the outcomes of th e water quality data analyses are presented, with specific item analysis and implications of the results. In Chapter V: Water Quality Results and Discussion, the outcomes of data analysis are discussed in the context of the implication of drinking water standards and public health issues for the potable use of the roof runoff water.

PAGE 31

10 Chapter VI: Model Development and Discu ssion presents the parameters that are needed for inclusion in the regional collec tion system design. This chapter discusses the modules, elements, and conditions in the deve lopment of a model. The model constraints and piping for optimal use of roof runoff are discussed. Chapter VII: Conclusions, describes th e outcomes of the investigation, field research and the augmentation model. Finally, Chapter VIII: Recommendations presents suggestions for further study and future investigations.

PAGE 32

11 CHAPTER II: BACKGROUND Overview This chapter background and literature re view was divided into smaller sections, such as meteorological effects on rainwater, biogeochemical processes, and stormwater compared to roof runoff. This interdisciplin ary approach to roof runoff identifies gaps within the literature in some of the secti ons, whereas other section topics had significant information available for review. The scientific literature discusses several approaches to the collection and use of rainwater. Sur, Bhardwaj, & Jindal (2001) repor ted that in Australia, India, and parts of Southeast Asia, DRRH is a traditional practice and part of the national water policy. Taiwan has successfully incor porated a DRRH system to supplement its potable water supply (Liaw & Tsai, 2004). Studies of rainwa ter collection in the U.S. have focused mainly on cisterns and cistern microb iology (Lye, 2002), while the Australian government recommends above-ground storage to prevent seepage and overflow into the tanks (Cunliffe, 1998). The thrust of most of this literature is the use of DRRH as a water source in remote or arid regions. It is understandable th at in arid locations, th e need for water takes precedence over the concern for its quality. Ho wever, there is new research emerging on the microbacterial interaction on roof surface and cisterns, in teraction of rainwater with roofing materials, and the subsequent health risks to the population. There are numerous studies throughout the world examining the di fferent concentrati ons of the various

PAGE 33

12 elements, metals, and other items of interest. The object was to review the literature, in the context of asking what are the processes and variables that affect the concentration of elements that might pose a risk. Meteorological Effects on Rainwater Water vapor can be considered the start of the hydrological cycle. The quantity of atmosphere water varies with location and time. On occasions, there may not be any relationship between the amount of water vapor over a regi on and resultin g precipitation. For example, at times there is more water vapor over the dry southwest than in the humid northern regions of the U.S. that receive the precipitation (Viessman & Lewis, 2003 ). The anthropogenic use of fossil fuel, slash and burn clearing of forest, forest fires, and smelting emissions all contri bute to the already existi ng fine particulate in the atmospheric addition to the natural occurring geochemical processes. Thunderstorms It is the unique chemical properties of rainwater in equilibrium with the atmospheric gases and the particulates comb ined with the geography of the state of Florida that creates the larg est number of thunderstorms in North America (Cooper, et al., 1998; Fernald & Purdem, 1998; Parker & Co rbitt, 1993; Viessman & Lewis, 1996, 2003). Cooper et al., (1998) examined the effects of a convective storm’s wind trajectory velocity both horizontally and vertically, inland temperatures, and the atmospheric pressures that develop as cooler sea breezes converge over the latent heated peninsula land in summertime. The undisturbed large-sc ale flow over the peninsula is strongly

PAGE 34

13 influenced by the relative position of the Azores Bermuda high. This position of the high during the summer creates the higher temperat ures over a peninsula and a strong breeze that induces rising air while en hancing rainfall. These summer months are the most active times on both coasts; the variation in rainfall is lowest during the summer because of the consistency of the summertime pattern of the daily rainfall from thunderstorm cells which form from the sea breeze. Studies have shown summer thunderstorms have a higher frequency on the east coast, and the convect ive storms occur earlier in the day on the west coast than they do on the east coast (C ooper, et al., 1998; Fernald & Purdem, 1998). Under these conditions, evaporation of the rain fall after a storm's passage is greater for the west coast, attenuating post-storm evapor ation and diminishing the amount of rainfall available for roof runoff. Nu merous studies have shown th at there is a general trend under a westerly flow; convection storms devel op earlier in the day on the west coast and then propagate eastward across the peninsul a (Cooper, et al., 1998; Fernald & Purdem, 1998). The opposite occurs for easterly flow: co nvective storms have a general tendency of convection starting on the east coast early in the day and then propagating towards the west coast later in the day (Cooper, et al., 1998; Fernald & Purdem, 1998). In the winter months, however, storms in Florida are gene rally cyclonic and ch aracterized as broad north-to-south trajectory cr oss-peninsula fronts. The remaining seasons, spring and autumn, are characterized by an uneven rainfall intermittent frontal sea breeze and random tropical storms (Cooper, et al., 1998; Fernald & Purdem, 1998). Large quantities of African dust are carri ed long distances by trade wind transport processes that affect Florida during the summer months (Garcia, et al., 2006; Guentzel, et al., 2001; Petersen, et al., 1998). Several Euro pean researchers have investigated the

PAGE 35

14 effects of the African dust. All have found that calcium (Ca) ions dominate as the precipitation-neutralizing agent of the strong acidity of marine aer osols. Thus, alkaline precipitation prevails, and when there is the lack of the calcium dust, more acidic rain is observed (Glavas & Moschonas, 2002; Loye-Pilot & Morelli, 1988). Scavenging Effects of Rain The origins of the trajectories are importa nt because of the scavenging activities of convective storms and frontal storms for the spatial and temporal trends in concentration and deposition. The transport of metals to the atmosphere is integral to biogeochemical cycling, and the dynamic nature of these transports accounts for deposits in remote areas far from the original sour ces. Long-range atmospheric transport occurs worldwide, and there are known natural sources of metals in the atmosphere from soil, sea salt, water, volcanic dust, and gas as we ll as anthropogenic emissions from fossil fuel combustion, industrial gas and particulates, a nd tillage. There is no national program in the United States or worldwide for assessing trace metals in atmo spheric deposition. A review and assessment of trace metals and atmospheric deposition data from numerous studies were compiled to give a reference poi nt for researchers (Galloway, et al., 1982). Galloway et al., (1982) reviewed the literature and compared th e concentrations of metals in rainwater emission rates from “human sour ces and natural sources with a mobilization factor.” A modified version of the fi ndings is illustrated in Table 2-1.

PAGE 36

15 Table 2-1: Global Mobilization Fact ors Based on Annual Emission Rates. Emissions(10 gy) Elements Natural Anthropogenic Mobilization Factor Cadmium(Cd) 2.90 55.00 19.00 Chromium (Cr) 580.00 940.00 1.60 Copper (Cu) 190.00 2,600.00 14.00 Lead (Pb) 59.00 20,000.00 340.00 Manganese (Mn) 6,100.00 3,200.00 0.52 Nickel (Ni) 280.00 980.00 3.50 Zinc (Zn) 360.00 8,400.00 23.00 Source: Galloway et al., (1980) Table 2-1 illustrates the mobilization fact or results which is one of the three different techniques examined by Galloway, et al., (1982). The mobilization factor is the measurement of the flux between the actua l metal emission between natural, and anthropogenic sources. Upon examining the mob ilization factor and enrichment factor, a comparison of atmospheric concentrations to the earth's crust concentration, which are predictive measures and the third techni que, is the actual measurement of metal concentrations over time, historical trends All three techniques, both the predictive conditions were in agreement with the historic al trends, for the concentrations of Cd, Cu, Pb, and Zn had an increased rate of deposit ion. Galloway et al. (1982) argues that the processes that control the rate of atmospheric deposition of Cd, Cu, Pb, and Zn in the

PAGE 37

16 eastern United States is at a minimum str ongly influenced by the anthropogenic process. At the time, because of a lack of data from a systematic collection of metals in wet deposition, the investigations were divided into three categories: remote, rural and urban. The working definition of "…remote [is] any area of the lo west concentration [excluding] Antarctica and Arctic. Rural is defined as representing the regional backgrounds and not directly influenced by local anthropogenic emissions, [and] Urban any site in a city or elsewhere directly influenced by local anthropogenic emissions” (Galloway, et al., 1982). Table 2-2: Concentrations of Meta ls Ranges Found in Wet Deposition. Urban Elements Range ( g )Median ( g ) Cadmium (Cd)0.48 2.300.7 Chromium (Cr)0.51 15.003.2 Copper (Cu)6.80 120.0041 Lead (Pb)5.40 147.0044 Manganese (Mn)1.90 80.0023 Nickel (Ni)2.40 114.0012 Zinc (Zn)18.00 280.0034 Rural ElementsRange ( g )Median ( g ) Cadmium (Cd)0.08 46.000.5 Chromium (Cr)<0.10 30.000.88 Copper (Cu)0.40 150.005.4 Lead (Pb)0.59 64.0012 Manganese (Mn)0.20 84.005.7 Nickel (Ni)0.60 48.002.4 Zinc (Zn)<1.00 311.0036 RemoteElementsRange ( g )Median ( g ) Cadmium (Cd)0.004 0.6390.008 Chromium (Cr) Copper (Cu)0.035 0.8500.06 Lead (Pb)0.020 0.4100.09 Manganese (Mn)0.018 0.3200.194 Nickel (Ni) Zinc (Zn)0.007 1.1000.22 Source: Galloway et al., (1982)

PAGE 38

17 Table 2-2 illustrates the wide ranges of ra inwater concentrations of metals from wet depositions for the three different categorie s, with the median within these categories. It is noted there are orders of magnitude differences in obs erved concentrations for any constituent, which reflects different locatio ns and sampling techniques. The urban median concentrations are higher, possibly because of the influence of point sources, whereas the remote sites were consistently lower. It is the physical characteristics of the metal and its compounds, in particle size, vapor pressure, heats of solution, and solubility, where the process affects the raindrop formation. Rain deposition is dependent on particle size and is determined by the rainout and the washout or scavenging. Fine particles and gases from convective thunderstorms which are consid ered tall and in the range of 12 to 16 km in altitude are generated in Florida in the we t season and have been recorded to scavenge particles from the middle and upper troposphe re and are transformed in rain drops (Guentzel, et al., 2001). Se veral researchers in Florida have, ''…reported these tall convective thunderstorms entrai n 60 percent of the air from the boundary level and 40 percent from the troposphere." (G arcia, et al., 2006; Guentzel et al., 2001; Petersen, et al., 1998; Viessman & Lewis, 2003). As the primary input for the hydrological cycle, precipitation type is defined by the vertical transport conditions generated, with the two most common found in Florida: convective precipitation in the summer and cycl onic precipitation in the winter (Fernald & Purdem, 1998; Viessman & Lewis, 2003). Conv ective type precipita tion is typical of the tropics where the precipitation is created by the process of heated water vapor at the land surface that rises, creating an upwe lling of vertical wind and downdrafts. The dynamic cooling of the water vapor results in condensation and precipitation; this

PAGE 39

18 typically takes the form of light showers or extremely high intensity thunderstorms. Cyclonic type precipitation is associated with the uneven heating of the earth that creates high and low pressure large-scale air movement of non-frontal or fr ontal origin. Hence, precipitation varies geographi cally, temporally, and seasona lly. Spatially, precipitation varies within the same storm precipitation and can considerably vary within a distance 20 feet apart from two recording devices as much as 20 percent (Fernald & Purdem, 1998; Viessman & Lewis, 2003). Biogeochemical Processes These complex processes and interac tions in a heterotrophic atmospheric environment is where the biogeochemical cyc ling of metal species occurs, and some are toxic (James N. Galloway, et al., 1982; Ta nner & Wong, 2000). Atmospheric gases such as NOx( g ) are adsorbed, causing acidification by nitric acid (HNO3 -) into the raindrop under the coexistence of gaseous sulfur dioxide (SO2( g )), gaseous nitrous oxide (HNO2( g )), and gaseous hydrogen peroxide (H2O2( g )) in washout and rainout scavenging. The atmospheric fluxes are importa nt because increased pollution emissions alter the pH levels, metal deposition, and c oncentration of toxic metals. The chemical reactions caused by the disassociation and oxida tion reactions of gases with the raindrops generates sulfate ions (SO4 2-) and hydrogen ions (H+) from the oxidation reaction of hydrogen sulfite ions (HSO3 -) with hydrogen peroxide (H2O2( aq )) in the raindrops. The oxidation reaction of hydrogen peroxide (H2O2( aq )) is an irreversible reaction, as shown in Equation 2.0. According to Henry’s Law of Constants (solubilities) characterize the equilibrium distribution between gas and liquid phases, where the equilibrium ratio in

PAGE 40

19 liquid phase concentration to gas concentrati on is much larger than in gaseous sulfur dioxide (SO2( g )). The gaseous hydrogen peroxide (H2O2( g )) and gaseous nitric acid (HNO3 -( g )) are more soluble in water than gaseous sulfur dioxide (SO2( g )) (Alfonso & Raga, 2002; Bachmann, et al., 1993; Mudgal, et al., 2007; Seinfel d, 1975). Equations 2.1, and 2.2 illustrate these processes. (Equation 2.0) (Equation 2.1) (Equation 2.2) Microbial Effects on Water Quality There are two modes of microbial contam ination of harvested water from roof runoff: roof contamination and cistern contamination. Roof Surface Numerous studies have analyzed the chemical composition of water. The prevalence of microbiological contaminants on rooftops has been less studied. The bacterial composition of roof runoff has not been widely explored in the published literature nor has the prevalence of microbiol ogical contaminants. The literature review for the microbial presence on the rooftop and in the catchment entrainment to the storage tank is still of controversy, with Lye's (2002) data in Kentucky where he states the high bacteria was from the process of the catch ment. Others disagree and regard the roof runoff as a source of clean water (Gould, 1999). Other researchers attr ibute the increased

PAGE 41

20 coliform and fecal coliform counts to the antecedent period on the roof surface area (Yaziz, et al., 1989). Yaziz et al. (1989) reports a positive correlation between the Heterotrophic Plate Counts (HPC) and the duration of the antecedent dry period. Meteorological Influences on Micr obial Concentrations on Roofs Aerobiological studies have found diurnal rhythms and a positive correlation with daily maximum temperatures, with monthly ra infall average, and temperature day-to-day spore levels for the fungal spores Alternaria (Corden & Millington, 2001). This research confirms the importance of rainfall and te mperature on spore concentration where the occasional rainfall resulted in hi gher monthly concentration of Alternaria spores. This further demonstrates the seasonal meteorologica l influences on concentrations, which has been positively correlated, with the inciden ce of allergic and infectious outbreaks in United States, Australia, and the United Kingdom (Corden & Millington, 2001; Evans, et al., 2006) The physical mass, size, shape of a vi rus, bacteria, and/or spore plays a role in efficient atmospheric dispersion. The wind ve locity and other meteorological conditions such as relative humidity, direction of the ai r trajectory, and the front al system have been shown under suitable conditions to spread ai rborne viruses more than 100 km (Jones & Harrison, 2004). Evans et al. ( 2007) reports wind direction infl uences the contribution of the total bacteria load on the roof area. Th ere is a strong correlation of the HPC and the wind velocity, which is a function of the prevailing wind and location of the source contamination (Evans, et al., 2006; Evans, et al., 2007). The natural processes of UV exposures, temperature of the roof, physio chemical reactions of contact, surface complexation reactions, the surface charges on the suspended particles, the rate of

PAGE 42

21 intensities and quantities of rainfall, a nd the microtransport and macrotransport mechanical movements across the roof surf aces are all processes that improve water quality. This improvement is supported by the observed microbial improvement in bacteria water quality in a collection sy stem (Benjamin, 2002; Coombes, et al., 2000; Evans, et al., 2006; Tchobanoglous & Schr oeder, 1985; Viesssman & Lewis, 2003; Zhu, et al., 2004). The roof surface collection sy stem is the sum of the process of the catchment water entrainment as the first phase towards an integrated system. Cisterns There are risks associated with rainwater storage; yet in arid regions such as southern Australia, approximately 800,000 sy stems are in use by the rural population along with the urban population of Aliadiae (Heyworth et al ., 1998). In a study of the five different types of cisterns in Micronesia, th e examination and report found that of the acceptable drinking quality, the ferro-cement ci stern had the best water quality, and the metal cisterns had the poorest water quality (Dillaha & Zolan, 1985). In a comparative study of the water quality of ci sterns in the area, receivi ng acidic deposition–Kentucky and Tennessee–compared to regions that had not received acidic deposition–St. Maarten, Netherlands Antilles–the rainwater was neut ralized upon contact with masonry cisterns (Olem & Berthouex, 1989). Samples from stored rainwater in tanks (in place) reported by Thomas and Greene (1993) were high in bacteria counts due to the tanks’ environments. In contrast, other researchers state that it is the stored rainwa ter tank environment that reduces the bacteria counts, and different water depths in the tanks foster different bacteria counts within that

PAGE 43

22 particular zone within the cistern tank (Spinks, et al., 2006). In some circumstances, geography dictates that the only viable water source is to harvest roof runoff, particularly on the islands of the South P acific (Connell & Lea, 1992; Ebi, et al., 2006; White, et al., 2007), on the United States Virgin Islands (Crabtree, et al., 1996; Heymann, 2004; Robertson, et al., 1992; Wyngaarden & Smith, 1985) and on some of the Greek islands (Sazakli, et al., 2007). Cisterns Founded in the South Pacific In these small island countries in the Paci fic, the water resources are limited due to the scarcity of potable fresh water. Thes e land masses are relatively small, preventing adequate groundwater storage on islands that ha ve an elevation of a few meters above sea level, whereas others are several meters above sea level (Con nell & Lea, 1992). The population of these islands range from inhab ited and rural to urban concentrations of large, unorganized population migr ations into urban areas that have no water or sanitation infrastructure, a situation typically found in most Third World cities. Many small island countries have relatively high rainfalls that are constrained by sma ll land areas and atoll geology, and water is usually treated as a co mmon resource (White, et al., 2007). The socio-economical pressures, combined with cu ltural value (or lack of value) of water resources, and an increasing population growth competing for limited resources such as water, housing, and employment creates an unsus tainable situation fo r these islands. In higher density population areas, pit latrines replace defecation on the beach, while the water supply source moves to shallow groundw ater wells because of demand, where the pits lead to groundwater contamination (Ebi, et al., 2006; White, et al., 2007).

PAGE 44

23 Usually in small tropical islands and rural ar eas of larger islands, drinking water is from roof catchments because the shallow well s have a high risk of contamination due to the lack of sanitation and th e presence of industr ial pollution (Taylor, 2001). The South Pacific island region is climatically sensitiv e. This region also has experienced extreme droughts as well as several severe storms a nd any changes in preci pitation or rising sea levels will present challenges to the water supply and public health (Ebi, et al., 2006). The United Nations’ recommendations for the at olls and many of the other small islands of the South Pacific include a strategy of conjunctive use of different sources of water such as combining the use of rainwater with groundwater (Taylor, 2001). The South Pacific’s haphazard approach to waters re sources is a contrast to the methodical development and reliance of the U.S. Vi rgin Islands on harvested roof runoff. Cisterns Founded in the U.S. and British Virgins Islands The U.S. and British Virgin Islands have a compulsory requirement in design and construction that a cistern must be used in every building (Crabtree, et al., 1996; Lye, 2002). Title 29, chapter 3, of the Virgin Isla nds Building Code, requires ten gallons-persquare foot of roof for a si ngle story dwelling and requires fifteen gallons-per-square-foot of roof for a multi-story dwelling. Crabtr ee et al. (1996) “…found no significant correlations…between cysts and oocysts and the bacteria or turbidity.” However, they reported a statistically signifi cant association between the he terotrophic plate counts and total coliform counts (r = 0.42638, P = 0.0061), and they also reported the association of the heterotrophic plate and the turbidity readings (r = 0.3249, P = 0.0305) (Crabtree, et al., 1996). The robustness of Giardia cysts' and Cryptosporidium oocysts' viability and

PAGE 45

24 hardiness to resist disinfection are a public health concern in dri nking water. There are many species of Cryptosporidium ; it can be found in 45 verteb rate species, including birds, rodents, reptiles, lizards, frogs, small mammals (squirrels, cats and dogs), and large mammals (cattle and sheep). Giardia is similarly found in a larg e variety of invertebrate species like Cryptosporidium (Crabtree, et al., 1996; He ymann, 2004; Robertson, et al., 1992). The species Cryptosporidium parvum from mammals is the only species associated with disease for humans, whereas Guardia lamblia is the species associated with human infections (Heymann, 2004; W yngaarden & Smith, 1985). An investigation of cisterns in the U.S. and British Virgin Islands over a one-year period concluded that 81 percent of the samples from public cisterns were positive for Cryptosporidium or Giardia, and only 47 percent of the private cisterns were positive for both (Crabtree, et al., 1996). However, the study did not examine the cysts and oocysts for viability. The researchers alluded to the fact that more research needs to be done in reference to viability and suggested that the warm temp eratures of 30C may facilitate inactivation. Investigators in a non-laboratory setting monitoring the oocysts' viability found desiccation was 100 percent lethal, freezing was also lethal (but a small portion can survive for extended periods of time), whereas a significant portion of oocysts were killed in all the environments inve stigated over a six-month pe riod (Robertson, et al., 1992). The vast majority of the invest igations in the lite rature, counted the presence of cysts or oocysts in excretions as infections, and the measures used are not determined by illness (Heymann, 2004; Stites, et al., 1987; Waterborne Pathogens: manual of water supply practices 1999). There are conflicting reports in the literature regarding the high percentage of asymptomatic children a nd adults in the population, which will be

PAGE 46

25 presented in more detail in the discussion section (Eisenberg, et al., 2007; Eisenberg, et al., 1998; Haas, et al., 1996; Haas, et al., 1991). These findings in the cisterns and water supply of the U.S. and British Virgin Islands are very different from the findings in Kefalonia Island, Greece. Cisterns Founded in the Greece Islands The island of Kefalonia has the similar limited water resources as all islands which have tourism as the major industry. Kefalonia’s population of 2,000 doubles in the winter and triples in the su mmer according to the investig ator (Sazakli, et al., 2007). The investigation of the water quality of rainwa ter, catchment runoff, and cisterns over a period of three years resulted in favorable physiochemical water qua lity for rainwater collection. The microbiological water quality of the rainwater is in concordance with previous studies (Crabtree, et al., 1996; Evan s, et al., 2006; Evans, et al., 2007), with the microbial indicators and pathogens counts f ound to vary greatly. Sazakli et al. (2007) investigation, found the microbi al indicators counts were in low numbers but in high percentages of the samples. The microbiologi cal quality using comm on microbial indices were contaminated in 80.3 percent of the sa mples (n= 156) (Sazakli et al., 2007). These results are similar to the U.S. Virgin Isla nds’ contamination of 81 percent (n=16) for public cisterns and 47 percent for private cister ns (n=30) (Crabtree, et al., 1996). In this particular investigation, the microbial contam ination was a result of the contact with the catchment area rather than the water itself (C rabtree, et al., 1996). Upon examination of a number of the samples taken during rainfall events, the investigators found there was no microbiological contamination present in a ra infall sample (Sazakli, et al., 2007). The

PAGE 47

26 investigators also detected a statistical di fference (p< 0.05) in the microbiological counts at 22C and 37C, indicating that there were higher count values with a higher temperature. This contrast s to Crabtree et al. (1996) who suggested the possible inactivation of bacteria cysts and oocysts at higher temperatures. Sazakli, et al. (2007) found the microbiological indicators showed the seasonal variations with a high count in autumn and a decreased count in winter. In addition, they found that microbiological indicators have a high negative correlation with chlorides, which illustrates the importance of location of the cistern, envi ronment, and meteorological effects on the water quality. Process That Changes Rainwater pH According to several resear chers, usually 80 percent of the wet deposition of the heavy metals are dissolved in rainwater (G arcia, et al., 2006; Kaya & Tuncel, 1997). The solubility of elements depends on a variety of factors including rainwater pH and the type of particles that are associated in the atmosphere. If the elem ent is already soluble in the rainwater, the higher the solubility and less significant effect the pH has on the element. Variation may be related to differences in pa rticle size and the efficiency of scavenging (J.N. Galloway, et al., 1993). The reactions w ithin the raindrop beco me a function of the drop size, where oxidation by H2O2 at a pH of less than 5 occurs in small raindrops where oxidation by O3 at pH values greater than fi ve occurs in large raindrops (Bachmann, et al., 1993). Bachma nn et al. (1993) states the radius variations of the raindrops appear to be dependent on the e fficiency of scavenged raindrops. Industrial aerosol particles such as Fe3+ and Mn2+ may influence the oxidation of SO2. Likewise, the

PAGE 48

27 aerosol particles of Ca2+ and Mg2+ may have a significant effect on the droplet pH (Williams, et al., 1988). The presence of H2S4, HN3, and organic acids, small ligands, organic macromolecules, and natural particles, depend on the availability of the acidic and basic species, but the reaction, a nd reaction between the reac tions can be neutralized predominantly by NH3 and CaCO3 (Kaya & Tuncel, 1997; Manahan, 1990, 1994). An example of neutralizing the ac idic raindrop conditions is the effect of long distance transport of soil particulates from Ca f ound in desert areas and African dust on the precipitation and atmospheric deposition. The main effect on rainwater pH is not fr om a local anthropogenic source; rather, this regional long transport of aerosol particles neutralizes an acidic rain event (Glavas & Moschonas, 2002; Herut, et al., 2000). Like wise, the examination of storm origin suggests Cu is present in highe r concentrations in continental storms when compared to marine dominated storms. These continenta l storms had elevated concentrations hydrogen ion in rainwater relati ve to marine dominated even ts, where a concentration of hydrogen ion has differed by an order of magnitude (Tanner & Fai, 2000; Tanner & Wong, 2000). This would imply that Cu is an anthropogenic source similar to Fe. In contrast to the impact and effect of storm origin, the Cr total concentrations were statistically equivalent, for both continental and marine origins have no effect on the concentration of Cr. (Kieber, et al., 2004; Kieber, et al., 2002). A negative correlation of the total copper concentrati ons in rain amounts indicates that Cu is a storm washout, which implies the origin is of anthropogenic local sources, and th e copper concentration decreases as rain increases (Kieber, et al., 2005; Kieber, et al., 2004; Kieber, et al., 2002).

PAGE 49

28 However, Fe and Cr had no correlation with ra in volume, suggesting that concentrations are the result of the long di stance transport at the site (Kie ber, et al., 2003; Kieber, et al., 2002). In addition to the complexities of thunders torms, there is a variety of processes occurring simultaneously, such as activity with in the storm itself, besides the oxidations of various metals (Bachmann, et al., 1993). R ecent literature suggests that lightning affects the dynamics of pH in the atmosphere of rainwater, subsequently affecting the solubility of metals within a storm event (Railsback, 1997). Railsback (1997) suggests the lightning generates in-cloud oxidization of SO2 and NO2, which contribute to rainwater having a lower pH associated with lightning. A comparison of solubilities of metals in rainwater from the literature in two different re search locations is illustrated in Table 2 3.

PAGE 50

29 Table 2-3: pH Effects on Sol ubilities of Metals in Rain water Between Locations in Turkey and Mexico. Ankara Turkey Ankara Turkey Rancho Viejo Edo. Mexico Ankara Turkey Rancho Viejo Edo. Mexico solubility of solubility ofsolubility ofsolubility ofsolubility of element in element in element in element in element in whole data sample withsample withsample withsample with setpH < 5pH < 5pH > 5pH > 5 %%%%% Mg2 +61 2664 23ma50 25na Cd88 1793 1082.679 t973.9 Cu49 2753 22na30 27na Cr35 2931 2866.7Il 1539.3 Zn43 2946 27na38 30na Pb40 3540 3362.4II 1543.3 Fe17 1621 19na12 11na Ni72 3184 1956.218 2163.4 Mnnana74.6na75.6 Element Source: (Baez, et al., 2006; Kaya & Tuncel, 1997) In Table 2-3, illustrates the solubilities ar e different and the range in variance between different locations. One example of th is was for Ni at the pH values > 5 in the individual samples; yet in the bulk sample s for Ankara, Turkey, the corresponding values are all in the same range as the Ra ncho Viejo Endo, Mexico samples.

PAGE 51

30 Public Policy for Drinking Water Federal Regulations for Drinki ng Water in the United States There was not a federal program to prot ect drinking water quality in the United States until Congress passed the Safe Dr inking Water Act (SDWA) in 1974, which was subsequently amended in 1996 and again in 1999. This Act created a federal-state partnership, which ensures compliance with fe deral regulation to pr otect the public from a variety of contaminants in drinking wa ter. The Environmental Protection Agency (EPA) establishes maximum contaminant leve ls (MCLs) for more than 90 biological, chemical, and radioactive pollutants (E.P.A., 2002a, 2002b). These are federal legally enforceable mandatory compliances as the primary and secondary drinking standards whereas the primary standards require compliance and secondary are recommendations of unregulated drinking water contaminants that may pose a health risk (E.P.A., 2002a, 2002b, 2005). The MCLs must be met by every community water system, which the EPA defines as any water conveyance having at l east 15 connections year -round or serving 25 or more people. Currently the EPA is investigating and researching an additional contaminant candidate list of 51 unregulated c ontaminants. If the i nvestigation and data show specific contaminants present a public he alth risk, a regulatory determination is made to add these contaminants to the primary drinking standards (E.P.A., 2005). State of Texas Regulations for Rainwater Harvesting The Texas government does not regulate private water systems. According to a 2004 report by the Texas Natural Resource C onservation Commission (TNRCC), “It is up to the individual to regul ate their own water system ” (Texas Natural Resource

PAGE 52

31 Conservation Commission, 2004). The Texas population shows a growing awareness regarding the limit of water res ources and the need for DRRH. For example, the cities of Austin and San Antonio are providing rebate s of up to $450 to homeowners who install DRRH, and other counties waiv e application fees and exempt the DRRH system from property taxes as an incentive (TNRCC, 2004). No federal or Texas water quality standards exist currently; however, the Texas legislature established a rainwater harvesting evaluation committee in May 2005 to recommend minimum standards (Texas Water Development Board, 2005). State of Florida Regulations for Rainwater Harvesting Florida's Administrative Code Chapter 64 E-8 sets the standards for private and limited use of water systems, and it establ ishes requirements and MCLs for community public works systems in the state of Florid a. The Department of Health administers Chapter 64E-8 through program coordination with all of Florida's 67 county health departments. A review of chapter 64E-8 a nd other Florida admini strative codes did not produce a regulatory code for roof runoff. A further investigation of chapter 373 of the Florida Statutes and the Stat e of Florida water policy set forth in Chapter 62-40 did not find regulations for use of roof runoff at this time.

PAGE 53

32 Regulations for Drinking Water in the Western United States In the states located to the west of the Mississippi River, water rights are classified much like real property and whoever diverts the water first retains priority. In contrast, in the states locate d east of the Mississippi River, water rights are based upon the British Riparian Rights system, which al lows a property owner the water rights to all water on their properties provided that usage will not impact downstream property owners. For example, in the Four Corners Re gion, which consists of the intersection of the states of Arizona, Colorado, New Mexico, and Utah, recently Colorado passed a law allowing limited rooftop rainfall collection. Prior to this 2009 ruling, it was illegal to gather the rainwater from a property’s rooft op unless the property owner also owned the water rights to said property. In Santa Fe New Mexico, it is now mandatory that new construction have rainfall catchments and in Tucson, Arizona, rainfall catchment is actively promoted for all new construction as well (Johnson, 2009). However in Utah, it is still illegal to harvest ra inwater from a property owner’s roof unless the property owner has water rights to said propert y. If the property owner does not have the water rights to a property, they must be appropriated via permit through the State Engineer and the original water rights' owner must agree to this arrangement. For most of the history of the United States, the water rights to states west of the Mississi ppi River were determined by diversion of a water body and have been sold off much like real estate properties (Utah Division of Water Rights, 2009).

PAGE 54

33 Regulations for Drinking Wa ter in Other Countries In Australia, 13 percent of all households use rainwater tanks as a source of drinking water. In southern Australia, the figure is 37 percent (Cunliffe, 1998). In the capital cities, 6.5 percent of households use th e tanks, but the usage ra te is 28 percent in the southern city of Adelaide (Australian Bu reau of Statistics, 1994). Of rural dwellers, 82 percent rely on DRRH as their primar y water source (Heyworth, et al., 1998). Australian rainwater tanks are constructed in accordance with the Australian/New Zealand standards for material selection, insta llation, and associated fi xtures and fittings. The Australian/New Zealand literature emphasi zes the proper select ion, construction, and maintenance of the tanks. The brief discussion of the types of roofi ng materials available (tiles, terracotta tiles, galv anized steel, polycarbonate sheeting, slate, and wooden shingles) recommends that consumers consu lt the manufacturer as to the materials’ suitability for DRRH. The Australian literatur e does not discuss the safety of different roofing materials. Regarding public attitudes toward DRRH a survey of the Australian population found that most citizens thought DRRH was bot h necessary and safe (Australian Bureau of Statistics, 1994). This populat ion preference is for rainwate r; therefore, they utilize rainwater tanks in both urban and rural sett ings and even when the municipal water supply is available. Researchers also reported considerable support for water conservation and recognition that water is a limited resource (Australian Bureau of Statistics, 1994). In summary, specific resear ch on rooftop materials’ effects on the quality of DRRH water is l acking. Research on public per ception of DRRH is limited to the Australian case.

PAGE 55

34 Stormwater Compared to Roof Runoff Numerous effects of urban stormwater runoff shows that the best management practices are to manage the stormwater us ing low-impact development and community design (Gaffield, et al., 2003). The authors rais e concern that the stormwater, while more readily accessible than rooftop runoff, contains more potential risk factors such as high nitrogen, vehicle exhaust, and various other sediments. In general, there is more literature on impervious area, and highway runoff conve yance such as solids, hydrocarbons, heavy metals, and chemicals. Researchers found high c oncentrations of zinc and other metals in the dissolved form in 35 to 65 percent of the stormwater runoff whole-water samples. The high proportions of the metals were bi oavailable in the water and soil sediment samples (Marsalek, et al., 1997; Stumm & Morg an, 1996). Tire-wear is a source of Zn where it is used in the manufacturing process to facilitate the vulcan ization of the rubber (Councell, et al., 2004). The research on stor mwater runoff quality in Texas found that the maximum contaminant levels MCLs for the EPA's drinking water regulations were exceeded 42 times for mercury (Hg) and 23 tim es for lead (Pb) in the total 272 samples (Zartman, et al., 2001). There is research that is beginning to fill the gap in the literature, quantifying water quality difference between the rooftop runoff a nd stormwater runoff per se in the context of urban roads an d highways (Gobel, et al., 2007). Table 2-4 illustrates the difference in the concentration of the various locations and the differences found in the literature review.

PAGE 56

35 Table 2-4: Comparisons of Concentra tions from Rainfall and Roof Runoff.

PAGE 57

36 CHAPTER III: METHODOLOGY This chapter addresses the methodology of water quality sampling and collection roof runoff, and includes the following subs ections: material selection and description, instrumentation, data collection pr ocedures, and data analysis. Material Selection and Description Roof Panels Two 4' X 8' roofing panels were constr ucted in accordance with the local and Florida state building codes. The panels were constructed from CDX roofing plywood, with 1"X 2" pine boards for the frame and 1" X 6" fascia boards for the gutter framing. The panel’s surface was covered with roofing paper, and the edges were encased with a galvanized drip edge. The wooden panels were then fitted with the five experimental surfaces. The first panel was topped with S4, galvanized steel, half painted with the manufacturer’s acrylic paint, and the other half with S5, galvanized steel unpainted. The second panel was fitted one-third with S1, a natural clay barrel, one-third with S2, a glazed clay barrel, and the remaining third with S3, a flat shaker impregnated tile. Each section of the panels had its own gutter and downspout made of painted galvanized steel, since conventional plastic materials tend to accumulate trace metals. The water samples drained directly into individual high-density polyethylene HDPE five-gallon containers. Each roofing panel was supported with concre te blocks; one side was 44 inches above the ground and the other was 30 inches a bove the ground. This slope was set to

PAGE 58

37 accommodate most building codes in the southeas tern United States, and similar to those of the majority of Florida roofs, w ith an approximate fifteen-degree pitch. Control or Reference Sample A sixth container was used for collection of the control samples. The HDPE fivegallon control container was elevated to the same height as the e xperimental panels, 30 inches above the ground, and placed three feet from the corner of the nearest experimental panel to prevent collection of deflected rainwater bouncing from the panels. This container was left open to the air and subject to natural rain events. The collected water was tested after each ra in event. The differences between the control sample and the rooftop-collected samples clearly showed the influence of surface materials on water quality. Instrumentation This section briefly describes the inst rumentation sensitivity, accuracy and detection limits of water quality testing. Field Sample Instrumentation The pH was measured using the Oakton Instruments Acorn series pH 5, with resolution at 0.01 pH, and with accuracy +/0.01 pH at the field site. The meter was calibrated using three point USA pre-pack standardized solutions for pH 4.01, 7.00, and 10.01. In order to avoid contamination, prior to inserting the probe into the next sample,

PAGE 59

38 the probe was rinsed with pre-pack standard ized rinse solution. Once the digital meter stabilized, the reading wa s recorded in the labora tory research notebook. The total dissolved solids (TDS) reading at the field site was accomplished using a Myron L deluxe DS meter model 532. The calibration of the meter was built into the instrument, with accuracy of +/-2 percent of the full scale. The measuring cup was filled three times with the sample to receive an accurate reading. On the third measure, the scale was selected, and the reading recorded. The instrument’s measuring cup was then rinsed with distilled water before proceeding to the next sample. The alkalinity at the field site was meas ured using a Hach test kit model Al-Ap colorimetric test in accordance with the kit’ s instructions and recommendations. After the titration was completed, the quantities of drops were recorded in the laboratory notebook and the titrated sample was disposed of, and th e small vial was rinsed with distilled water prior to proceeding to the next sample. Biological Instrumentation If there was any sample volume left in th e field, a sterile 1320 ml sample (volume permitting) was taken for biological testing. It was sealed and refrigerated, or transported to the laboratory in an ice chest. The sample was kept cold in the University of South Florida’s College of Public Health's walk -in laboratory research cooler until it was analyzed, and was plated within eight hour s. Biological testing was according to the Millipore method. If sample volume permitte d, the Standard Pour Plating Method was also performed. According to the Millipor e method, each 500 ml of the sample was individually filtered through a sterile glass funnel, which has a sterile 45 g filter to

PAGE 60

39 collect microorganisms. Using a standard labo ratory vacuum pump, the water sample was filtered through a sterile paper filter with the water passing into a fritted flask. Once the sample had passed through the assemblage, th e filter was removed with sterile tweezers and placed on a certified sterile media-specific agar plate fo r heterotrophic colonies. All of the above occurred in the University of S outh Florida’s College of Public Health’s Cell Media Laboratory in a sterile, negative hood environment. This procedure was performed twice for each sample, hence two individual pl ates per sample. The closed agar plate was sealed with parafilm from the outside envir onment, retaining its moisture content. The plates were placed in the Co llege’s walk–in incubator assigned to this project at the temperature of 35 C for 48 hours to develop the culture colonies (Clesceri, et al., 1998). The plates were then examined in the media laboratory under the microscope for heterotrophic colonies. The sp ecific heterotrophic agar’s co lorimetric system aided in identification and colony counts, but quantif ying the organisms precisely was out of the scope of this investigation. Af terwards, the assemblage used was sent for cleaning. After all the U.S. E.P.A. certified biological sterile samples jars had been used for the sample collection, a method was applied and used fo r the reuse of the bottles and caps in accordance with Standard Methods (Cles ceri, et al., 1998; E.P.A., 1992). All the apparatus components and sample bottles al ong with sealing caps were washed and cleaned in the automatic laboratory instrume nt washing and drying machine. Then they were placed in the autoclave fo r sterilization of the apparatu s and the dark certified glass sampling jars for future collection in accord ance to operating procedures and Standard Methods (Clesceri, et al., 1998; E.P.A., 1992). There were numerous occasions that there was insufficient volume for the Standard Methods' Millipore plate method.

PAGE 61

40 Meteorological Instrumentation Due to the specificity of the rain events in this subtropical region (weather can vary significantly within 20 m), the primar y data was obtained at the research sitespecific weather station dur ing the period of 28 August 2005 through 25 February 2006, using the Davis Instruments “Vantage Pro2 Plus.” At the station the following variables were monitored: Date and time in minute intervals Ambient temperature, temperature unde r each of the five surfaces, and temperature near collection containers Humidity, dew point, and evapotranspiration rate Wind speed, wind direction, wind r un, wind chill, and wind sample High wind speed and high direction of wind (shows trajectory of the rainstorm) Barometric pressure Rain and rain rate Solar energy, solar radiation, and UV dose Hot and cool days be tween rain events Temp-humidity-wind (THW) index and temp-humidity-sun–wind (THSW) index Wind chill factor and heat index.

PAGE 62

41 The meteorological data recorded 296,113 individual records and resulted in 11,844,520 data points. The data was collected on one-minute intervals continuously until the end of the research program. Measurements of ISS reception and Arc Interval monitored the function of the w eather station itself. This indicator provided a mechanism as to the quality and integrity of the data at that particular time. The ISS and Arc Interval provided the strength of the wireless recepti on between the outside monitoring device and the data logger to the computer inside the hous e. It was scientifically prudent to data log all the variables possible so that bias and se lection type errors c ould be diminished. Any of these variables may affect th e quality of the samples. Temp eratures of the panels were monitored because of possible effect on organi sm growth and of water evaporation on the panel. After the data was analyzed, only those variables that show statistically significant correlations were retained in the model. Figur e 3-1 is a picture of the residential location of the meteorological apparatus and the expe rimental material in-situ used in the research.

PAGE 63

42 Figure 3-1: Photo of the Appa ratus and Monitoring Station Analytical Instrumentation Atomic Absorption Spectrophotometry (AAS ) was used to measure the analyte concentration of a few trace metals found in se veral of the collected water samples from the six different panels (more details on th is are in the Data Collection Procedures section, under Preliminary Testing). Some samp les were split, with one part analyzed with instrumentation at the University, and a ll the others were anal yzed at a certified reference laboratory: Benchmar k Enviroanalytical Laboratorie s. An aliquot was retained for quality assurance. Using Varian Graphite Furnace AAS permits determination of the trace metals with sensitiviti es and detection limits 20 to 1000 times better than those of conventional flame techniques, without the ne ed for extraction or sample concentration (Clesceri et al., 1998). Many elements can be dete cted at concentrations as low as 1.0 g per liter. Some preliminary elements of intere st for the investigati on were Cd, Co, Cr, Cu, Fe, Ni, Mn, Pb, and Zn because of their known health effects; however, Pb was the

PAGE 64

43 chosen for the preliminary go or no go anal yses. Ascorbic acid and palladium solution (500-2000 g/ml) were used as matrix modifiers to reduce background noise of the detector. These analyses confirmed the presen ce of the constituents’ concentration levels and different concentrations on different su rfaces, and this expl oratory investigation provided a valid reason to proceed with this completed study and final analyses. The data for the preliminary investigation is presented in a table format in Chapter IV. All of the final analyses were perfor med using Inductive Coupled Plasma-Mass Spectrometry (ICP-MS) which conforms to EPA standards at the certified EPA, Florida, and nationally certified reference laboratory, Benchmark Enviroanalytical Laboratories. The samples required that 5,364 individual ch emical analyses were performed in triplicate to establish the m ean for the final concentration of 1,788 individual chemical observations of the following elements: Cd, Cr, Cu, Fe, Ni, Mg, Mn, Pb, and Zn. These concentration levels included all the auto mated quality assurances and standard laboratory referenced calibrations after ever y twenty samples. The calibration required calibration for each element, several blanks of known reference solutions for the specific element, and quality assurance references for the sensitivity and accura cy of the specific. As discussed in the literat ure review, the concern for roof runoff water was to meet the EPA primary drinking water standard s for Cd, Cu, Pb, whereas Fe, Mg, and Zn were secondary standards. The objective was to determine the concentrations of some elements that could pose a health risk in water roof runoff and to determine if the concentration exceeds the national drinking stan dards; if the concen tration exceeded the standards, the objective was then to assess that risk (Aldrich & Griffith, 2002). The data for the elements' concentration analyses is presented in a table format in Chapter IV.

PAGE 65

44 Data Collection Procedures Preliminary Testing In order to provide a baseline for futu re analysis, preliminary testing was performed before the actual research projec t commenced. The rainwa ter collected in the control container and rainwate r runoff from the five panels collection containers were tested for pH, alkalinity, and total dissolved solids, and Pb concentration was analyzed using the AA. There was a significant differenc e in pH and in levels of metals between the control rainwater and the pa nel runoff. The metals' concentration results from the five different panels varied significantly, enough to warrant proceeding with the multiple variable investigation. The preliminary data is discussed and illustra ted in Chapter IV. Primary Meteorologi cal Data Collection Prior to starting the research, a prelim inary study, was conducted to ensure the meteorological station would properly function and integrate with the collection database. This site-specific station monitored 40 separa te weather variables at one-minute intervals during this period. The meteorological system recorded 296,113 individual records and 11,844,520 individual data points. The type of ra infall events in the region required a station on-site because of variation in rain and convective nature of storms, which can change within 20 meters of a location. Bias a nd selection type errors were diminished by logging all the variables possibl e. Any of these variables ma y affect the quality of the samples. Temperatures of the panels were monitored because this may affect organism growth on the panel and evapor ation. After the data was an alyzed, only those variables that showed statistically significant corre lations were retained in the model.

PAGE 66

45 Primary Water Data Collection Rainwater washed down each sloped su rface, which was collected in each surface’s individual gutter, and then drained into individual collection containers. The composite samples were tested according to the EPA (E.P.A., 2002a, 2002b) and Standard Methods (Clesceri, et al., 1998). The water runoff samples contained dissolved, suspended material, and the particulate matter that had accumulated. Each vial or bottle met specifications established in the EPA's “Specification and Guidance for Contaminant–Free Sample Containers” (E.P.A ., 1992). From each collection container, a one 40 ml sample was preserved with 1 ml trace metal-grade HNO3 that was added in the field at time of sampling to prevent speciati on of the metals. The vial was sealed and taken to the laboratory for metals testing. The remainder of the sample was tested for pH, total dissolved solids, and alkalinity at the field site. If the volume remaining permitted, a sterile 1320 ml sample was taken for biological testing; the sterile bottle was sealed with the cap and refrigerated or transported to the laboratory in an ice chest. The sample was kept cold until it was analyzed and was plated within eight hours. The remaining water was discarded, and the container was placed back under its respective waterspout for the next rain event. The chemical and atmospheric processes at the surface of the panel are a major concern in periods of wet and dry deposition in the harvesting roof ra inwater. The control sample provided the reference for establishi ng possible interaction between the roofing material and the rainwater. Water samples fr om each panel and the control were analyzed for dissolved heavy metals after each rain event.

PAGE 67

46 The literature indicates that low dosage chronic exposure to metals in drinking water may lead to serious health conditions (Aldrich & Griffith, 2002; Hee, 1993; Louvar & Louvar, 1998; Manahan, 1991; Ness, 1994). The an alysis of the water quality is critical in the development and design of the water m odel for water use. For drinking water, the EPA has established maximum contaminant level (MCL) for primary trace contaminants, which are arsenic (As), cadmium (Cd), chro mium (Cr), copper (C u), lead (Pb), and mercury (Hg); and secondary trace contaminan ts, which are iron (Fe), magnesium (Mn), total dissolved solids, and zinc (Zn). Based upon these cr iteria, it was decided that each panel sample would be analyzed for Cd, Cr, Cu, Fe, Ni, Mg, Mn, Pb, and Zn. The sampling and analyses for these trace meta ls occurred over a period of nine months, which permitted capturing conditions during the wet and dry season. These analyses contributed to the economics and public health portion of the model for determining if a particular roofing material increased or reduced water quality through the leaching or absorption processes.

PAGE 68

47 Data Analysis Several programs were used to process and manage the dataset, in particular: Microsoft Access, for managing the 11,844,560 weat her data-points; Microsoft Excel, for managing the parse data in a more efficient structure; and Origin Labs Origin 7.5, for analysis and charts. The stat istical program used for the analysis was SAS 9.1.3 (SAS Institute, Carey, NC.); a number of statistical analyses were used in this study such as groups paired student t-te st analyses, correlation te sting, and parametric and nonparametric testing. Meteorological Data Analysis The weather data from the site locati on recorded 296,114 records of 40 different variables. The size of the data base required the use of Micros oft Access to create queries to parse the data to only thos e events that produced sufficient rainfall to allow collection of water samples, since there were rain events that occurred which did not produce a sufficient volume of rainwater for a chemi cal analysis. Microsoft Access queries were used to determine the time the rain event occu rred and to extract all data related to that event. The query rule for an event would require a positive indication of both rain and beginning intensity, while the query rule for th e end of a storm was indicated by zero (0) for both rain rate and accumulation. In additi on, the various Access queries were used to calculate the number of antecedent days prior to the rain event, and this information was added to the dataset. The parse data then was transferred to Microsoft Excel for additional graphing and analyses, creating a mo re manageable dataset. The categorical variables, such as wind direction, were analyzed using the program SAS (SAS Institute,

PAGE 69

48 Carey, NC.), for the prevailing wind direction over the storm periods or periods during that sample date. The weather data was then compiled with concentration analyses from the water samples using Inductive Plasma–Mass Spectrometry (ICP-MS). Antecedent Historical Analysis The historical data obtained from the S outhwest Water District Management was analyzed using SAS for the frequency analysis of the antecedents from 01 January 2000 through 25 December 2006 for site-55, site-56, and site-396 (the closest government weather stations to the research site-specifi c station). These numbers were also used as parameters and as means for the developmen t of the economical portion of the rational model. Rainfall Data Records for the Model The site locations provided accurate mete orological data over nine (9) months for both the dry and wet periods. However, rainfa ll data for a longer period was required to develop the model accurately. A digital data fi le was obtained from the National Climatic Data Center (NCDC) for a station site COOPID 84797 Lakeland, Florida (National Climatic Data Center, 2009). This weather sta tion was the nearest to the research location site that was representative of general area conditions This weather station had continuous 15-minute interval rainfall record s over the most recent period of time from1997 through 1998. If any of the individual records were flagged or the entry was questioned, it was removed from the dataset. Th e use of this data file in the development of the model is discussed in Chapter VI.

PAGE 70

49 Experimental Design Precipitation and runoff are often considered random variables because of the complexity of understanding the atmosphe ric processes that are known to drive precipitation (Viessman & Lewis, 2003). An experimental design is based on collection of a control sample and the difference be tween the control concentration and the collected samples' concentrations from the various roof surfaces. The paired t-test analysis of data was used because the samp les are not independent samples (Box, et al., 2005; Frigon & Mathews, 1997; Kachigan, 1991; Sirkin, 1994). This statistical methodology provides a mechanism of compar ison between the control and the other surface concentrations. Due to th e Central Limit Theorem, a sa mple size of thirty will often result in sample distribu tions that appear normally dist ributed even if the original population deviates from a normal distribution, hence the need for thirty or more rain events for the experiment (Box, et al., 2005; Kachigan, 1991). Sample Size All measurements were performed in trip licate, and the number of event samples were sufficient (30+) to be statistically significant. The data collection enabled the parsing and combining of the various datasets into a dataset for water sample analyses and weather data that resulted in 31 unique events over the two dis tinct seasonal weather patterns. The data set was considered of suff icient and significant sample size to permit statistical analysis of the interaction of the variables. This comprehensive dataset permitted the development of a management decision-making model for water resources for roof runoff.

PAGE 71

50 Data Exclusion Criteria The plots also revealed the need to analy ze the outliers: thus the use of exclusion criteria. The data was examined first without any exclusi on criterion. Graphical examination of the data illustrated that some exclusion criterion must be applied to the dataset. Outliers have an important impact on the conclusion of this study; by using the extreme studentized deviate (or ESD statistic) which allows the creation of decision rules for the outliers, the data is more readily ma nageable. After applying the decision rules for outliers, the analysis was run again without the outliers. The analysis of the ESD statistic wa s based on the following conditions: we hypothesize that = no outliers are present versus and = a single outlier is present with a type I error of (Equation 4-0) The sample value such that is refer to as Therefore, if then we reject and declare is an outlier; if then we declare that no outliers are present. Figures 4-1 and 4-2 are examples of th e outlier’s effect on the study (Appendix I should be referred to for the complete analysis of all the elements ex amined in the study).

PAGE 72

51 S5 S4 S3 S2 S1 ConPb 1.2 1.0 0.8 0.6 0.4 0.2 0.0 mg/LBoxplot of ConPb, S1, S2, S3, S4, S5* outlier Figure 4-1: Outliers' Effect on the Lead Concentration Data S5 S4 S3 S2 S1 ConPb 0.010 0.008 0.006 0.004 0.002 0.000 mg/LBoxplot of ConPb, S1, S2, S3, S4, S5 Figure 4-2: The Results of Removal of Outlier from the Lead Concentrations

PAGE 73

52 CHAPTER IV: RESULTS This chapter reports the results of the e xperimental research and methods used to analyze the data collected. As previously noted, this resear ch encompassed both field and analytical laboratory measures of key vari ables within meteorological, chemical, and biological areas. To ensure that all the m easures were reliable and valid, replicate measures were taken prior to logging the da ta in the laboratory notebook. The objective of this investigation was to determine the ch emical and biological water quality of roof runoff across the five selected roof surfaces commonly found in the southeastern United States. The purpose of the investigation was to quantify the potential health risk of small concentration of metals in pot able water, due to the high co st to remediate those small concentrations of metals in potable water supply. Table 4-1 is the preliminary data from AA spectrometry that showed a difference in concentration of Pb over the different su rfaces. The highlighted cells in the table are indication that Relative Standard Deviation in percent %RSD was at 100 percent. This is a measure of the reproducibility of the results of multiple measurement of the same sample. The measurements are more reproducib le the larger the %RSD. For example, in row 2 the water off the clay tile had a lead concentration of 0.0014. mg -1,whereas the water from the shaker tile had 0.0017 mg -1. The interesting observation was that the actual rain contai ned only 0.0013 mgl-1 (control sample). The data showed that both clay tile and shaker tile were releasing lead. Howe ver, the data from the four rain events showed that the water from the clay tile was 0.0013 mgl-1 and the glazed tile was 0.0134

PAGE 74

53 mgl-1 of lead. Table 4-1 is the summary of th e data for this study. The values for the individual elements are detailed in Tables 4-2 through 4-14. The c oncentrations in all these tables are expressed in mg -1.

PAGE 75

54 Table 4-1: Preliminary Data Results fo r Samples at the Site Using the AA. Rain Tile Glazed Tile Shaker Tile Coated Tin Tin Control 12345 pH 5.826.86 -1.04 6.89 -1.07 6.89 -1.07 6.03 -0.21 5.94 -0.12 Alk 6.8013.60 -6.80 13.60 -6.80 13.60 -6.80 13.60 -6.80 13.60 -6.80 TDS 9.009.50 -0.50 14.50 -5.50 20.00 -11.00 7.50 1.50 8.00 1.00 Pb 0.00120.0007 0.0005 0.0025 -0.0013 0.0023 -0.0011 0.0031 -0.0019 0.0004 0.0008 %RSD >100>100 0.00 43.40 56.60 9.70 90.30 >100 0.00 >100 0 pH 4.154.33 -0.18 6.25 -2.10 6.63 -2.48 4.20 0.05 4.33 -0.18 Alk 13.66.8 6.8 6.8 6.8 13.6 0 13.6 0 13.6 0 TDS 1915 4 12 7 18 1 16 3 13 6 Pb 0.0013 0.0014 -0.00010.0009 0.0004 0.0017 -0.0004 0.0007 0.0006 0.0027 -0.0014 %RSD 89.90>100 -10.10 25.80 64.10 >100 -10.10 4.30 85.60 15.50 74.4 pH 3.74.27 -0.57 6.75 -3.05 7.2 -3.5 4.13 -0.43 4.57 -0.87 Alk 13.613.6 0.00 13.6 0 13.6 0 13.6 0 13.6 0 TDS 31.128 3.10 37 -5.9 45 -13.9 39 -7.9 30 1.1 Pb 0.0027 0.0045 -0.0018 0.0024 0.0003 0.0005 0.0022 0.0031 -0.0004 0.0039 -0.0012 %RSD 31.15 26.10 38.8 -7.7 41.3 -10.2 20.5 10.6 0.3 30.8 pH 7.57.28 0.22 7.26 0.24 7.37 0.13 4.85 2.65 5.83 1.67 Alk 13.613.6 0.00 13.6 0 13.6 0 13.6 0 13.6 0 TDS 5025 25.00 3020 36 14 25 25 25 25 Pb 0.0030 0.0003 0.0027 0.0158 -0.0128 0.0016 0.0014 0.0036 0.0006 0.0014 0.0016 DIFF. Elements DIFF.DIFF.DIFF.DIFF.DIFF. Mean Temp FHigh Temp F 81.4094.70 -13.30 NE 79.0094.90 -15.90 80.9094.30 -13.40 Wind Dir Date Event 08-31-05 80.3094.00 -13.70 SW SSE NE 09-02-05 09-03-05 09-06-05 1 2 3 4

PAGE 76

55 Table 4-2: Roof Runoff C oncentrations Summary Results from the Site.

PAGE 77

56 Table 4-3: Roof Runoff Analyzed for pH Analyses from the Site.

PAGE 78

T T able 4-4: R o o of Runoff A A nalyzed fo r 57 r Alkalinity f f rom the Sit e e

PAGE 79

58 Table 4-5: Roof Runoff Analyzed for Total Dissolved Solids at the Site.

PAGE 80

59 Table 4-6: Roof Runoff Analy zed for Zinc from the Site.

PAGE 81

60 Table 4-7: Roof Runoff Analy zed for Lead from the Site.

PAGE 82

61 Table 4-8: Roof Runoff Analy zed for Cadmium from the Site.

PAGE 83

62 Table 4-9: Roof Runoff Analyzed for Nickel from the Site.

PAGE 84

63 Table 4-10: Roof Runoff Analy zed for Iron from the Site.

PAGE 85

64 Table 4-11: Roof Runoff Analyzed for Manganese from the Site.

PAGE 86

65 Table 4-12: Roof Runoff Analyzed for Chromium from the Site.

PAGE 87

66 Table 4-13: Roof Runoff Analy zed for Copper from the Site.

PAGE 88

67 Table 4-14: Roof Runoff Analyzed for Magnesium from the Site.

PAGE 89

68 Regression Model and Transformation of the Data Based upon preliminary analysis, there wa s no apparent benefit in using more complex statistical methods, such as regressi on analysis. Chandler (1994) indicates that more complex statistics are ju stified only when considering similar results and methods derived from the same dataset. Therefore, th e data was not transformed to create a better fit to the model, because the numeric tran sformation of the numeric concentration of chemical analysis would have no meaning and simply because there were only small differences exhibited overall. Correlation Matrix The next step required creating a correla tion matrix, comparing data from the control with that from the va rious variables of interest. The significance of a correlation matrix is that the descriptive has the power potential for predicting information about the values on another variable. The correlation in the descriptive form serves as a mechanism in data reduction. Nevertheless, most of a ll the existence of a correlation between two variables does not imply causa lity; it is possible that th ere were confounding variables that were responsible for the observe d correlation in whole or in part. The chemicals' analyses data were analy zed using descriptive statistical methods and presented in graphical plots and matrix, as shown below in Figure 4-2. Analyses of all of the variables of interest are also shown in Appendix I.

PAGE 90

69 Table 4-15: Roof Surfaces and Variab les’ Effect on the Copper Correlation. Correlation MatrixConCu S1S2S3S4S5ancedentrain ConCu 1.000 S1 .953 1.000 S2 .009 -.081 1.000 S3 .736 .759 -.065 1.000 S4 .912 .940 -.017 .839 1.000 S5 .904 .934 -.042 .843 .985 1.000 ancedent -.010 .045 -.151 .068 .025 .016 1.000 rain -.185 -.150 -.087 -.172 -.063 -.130 -.182 1.000 32sample size It should be noted that the correlation ma trix in Table 4-15 is a square, with as many rows as there are columns. The multivariate data matrix’s first characteristic to be noticed is that the diagonal coefficients are equal to one by the perfect correlations with

PAGE 91

70 each other. Each cell of the matrix contains a correla tion coefficient between the variables represented by the pa rticular column that the cell occupies. A visual inspection of the matrix provides information as to the relationship that exists among the variables. The darker yellow highlights represent a critic al value of the two-ta iled significance level at 0.01, and the lighter yellow highlights represent a critic al value of the two-tailed significance level at 0.05. For example, the matr ix in Table 4-15 illustrates that there is a positive correlation at the significance leve l 0.01 between the copper in the control sample and the samples for S1, S3, S4, and S5. The matrix also illustrates that S1, S4, and S5 are highly correlated to each other, wher eas S3 is correlated to a lesser value to the other samples. In contrast, S2 is not corr elated to the control for copper or the other samples and is not statistically significant. An inspection of the correlation matrix cannot assess the extent or joined eff ects of two variables with one an other or to the extent of the effects of a third or fourth variable, et c. (Frigon & Mathews, 1997; Kachigan, 1991). The variables analysis require s the use of other analytical techniques to determine the relationship. Descriptive Statistical Analysis Paired T-Test After completing the various plots of the data, the next step was to analyze the data using the paired t-te st and the Wilcoxon signed ra nk test to determine the relationship between the control and the follo wing surfaces: S1-natural clay barrel, S2glazed tile, S3-flat, shaker impregnated ti le, S4-painted, galvanized steel, and S5unpainted, galvanized steel. The data was co llected simultaneously as samples for the

PAGE 92

71 rain event; the data does not represent inde pendent random sampling, and therefore, the paired t-test is appropriate. The assumptions for a valid t-test are that the samples are random from the population of differences and th e set of differences come from a normal population. The assumption of normality for a t-test is only necessary for small samples (since for larger samples, the distribution of the sample mean has a normal distribution, regardless of the shape of the population fr om which the samples were selected). A sample size of 30 has been trad itionally used to distinguis h between “large” and “small” samples. However, it has been shown that as the sample size approaches 30 samples, the sample mean rapidly approaches normality. Si nce the samples sizes in the datasets were close to 30, there is no reason to doub t the validity of the procedure. The paired t-test analyses tested the null hypothesis that the population means of the control group for each of the contaminants is equal to the population mean of the “treatment” grouping (in this case the pane l roof surface samples). The data for the analysis was a set of differences between th e set of the treatment group and the control group. The null hypothesis is, therefore, the same as saying that the mean of the population of such differences equals zero. Symbolically the null hypothesis is: 0:0lreatmentControlH (Equation 4-1) SincedtreatmentControl we can write the null hypothesis as: 0:0dH (Equation 4-2)

PAGE 93

72 The alternative hypothesis is: :0AdH (Equation 4-3) The sample mean d x was calculated from the set of differences in the sample and sample standard deviation,d s The test statistic in the analysis is: 01d dx t s n (Equation 4-4) Since the assumed value of 0 in the null hypothesis is zer o, the test statistic is: 1d dx t s n (Equation 4-5) where n is the number of matched pairs in the sample, D is the difference for each pair of scores in the sample, d x is the mean of all the sample’s differences scores, d s is the sample deviation of the difference scores, and 0u is the mean of the difference scores for all possible pairs in the population. 2 dDD S n (Equation 4-6) If the null hypothesis is tr ue, each value calculated by this equation can be considered a randomly selected obser vation from a t distribution with 1 n degrees of freedom (sometimes called the null distribution).

PAGE 94

73 The decision as to the validity of the nu ll hypothesis or the alternative is based on the p-value. The p-value is th e probability that a randomly selected observation from the stated t distribution is extreme or more extrem e than the observed test statistic. If this probability is very small, then there is str ong evidence that the observation did not come from the null distribution, and it can be concl uded that the alternative is true. If the pvalue is not small, there is no reason to susp ect the test statistic came from a distribution other than the null distributi on, and there is no reason to di sbelieve the null hypothesis. The significance level is usually determin ed by convention as p-values are normally considered either 0.05 or 0.01. For example, Tabl e 4-16 illustrates the paired t-test of the control sample and S1-natural clay barrel; the significance level is 0.05, and there are statistically significant differences between the control and S1 sample.

PAGE 95

74 Table 4-16: Copper Paired T-Test S1 and Control. 0.0000000 hypothesized value 0.0058903 mean S1 0.0065871 mean ConCu -0.0006968 mean difference (S1 ConCu) 0.0018837 std. dev. 0.0003383 std. error 31 n 30 df -2.06 t .0482 p-value (two-tailed) -0.0013877 confidence interval 95.% lower -0.0000058 confidence interval 95.% upper 0.0006909 half-width

PAGE 96

75 Wilcoxon Signed Rank Test The chemical data was analyzed using the non-parametric equivalent of the paired t-test, the Wilcoxon signed rank test. The Wilcoxon signed rank test, also known as the Wilcoxon matched pairs test, is a non-parametric test used to test the median difference in paired data. As in the t-test, the null hypot hesis is that the median of the population of differences (treatment–control) is zero, and the alternativ e is that the population of differences has a value other than zero. The Wilcoxon signed rank test is based on the concept of asymptotic relative efficiency. This test is appropriate for small sample sizes with an unknown distribution, as this test is more sensitive than the t-test. The p-values for this test are interpreted the same as the p-values for the t-test. Small p-values are evidence that we should reject the null hypothesis and conclude the alternative is true. Large p-values do not provide evidence against the assumption that the median is equal to zero. This provides a basis upon which to develop further analysis of the data. For example, in Table 4-17 the control and the gl azed tile, S2, at the .01 significances level the control, and S2 is statistically significant. Table 4-17: Copper Wilcoxon S2-Control. variables:S2 ConCu 104sum of positive ranks 361sum of negative ranks 30 n 232.50 expected value 48.34 standard deviation -2.66 z, corrected for ties .0079 p-value (two-tailed)

PAGE 97

76 CHAPTER V: WATER QUALITY RESULTS AND DISCUSSION The objective of this investigation was to determine the water quality of roof runoff across the five selected surfaces. The investigation analyzed and characterized the chemical composition for the various heavy me tals in the roof runoff. The study samples were obtained using approved standard met hods and approved EPA containers to avoid possible contamination of the sample. Ten (10) rain events that did not produ ce sufficient quantity of roof runoff for the analyses were: 1-Sep-05, 5-Sep-05, 9-Se p-05, 4-Oct-05, 7-Oct-05, 8-Dec-05, 9-Dec-05, 11-Dec-05, 16-Dec-05, and 20-Jan-06. The paired t-test and the nonparametric test, the Wilcoxon test, were used in the statistical analyses of the data, which allows the examination of roofing material effects on the water quality. Using the National Primary Drinking Water Regulations as a standard, all results of the chemical analyses were examined for compliance with the maximum contaminant level (MCL) and their action level for all the roofing materials of S1, S2, S3, S4, and S5. If a specific substance exceeds the MCL, then the EPA requires that action be taken to lower the concentration of sa id substance. There was no exceedance of the regulatory standard in any of the thirty -one (31) samples obtained from the roofing materials S1, S2, S3, S4, and S5 over the nine (9) month inves tigation period. The exam ination of the data for all surfaces showed that there was no exceedance of the standard for chromium (Cr) 0.1 mg -1, and copper (Cu) 1.3 mg -1 with an action level of 1.3 mg -1. The national secondary drinking water regulations which are manganese (Mn) 0.05 mg -1, iron (Fe)

PAGE 98

77 0.3 mg -1, pH 6.5 – 8.5, zinc (Zn) 5.O mg -1, and total dissolved solids (TDS) 500 mg -1 were not exceeded. The specific cause of a response to the environmental variable was difficult to determine due to the concentration levels a nd the spatial occurrence. For example, the examination of the data for Pb indicated that there were some underlying processes occurring with the S5-unpainted, galvanized steel surfaces, because the samples for S5 were consistently lower than the control sample and lower than the S4-painted, galvanized steel. The data showed that S1 -natural clay barrel exhibited apparent adsorptive properties for l ead (Pb) and zinc (Zn). The Zn concentrations' analyses are shown in Table 5-1. The samples for S5 were higher than all other roof samples, w ith concentrations as high as 3.7 mg -1, which is approaching the MCL level of 5.0 mg -1. The surface S2-glazed tile, had the lowest mean concentration for Zn at 0.0585 mg -1, followed by S1-natural clay barrel, at 0.0722 mg 1, S3-flat, shaker impr egnated tile, at 0.1164 mg -1, and S4-painted, galvanized steel at 0.1345 mg -1. Only the runoff from surface S4 had Zn concentrations statistically equal to that of the control samples. This means that zinc was being ab sorbed or exchanged. Table 5-1: The Zinc Concentrations Analyses of the Roof Runoff mg -1 at the Site.

PAGE 99

78 Table 5-2: A Comparison of pH Levels at the Site. Table 5-2 shows that the pH of the samples taken from surfaces S2-glazed tile, and S3-flat, shaker impregnated tile, were consistently more ba sic than the control sample. Overall, the mean pH of the samples from S2 and S3 were 7.04 and 7.24, and were 0.77 and 0.97 larger than that of th e control, respectively. The hydroxide concentration in the samples of S2 and S3 are approximately 9.3 times greater than the controls. Conversely, the pH of samples from surfaces S4 and S5 are consistently more acidic than the control. The S4 and S5 samples, with mean pHs of 5.83 and 5.92 hydroxide concentration, is only 0.41 of that of the control, respectively. Only surface S1-natural clay barrel, had a sample pH clos e to the control. The control pH mean was 6.30, with a minimum of 3.67 and a maximum of 8.21. The difference in pH (acidic versus basic) did affect the metals' removal or concentrations. The total dissolved solids' (TDS ) mean concentration was 26.72 mg -1 for the surface S2-glazed tile, and this was the only surface found to be statistically representative of the control. For surf ace S3-flat, shaker impregnated tile, TDS concentrations were consistently higher than the control sample, with mean concentrations of 37.73 mg -1, which is approximately 1.43 tim es the control. This would imply that the material was dissolved from the roofing material, or atmospheric deposits

PAGE 100

79 are being retained and re-disso lved during the rain event. Th e opposite is true for surfaces S1, S4, and S5, where the TDS c oncentration levels are consistently lower than the mean control concentr ations, 26.41 mg -1. In the latter case, the ma terial would appear to be absorbed into the surface, as discussed previously. Samples from S1 had the lowest mean concentration levels 14.35 mg -1 with S4 at 17.67 mg -1 and S5 at 18.89 mg -1. A change in the TDS did not affect the activity of th e solution and any release would be a function of the pH rather than the TDS. Table 5-3 il lustrates the comparison of the different roof runoff and the control for TDS. Table 5-3: Comparisons of Total Dissolved Solids mg -1, Levels at the Site. The adsorption processes of S1-natural cl ay barrel consistently created lower mean concentrations of Fe, 0.0532 mg -1, while all the other surfaces were statistically representative of the control. The unglazed su rface of the clay barrel tile provides pores and adsorption sites for the deposition of iron (Fe). Many researchers suggest that the low surface energy charge between the tile surface and the adso rbed ions arises from adsorbate quadrupole interaction, with varying electrostatic field grad ients lattice of the solid (Benjamin, 2002; Jensen, 2003; Schwar zenbach, et al., 1993). Hydrogen is the

PAGE 101

80 simplest chemical substance that accounts fo r adsorption and surface diffusion, as well as distribution and character of the adsorption of the site s (McMillan, 1960). The lower concentration of Fe on S1, while all other surfaces were statistically equivalent to the control, could be explained via this adso rbate quadrupole inter action mechanism on the clay barrel tile. From Table 5-4, it is evident that the Cr concentrations in the S1-natural clay barrel runoff samples were consistently lower than the control due to the adsorptive quadrupole interaction mechanism. The Cr m ean concentrations in the sample runoff from the S2-glazed tile and S3-f lat, shaker impregnated tile, were consistently higher than the control sample, with mean concentrations of 0.0012 mg -1 and 0.0014 mg -1, respectively. This is 1.50 and 1.75 times the mean concentration of the control. The samples from surfaces S4-painted, galvanized steel and S5-unpainted, galvanized steel were found to be statistically representative of the control sample Based on my viewing of various tile-manufacturing machines, the rollers used to apply the ceramic glaze are manufactured in part from chromium (Cr). It is possible that there is a transfer effect that would add trace chromium (Cr) during the manufacturing process (Lyubenova, et al., 2009).

PAGE 102

81 Table 5-4: Comparis ons of Chromium mg -1, Levels at the Site. The analysis of Cu in the samples, is illustrated in Table 5-5, revealed that only surface S3 contained concentrations representa tive of the control mean concentration of 0.0066 mg -1, whereas all other surface samples ha d reduced concentrations of copper (Cu) except S3-flat, shaker impregnated tile. These concentrations are well below any action level of 1.3 mg -1, required by the EPA primary drinking standards. An action level (AL) is the concentrat ion of a contaminant over which a treatment is required. Table 5-5: Compar isons of Copper mg -1 Levels at the Site.

PAGE 103

82 The concentration dataset illustrated in Ta ble 5-6 for the Mg is limited due to the sample size. The data suggests that the S1 adsorptive quadrupole interaction processes also operate with Mg as seen with other elem ents, but due to the small sample size, it is difficult to fully infer this interaction. The standard deviation of the control was larger than that of the all the tiles. Table 5-6: Comparisons of Magnesium mg -1 Levels at the Site. Table 5-7: Roof Surfaces and Vari ables' Effect on the pH Correlation. ConpHS1S2S3S4S5ancedentrain ConpH 1.000 S1 .810 1.000 S2 .703 .871 1.000 S3 .562 .701 .916 1.000 S4 .729 .836 .790 .615 1.000 S5 .718 .902 .895 .747 .952 1.000 ancedent .510 .575 .518 .315 .688 .654 1.000 rain -.463 -.494 -.423 -.369 -.315 -.385 -.315 1.000 28sample size .374 critical value .05 (two-tail) .479 critical value .01 (two-tail)

PAGE 104

83 Table 5-8: Roof Surfaces and Variables' Ef fect on the Total Dissolved Solids Correlation. ConTDSS1S2S3S4S5ancedentrain ConTDS 1.000 S1 .901 1.000 S2 .907 .808 1.000 S3 .872 .747 .971 1.000 S4 .848 .888 .740 .750 1.000 S5 .896 .875 .775 .766 .965 1.000 ancedent .332 .286 .217 .273 .444 .536 1.000 rain -.551 -.505 -.658 -.647 -.503 -.530 -.315 1.000 28sample size .374 critical value .05 (two-tail) .479 critical value .01 (two-tail) Table 5-9: Roof Surfaces and Variables' Effect on the Zinc Correlation. ConZnS1S2S3S4S5ancedentrain ConZn 1.000 S1 .887 1.000 S2 .870 .885 1.000 S3 .755 .882 .684 1.000 S4 .779 .755 .832 .547 1.000 S5 .129 .092 .267 -.016 .510 1.000 ancedent -.050 -.013 -.041 .068 .054 .452 1.000 rain -.145 -.163 -.220 -.092 -.375 -.632 -.353 1.000 29sample size .367 critical value .05 (two-tail) .471 critical value .01 (two-tail)

PAGE 105

84 Table 5-10: Roof Surfaces and Variab les' Effect on the Lead Correlation. ConPbS1S2S3S4S5ancedentrain ConPb 1.000 S1 .487 1.000 S2 .738 .537 1.000 S3 .578 .486 .334 1.000 S4 .590 .426 .661 .217 1.000 S5 .381 .566 .460 .352 .359 1.000 ancedent .353 .324 .248 .296 .109 .057 1.000 rain -.244 -.425 -.221 -.095 -.104 -.107 -.353 1.000 29sample size .367 critical value .05 (two-tail) .471 critical value .01 (two-tail) Table 5-11: Roof Surfaces and Variab les' Effect on the Cadmium Correlation. ConCdS1S2S3S4S5ancedentrain ConCd 1.000 S1 -.043 1.000 S2 .162 .022 1.000 S3 .212 .373 -.068 1.000 S4 -.014 .259 -.209 -.116 1.000 S5 -.028 .076 -.209 .033 .165 1.000 ancedent -.089 -.093 .254 .185 -.173 -.118 1.000 rain .432 -.200 -.083 .422 -.171 .261 -.215 1.000 30sample size .361 critical value .05 (two-tail) .463 critical value .01 (two-tail)

PAGE 106

85 Table 5-12: Roof Surfaces and Variables' Effect on the Nickel Correlation. ConNiS1S2S3S4S5ancedentrain ConNi 1.000 S1 .806 1.000 S2 .786 .808 1.000 S3 .396 .018 .194 1.000 S4 .647 .547 .711 .102 1.000 S5 .730 .632 .695 .088 .909 1.000 ancedent .020 -.058 -.035 -.159 .083 .061 1.000 rain -.170 -.113 -.150 .383 -.207 -.133 -.355 1.000 23sample size .413 critical value .05 (two-tail) .526 critical value .01 (two-tail) Table 5-13: Roof Surfaces and Variables' Effect on the Manganese Correlation. ConMnS1S2S3S4S5ancedentrain ConMn 1.000 S1 .468 1.000 S2 .372 .958 1.000 S3 .171 .681 .622 1.000 S4 .321 .809 .755 .975 1.000 S5 .230 .768 .723 .973 .980 1.000 ancedent .025 .577 .509 .694 .697 .727 1.000 rain -.404 -.545 -.510 -.320 -.414 -.384 -.355 1.000 23sample size .413 critical value .05 (two-tail) .526 critical value .01 (two-tail)

PAGE 107

86 Table 5-14: Roof Surfaces and Variables' Effect on the Chromium Correlation. ConCrS1S2S3S4S5ancedentrain ConCr 1.000 S1 .483 1.000 S2 .148 .290 1.000 S3 .161 .181 .078 1.000 S4 .502 .428 .296 .298 1.000 S5 .304 .575 .085 .037 .657 1.000 ancedent .056 .010 -.120 .054 .344 .061 1.000 rain .113 -.262 -.130 -.293 -.319 -.250 -.215 1.000 30sample size .361 critical value .05 (two-tail) .463 critical value .01 (two-tail) Table 5-15: Roof Surfaces and Variab les' Effect on the Copper Correlation. ConCu S1S2S3S4S5ancedentrain ConCu 1.000 S1 .967 1.000 S2 .980 .987 1.000 S3 .752 .755 .792 1.000 S4 .927 .942 .954 .838 1.000 S5 .926 .936 .950 .840 .985 1.000 ancedent -.002 .017 .012 .038 .001 -.016 1.000 rain -.183 -.167 -.162 -.192 -.077 -.150 -.215 1.000 30sample size .361 critical value .05 (two-tail) .463 critical value .01 (two-tail)

PAGE 108

87 Table 5-16: Roof Surfaces and Variables' Effect on the Magnesium Correlation. ConMgS1S2S3S4S5ancedentrain ConMg 1.000 S1 .853 1.000 S2 .894 .965 1.000 S3 .861 .840 .946 1.000 S4 .913 .943 .986 .938 1.000 S5 .921 .933 .981 .955 .994 1.000 ancedent .574 .646 .624 .595 .698 .731 1.000 rain -.109 -.536 -.541 -.492 -.478 -.449 -.302 1.000 7sample size .754 critical value .05 (two-tail) .875 critical value .01 (two-tail) Table 5-17: Roof Surfaces and Variab les' Effect on the Iron Correlation. ConFeS1S2S3S4S5ancedentrain ConFe 1.000 S1 .617 1.000 S2 .391 .613 1.000 S3 .327 .636 .799 1.000 S4 .201 .494 .534 .479 1.000 S5 .479 .684 .628 .501 .238 1.000 ancedent .150 .084 -.089 -.083 .117 .133 1.000 rain -.063 .046 .180 .462 .092 .047 -.215 1.000 30sample size .361 critical value .05 (two-tail) .463 critical value .01 (two-tail) The correlations in Table 5-7 through Tabl e 5-17 show a negativ e correlation with the rain and this suggests that the deposition process is local; this finding is consistent with the literature (Kie ber, et al., 2003; Kieber, et al ., 2005; Kieber, et al., 2004; Kieber, et al., 2002; Luidold & Antrekowitsch., 2007). Similarly, the positive correlation of Fe with rainwater is consistent w ith the long transport cycle of Fe that is not of local sources (Kieber, et al., 2003; Kieb er, et al., 2002; Tanner & Fai, 2000; Tanner & Wong, 2000).

PAGE 109

88 Summary of Chemical An alysis of Roof Runoff The metal concentration levels of (Cd, Cr, Fe, Mg, Mn, Ni, Pb, and Zn) in over thirty-one (31) samples collected in this study from each of the five (5) roof surfaces were within EPA drinking water quality criter ia standards, as shown in Table 4-2. There was no exceedance of the primary and secondary drinking water standards.

PAGE 110

89 Item Analysis Paired t-test and Wilcoxon st atistical analysis are show n in Table 5-18. It can be seen from the data analysis that the roof surfaces are changing the quality of the water running off over these surfaces. The summary an alysis suggests there is a preferred roofing material for collection of roof runoff. Table 5-18 suggests that the S1-natural clay barrel roof material and S4-painted, galvani zed steel roof material, were the preferred roofing materials for collection of roof runoff. The S1 data illustrates that the adsorption properties are beneficial in th at they lower the metals concentration below that of the control sample. S1-natural clay barrel, then later releases the adsorbed contaminant in a lower concentration, reducing the overall aver age concentration found in the runoff, but S1 did not reach an equilibrium. The data exhibits a strong decrease in the zinc concentration found in the S1-natural clay ba rrel, whereas S5-unpainted, galvanized steel, had concentrations that were 7.45 times that of the control samples of zinc (Zn). This research has shown that the analyte concen trations meet the primary and secondary drinking water standards set by the EPA. Thus the study suggests the roofing material examined has only minimal impact on water quality.

PAGE 111

90 Table 5-18: Summary of Sta tistical Analyses of the Su rface Runoff Data for S1-S5. Roofing Surfaces AnalyteS1S2 S3S4S5 pHT-Test .8531.0004**.0001**.0099**.0189* Wilcoxon .7211.0003**.0002**.0175*.0378* TDST-Test 4.48E-05**.91991.20E-06**.0002**.0001** Wilcoxon 6.48E-07**.63282.81E-05**2.73E-06**6.10E-06** ZnT-Test .0009**.0008**.0563.19206.57E-07** Wilcoxon 1.02E-05**8.47E-06**.0060**.68841.92E-06** PbT-Test .4580.1396.3777.0966.0586 Wilcoxon .5855.3280.2560.1893.0901 CdT-Test .6979.5087.6201.9006.4671 Wilcoxon .3942.7454.8484.9199.7677 NiT-Test .1466.0578.3705.1257.3382 Wilcoxon .1727.0853.7578.3229.8129 FeT-Test .0066**.2018.7678.3782.0570 Wilcoxon .0073**.3886.8446.6735.1264 MnT-Test .1832.1193.2282.3453.7477 Wilcoxon .4823.0484*.1225.7076.3452 CrT-Test .0084**.06157.89E-07**.2102.4232 Wilcoxon .0048**.0003**1.13E-05**.1212.3409 CuT-Test .0482*.0089**.7482.0283*.0229* Wilcoxon .0086**.0079**.5958.0229*.0047** MgT-Test .0544.0856.1669.1169.1150 Wilcoxon .0180*.0910.1282.1763.2367 P=0.05 P=0.01**

PAGE 112

91 Wind Direction Analysis The preliminary analysis suggested that th e wind direction coul d have an effect on the outcome of the results, with eleven (11) ev ents from the Southwes t and five (5) events from the Northeast. The data analysis then sorted the data according to wind direction and examined if there were correlations using SAS between the wind direction and outcome. The preliminary data suggests that wi nd direction could ha ve an effect on the water quality; however, due to the small samp le size, further investigations would be required to determine this with certainty.

PAGE 113

92 Figure 5-1: The Prevailing Wind Direction at th e Research Site Over the Study Period.

PAGE 114

93 There was an attempt to correlate the wind direction with the biological plate count where the thought was the dispersion pr ocess would affect the plate counts, as shown in the following Table 5-15. The effects of the wind direction and the velocity of the wind was a major contributor to biological media and dispersion of biological media found on rooftops. Figure 5-1 represents the prevailing wind directi on of the study period of nine months, with 37 percent of the pr evailing wind during this period coming from the southwest. Then there was a consolidati on of the wind directions to the North and South to examine if there was an effect on th e concentrations and bi ological counts. An effect of the Northern winds on concentra tion was observed, which is plausible with continental land mass fronts. In Tabl es 5-19 and 5-20, the mean Zn was 0.0844 mg -1, from a Southern wind, whereas the Northe rn wind mean Zn concentration was 0.3297 mg -1, approximately 3.9 times greater. The mean concentration of copper exhibited a large difference in the mean concentration of 0.0049 mg -1 from the South and the mean copper concentration from the North was 0.0085 mg -1 approximately 1.73 times greater. Table 5-19: The Southern Wind Effects on Concentration of the Control.

PAGE 115

94 Table 5-20: The Northern Wind Effect s on Concentration of the Control. Unfortunately, the study was unable to obtai n sufficient biological plates count events. However, it was noted that Yaziz et al. (1989) reported in the literature that bacteria is always present in the air and on the roof surfaces of the 24 samples collected. Yaziz et al.’s (1989) plate c ount ranged from six (6) minimum to a maximum of 39 times 10 /100 ml, contrasting with some of the findi ngs at our site location. In this study, there were several samples that did not produce heterotrophic colonies. For example, 08 Oct 05 sample did not produce any colonies from any surfaces, whereas 02 Oct 05 sample did not produce any colonies on S1, S4, and S5, but colonies were present on S2 and S3, which is in contrast to Yaziz et al. (1989). In both cases, this study and that of Yaziz et al. (1989) need more data points for a conclusion.

PAGE 116

95 Table 5-21: Effects of Changes in West by Southwestern Wind Direction on Plate Counts. Plate 1Plate 2Median Avg Std dev22.20.022.20.00.0 25 Dec 05control 25 Dec 051WSW282825 Dec 052WSW202025 Dec 053WSW191925 Dec 054WSW282825 Dec 055WSW1616Sample date Total Values Suface Wind Dir. Table 5-22: Effects of Change s in East by Southeastern Wi nd Direction on Plate Counts. Plate 1Plate 2Median Avg Std dev19.119.419.019.76.1 23 Feb 06control 23 Feb 061ESE28.028.023 Feb 062ESE11.011.023 Feb 063ESE21.021.023 Feb 064ESE0.00.023 Feb 065ESE31.031.01 Nov 05control 1 Nov 051ESE14.022.018.018.05.7 1 Nov 052ESE42.027.034.534.510.6 1 Nov 053ESE36.045.040.540.56.4 1 Nov 054ESE0.03.01.51.52.1 1 Nov 055ESE8.00.04.04.05.7 Sample date Total Values Surface Wind Dir.

PAGE 117

96 Table 5-23: Effects of Ch anges in Southern Wind Direction on Plate Counts. Plate 1Plate 2Median Avg Std dev15.214.514.814.82.5 8 Oct 05control 8 Oct 051S0.00.00.00.00.0 8 Oct 052S0.00.00.00.00.0 8 Oct 053S0.00.00.00.00.0 8 Oct 054S0.00.00.00.00.0 8 Oct 055S0.00.00.00.00.0 2 Oct 05control 2 Oct 051SE0.00.00.00.00.0 2 Oct 052SE9.02.05.55.54.9 2 Oct 053SE3.06.04.54.52.1 2 Oct 054SE0.00.00.00.00.0 2 Oct 055SE0.00.00.00.00.0 29 Jan 061SE34.038.036.036.02.8 29 Jan 062SE31.033.032.032.01.4 29 Jan 063SE31.031.031.031.00.0 29 Jan 064SE27.036.031.531.56.4 29 Jan 065SE31.016.023.523.510.6 3 Feb 06control 3 Feb 061SSE28.018.023.023.07.1 3 Feb 062SSE31.030.030.530.50.7 3 Feb 063SSE29.033.031.031.02.8 3 Feb 064SSE20.026.023.023.04.2 3 Feb 065SSE29.020.024.524.56.4 Sample date Total Values Season Wind Dir.

PAGE 118

97 Table 5-24: Effects of Changes in North Wind Direction on Plate Counts. Plate 1Plate 2Median Avg Std dev15.813.014.114.14.2 7 Dec 05control 7 Dec 051NE20.018.019.019.01.4 7 Dec 052NE37.023.030.030.09.9 7 Dec 053NE29.022.025.525.54.9 7 Dec 054NE26.022.024.024.02.8 7 Dec 055NE25.031.028.028.04.2 23 Oct 05control 23 Oct 051NE0.00.00.00.00.0 23 Oct 052NE0.016.08.08.011.3 23 Oct 053NE42.06.024.024.025.5 23 Oct 054NE0.00.00.00.00.0 23 Oct 055NE0.00.00.00.00.0 17 Dec 05control 17 Dec 051NE10.07.08.58.52.1 17 Dec 052NE0.00.00.00.00.0 17 Dec 053NE0.0.0.00.0 17 Dec 054NE16.03.09.59.59.2 17 Dec 055NE0.00.00.00.00.0 1 Oct 05control 1 Oct 051NNE0.00.00.00.00.0 1 Oct 052NNE10.017.013.513.54.9 1 Oct 053NNE35.030.032.532.53.5 1 Oct 054NNE0.00.00.00.00.0 1 Oct 055NNE0.00.00.00.00.0 18 Dec 05control 18 Dec 051NW12.09.010.510.52.1 18 Dec 052NW33.032.032.532.50.7 18 Dec 053NW36.028.032.032.05.7 18 Dec 054NW34.020.027.027.09.9 18 Dec 055NW30.027.028.528.52.1 Sample date Total Values Season Wind Dir.

PAGE 119

98 Table 5-25: Effects of Cha nges in Southwestern Wind Direction on Plate Counts. Plate 1Plate 2Median Avg Std dev15.412.113.214.65.0 28 Sep 05control 28 Sep 051SW0.00.00.00.00.0 28 Sep 052SW0.00.00.00.00.0 28 Sep 053SW44.014.029.029.021.2 28 Sep 054SW0.00.00.00.00.0 28 Sep 055SW2.00.01.01.01.4 6 Oct 05control 6 Oct 051SW0.00.00.00.00.0 6 Oct 052SW0.00.00.00.00.0 6 Oct 053SW0.00.00.00.00.0 6 Oct 054SW0.00.00.00.00.0 6 Oct 055SW0.00.00.00.00.0 20 Nov 05control 20 Nov 051SW0.00.020 Nov 052SW17.017.020 Nov 053SW0.00.020 Nov 054SW0.00.020 Nov 055SW0.00.028 Nov 05control 28 Nov 051SW8.016.012.012.05.7 28 Nov 052SW35.016.025.525.513.4 28 Nov 053SW12.015.013.513.52.1 28 Nov 054SW7.02.04.54.53.5 28 Nov 055SW4.05.04.54.50.7 2 Jan 06control 2 Jan 061SW47.027.037.037.014.1 2 Jan 062SW32.029.030.530.52.1 2 Jan 063SW15.04.09.59.57.8 2 Jan 064SW16.00.08.08.011.3 2 Jan 065SW26.03.014.514.516.3 30 Jan 06control 30 Jan 061SW20.024.022.022.02.8 30 Jan 062SW30.032.031.031.01.4 30 Jan 063SW35.042.038.538.54.9 30 Jan 064SW41.031.036.036.07.1 30 Jan 065SW48.041.044.544.54.9 11 Feb 06control 11 Feb 061SW29.032.030.530.52.1 11 Feb 062SW9.00.04.54.56.4 11 Feb 063SW36.026.031.031.07.1 11 Feb 064SW14.00.07.07.09.9 11 Feb 065SW0.00.00.00.00.0 5 Dec 05control 5 Dec 051SSW12.08.010.010.02.8 5 Dec 052SSW16.07.011.511.56.4 5 Dec 053SSW37.030.033.533.54.9 5 Dec 054SSW5.013.09.09.05.7 5 Dec 055SSW19.07.013.013.08.5 Sample date Total Values Season W ind Dir

PAGE 120

99 Limitations of the Study and the Data One of the limitations of this study was th e sample size of a few elements, such as Mg. However, the sample size was sufficient to examine the majority of the selected metals found in roof runoff, such as meeti ng the water quality standards for the metal in question because of the costly remediation to be in compliance. With respect to the chemical analysis because of cost and time constraints, it would have been advantageous to have a c onfirmation by another certified laboratory to ensure no variance within the analysis. Wh ile the analysis was performed by the same individual in the same readings in triplicate, there is a possibility of a variance between laboratories. Another limitation of the study was that the roof material was new. Further investigation and future study should include new material that has been followed through time to ascertain if corrosion or oxi dation increases or d ecreases concentration release. Another reason for long-term research is that weather phenomena are spatial, temporal, and random. During the period of this study, these weather events were not indicative of nor representative of a norma l cycle of weather phenomenon. A study of longer duration would be beneficial to analyze trends over time. This research has provided information on roof runoff water quality for five different roofing materials. In conducting this part of the research, the following conclusions were reached: None of the water quality samples collect ed from the five roofing materials exceeds the primary or secondary drinking wate r standards. The metals were selected

PAGE 121

100 because they were among the most expensive contaminants to remove, in order to meet these standards. Biological a nd TSS of the contaminants would be removed from regional treatment systems as part of the normal pr ocesses and therefore were not considered critical. There are preferred roofing materials such as the clay barrel tile, which was found to have adsorptive quadrupole in teraction properties. The othe r preferred roofing material is the painted galvanized roofing material, b ecause it decreases the c oncentrations of zinc (Zn) when compared to the unpainted galvanized material. The glazed tile, the flat shaker impregnated tile, and the unpainted galvanized steel were found to be less desirable roofing material for roof runoff collection, be cause of higher metal concentration levels. The research findings and analyses were congruent with the Cu, Fe, and Cr correlations associated with the long trans port systems and anthropogenic sources (J. N. Galloway, et al., 1993; James N. Galloway, et al., 1982; Kieber, et al ., 2003; Kieber, et al., 2002; Mudgal, et al., 2007). Another sim ilar outcome of these analyses was the association of the wind direction on the HPC (Yaziz, et al., 1989). It is important to note that the experimental data we re not compromised by using PVC or plastic products, which have been know n to leach trace metals. Galloway et al., (1982) has stated that the older data in we t deposition are unreliable because at the time unknown leaching from the sample containers and plastics could raise the estimates higher by a factor of 10 for Cd, Cu, Pb and Zn (Barrie, et al., 1987). Because none of the water quality samples collected from the five roofing materials exceeds a concentration action leve l or a concentration for the primary or secondary drinking water standards, this wate r met potable water standards. This enables

PAGE 122

101 us to examine the roof runoff as an altern ative source for augmenta tion of potable water supplies. The results of this research potenti ally support the use of this roof runoff water to augment potable water supplies as a high quality source.

PAGE 123

102 CHAPTER VI: MODEL DEV ELOPMENT AND DISCUSSION This chapter presents relevant design and cost considera tion to address the economic feasibility of using roof runoff and re gional treatment. The crucial aspect in the development of this model was to align a nd reflect the conditions of the community, followed by needs and inputs of the user. The ap plicability of this model design is that it incorporates the flexibility fo r the user to change a variet y of variables and conditions found in the natural systems. This versatility al lows the evaluation of the feasibility of the roof runoff augmentation system for large and diverse conditions found in a smalldefined community. This permits the user to balance between the economics and the integration system, which includes roof runo ff augmentation to municipal supplies. This chapter discusses a strategy and a ssessment model that has been developed for determining the collection and cost of augmenting availabl e water sources using roof runoff as a potential water source. The Ra tional Concept shown in Figure 6-1 is the cause-and-effect inputs on the outcome of the model. The water quality and quantity, the geographic and demographic, and roof ma terials all affect the feasibility and sustainability of model. These conceptual input s, depicted in Figure 6-1, were parameters used to develop the modules of the Augmenta tion Model for the roof runoff as a potable water resource.

PAGE 124

103 Resource Water as a Potable Roof Runoff and Demographics Site Environment Water Quality Roof Materials Water Quantity Feasability Economic Sustainability A Rational Concept in the Development of a Roof Runoff Resource Model Figure 6-1: Conceptual Inputs for Creating the Augmentation Model Matrix

PAGE 125

104 The Design and Processes Used in the Model Development The framework used for the developmen t of the Augmentation Model utilized a location in the city of Temple Terrace, Florida, as a test case. The model takes into consideration that each city has unique characteristics and therefore, some of the input variables must be changed in order to accura tely represent the application. The model’s methodology uses three (3) categor ies for the water quality data, the meteorological data, and the geographical data analyses. These data were analyzed to ensure compliance with standards, as shown in Figure 6-2.

PAGE 126

105 Water Quality Parameter in the Model Development Water quality was found to be a function of the environment, such as relevant location of potential pollut ers, like industrial complexes that produce atmospheric emissions. Thus, the water quality and environm ent were the first parameters assessed in the model. Water quality parameters dictat e economics outcome, because water quality standards will determine the collection and treatment methods that are used. As previously discussed in Chapter V, this re search has shown that there was no exceedance of the EPA's primary and secondary drinking water standards for the metals in question over a nine-month period. Hence, the metals do not present a remediation issue or additional treatment costs. Roof runoff wa ter is a viable potable water resource. Meteorological Parameters in the Model Development In a preliminary analysis, the rainfall record from years 2000 through 2006 from the nearest gauging site from Southwes t Florida Water Management District (SWFWMD) were analyzed and showed the average rainfall for this period was 47.38 inches-per-year, which is similar to the findi ng of several researcher s (Fernald & Purdem, 1998; Wanielista, et al., 1997). Typical storms during the summer months in Florida are convective storms with high quantity of ra infall in short time periods. This water represents a substantial suppl emental water supply to the city, based upon the research data collected. The meteorological characteristics will be unique for each site, thus effecting quantity of rainwater capture based upon demographics.

PAGE 127

106 Figure 6-2: The Flow Chart of Proce sses for the Development of the Model

PAGE 128

107 Demographic Parameters of the Model Development Use of a stormwater runoff fr om rooftops represents an overlooked water resource for small communities such as Temple Terrace, and other communities that have a population of approximately 30,000 to 40,000. Th e uniqueness of this design allows water from the roof to be transported dire ctly to the water treatment facility through a collection system. This represen ts a substantial change in wa ter source strategy for water treatment and delivery to the customer. For example, Temple Terrace has a population of approximately 30,000, and 11,600 residential re sidences with an occupancy of 2.5 persons per home, as an average, within th e community. This represents approximately 12,000 homes in the corporate limits of the city of Temple Terrace. The average rooftop of these homes is approximately 3000 square feet (City of Temple Terrace Planning Department, 2006; City of Temple Te rrace Public Works Department, 2006). Geographic and Demographics Conditions on the Model The structure of the collection and dist ribution systems of the current water distribution systems can be used in developi ng this new strategy. Ho wever, the collection process requires piping and routing of water to relatively small transfer storage tanks. The augmentation strategy is to have the water treatment plant reduce the dependence on well water systems during and following a rain even t, thus maintaining a sustainable water strategy for the city. This research has shown that roof runoff can serve as an alternative potable source, without additional treatme nt processes for metals, hence minimal treatment. This proposed Augmentation stra tegy allows throughput directly to the demand, thus reducing the need for othe r more expensive alternatives.

PAGE 129

108 Sector Concept for the Feasibility and Viability of the Model Demographics Conditions The model area was defined from data provided by the city of Temple Terrace. For example, the area noted in yellow in Figure 6-3 is one section of the city showing the density and the typical grid c onfiguration of the neighborhoods. Figure 6-3: Temple Terrace Map Section of the Houses Used in the Model This sector area depicted in Table 6-3 is the single-family residential area between East 113th Avenue and Druid Hills Drive, and North 56th Street North to the Hillsborough River. This sector was utilized in the development of a model configuration. This neighborhood section is primarily a typical gr id North-South, East-West configuration, which is representative of most neighborhoods in Temple Terrace and of most of the Southeastern United States. The area is defi ned by 1,031 lots with approximately 24 lots per street with the gross roof area of 3,041,108 square feet (City of Temple Terrace S E N W

PAGE 130

109 Planning Department, 2006). Temple Terra ce has a population of approximately 29,000,with total residential wa ter connections of 11,600 with a daily demand of 4.1674 million-gallons-per-day (City of Temple Te rrace Public Works Department, 2006). The state of Florida’s average daily demand is 174 gallons-per-capita-day, whereas Temple Terrace has an estimated water usage per cap ita of 143 gallons-per-capita-day, which is below the state average (City of Temple Terrace Public Works Department, 2006). The total numbers of homes in Temple Terrace are estimated to be 11,600 with a total roof surface area of 35.275 106 square feet (City of Temple Terrace Planning Department, 2006). Geographic Conditions The total volume of roof runoff availabl e from the average rainfall annually of 47.38 inches-per-year is 139,268,068.8 cubic-feet -per-years, which equates to 1,041x109 gallons-per-year. Temple Terrace’s annual demand is 1.523 x 109 gallons-per-year (City of Temple Terrace Public Works Departme nt, 2006). The Augmentation strategy using the runoff has the potential of reducing gr oundwater withdrawals by 56.93 percent. This strategy is a significant leap toward sustai nability and conserva tion of our precious resource: the Floridian aquifer. According to the State of Florida De partment of Environmental Protection (FDEP) and the Florida Department of Tran sportation (FDOT), a “s tatistical analysis from Florida rainfall data and field inves tigations found that near ly 90 percent of all storm events that occur in any region of Flor ida in a given year will provide 1-inch of rainfall or less” (Florida Department of Environmental Protection, 2002). Based on this,

PAGE 131

110 we assume that a rain event of 60 minutes duration would have a volume of rainwater 1inch or less over 1,000 roof units in the m odel area, producing 215.4 or less, cubic-feetper-hour-per-roof or 26.85 or less, gallons-pe r-minute-per-roof of potable water. This was a preliminary analysis, and it was useful in determining if the strategy was feasible prior to modeling. The Configurations and Hydraulic Conditi ons Incorporated into the Model The hydraulic element of this model wa s based upon a gravity flow collection system and a forced main system returning collected water to the water treatment plant. The design of the system is a rational model that addresses the average approximate costs associated with components and appurten ances generally associated with the construction. All listed costs are presented as “general estimates” fo r the site: purchasing, designing, permitting, and construction; while al so recognizing that each project is unique. A wide variety of site specificatio ns and factors will come into play with individual projects in diffe rent areas, but the model purpos e is to provide feasibility calculations for the Augmentation system. Conceptual Description of the Configuration Design In this analysis it was assumed that th e homes’ roof gutters are connected to a leader pipe, which is subsequently connected to a lateral pipe. A typical lateral within Temple Terrace would have 12 to 24 connect ions, which equates to a flow of 322 to 644.4 gallons-per-minute. These laterals are then connected to the mains, which transfer the water to a central collection point. In one scenario, there are mains located under

PAGE 132

111 Druid Hills Road, Whiteway Drive, and East 113th Avenue, which are parallel and have diminishing elevation from North 56th Street towards the Hillsborough River. The ground storage tanks are to be located on a pr operty currently owned by the city, which represents a savings for the city of Temple Te rrace that is not realized in this analysis. The model includes the capacity input for the land costs variable in the calculations to be applicable to other municipalities. Hydraulic Conditions In the creation and analysis of the m odel components, the capacity of the collection system during a specific rain event was constrained by the configuration of the piping routing system. Another co nstraint of the system was the plant's ability to process the volume of water received from the roof runoff piping configur ation. The storage tank size is also constrained by land availability and economic cost factors. This fluid mechanics problem is constrained by econom ics and operational strategy implementation (Chase, 2004). The strategy of augmenting use of the ground water source to the storage catchments requires that the inflow is greater than the pumping rate of the force main out of the storage tanks. Manning’s equation was used to calculate the piping system for the gravity flow systems, while Darcy's equati on was used for the pr essure-piping network (Sincero & Sincero, 1996; Viessman & Le wis, 2003 ; Wanielista, et al., 1997).

PAGE 133

112 The Model Data Modules for the Input and Output Screens Water Data Module The model is a culmination of several models and numerous refinements that were developed over several months. Using a three-pronged approach shown in the flow chart in Figure 6-2, the water quality module used the data co llected during this research that demonstrated that the water quality parameters met the EPA standards. Meteorological Data Module The meteorological module used 19-year s of rain records from the National Climatic Data Center (NCDC) for the near by town of Lakeland, Fl orida. These rain records were used because of the proximity to the site location and the long record will not skew the results due to the random, spa tial, and episodic natu re of rainfall. The NCDC had to be transformed because the data file only contained rainfall events. In order to proceed with the time series analysis of data, it had to be transformed to include all 15minute intervals throughout the record with or without indication of rain (Carnahan, et al., 1969).

PAGE 134

113 Geographical and Demographic Module The model used the previously discus sed variables and constraints for the feasibility of the Augmentation Model. Th e configuration was based upon the sector previously described in Temple Terrace, whic h is representative of small towns in the southeastern United States. The model was developed in Microsoft Excel. The hydraulic elements of the model, sizing the pipes, a nd the various calculations, are contained in Appendix II. The Model Elements for the Variable s Sheet Input and Output Screens Figure 6-4 is the user input variables page where the user inputs the values into cells to produce the calculations and graphs.

PAGE 135

114 Figure 6-4.: Variable Input Screen

PAGE 136

115 User Inputs Variables The variables can be changed by the us er and are denoted by the white cells whereas the light blue tinted cells are calcula ted, or references that are not changeable. Under the model heading the “Catchment Vari ables” the user can adjust the number of groups, the number of roof sizes in square -feet, per capita info rmation 2.5 persons-perhome. For convenience under the same column, the light blue cell la beled the “Derived Participant the Demand/Day” calculates the demand for the model catchment. Under the heading “Community or Participant Demand per Day” the user must input the demand. The demand can be a sector, or a catchment demand found in the “Derived Participant Demand per Day.” The user can also input another demand, such as a larger community demand. The model is designed to allow the ca lculation of combining multiple sectors or catchments. The “Demand per 15 Minutes” is a calculated value based upon the “Community or Participant De mand per Day,” the result is found in the light blue tinted cell. This results in the value used in the subroutine of the mode l for the time series analyses and calculations of the data output. User Input Constraints This variable page, Figure 6-4, also allows the constrai nts to be entered into the model. Under the same heading “Model a Ca tchment Variables,” the user will find a “Maximum Reservoir Tank” size as a constraint of the model. This tank constraint value is zero, 4 million, 6 million, 8 million, or 10 million because this is linked to the financial reference module of the model. Using a value ot her than specified above will result in an output error.

PAGE 137

116 Another constraint of consideration is the “Effective Rain Variables.” These variables are dependent on the location of th e site and the surface interception value in inches, which accounts for the volume of wate r required to wet the surface. This volume is then lost subsequent to th e rain to evaporation. For this research site 0.05 inches was used as the “Interception capture volume” value to be subtracted from the volume of the first rain period in any event. The “Interevent” in hours is the time that has elapsed since the last rainfall event, for example is not uncommon during the Florida summers that there is a morning rain followed by afte rnoon rainfall. The increased temperature increases the evaporation of the morning rain, hence the roof surface is no longer saturated for afternoon rainfall flow over the roof. The interevent variable constraint takes the intercept value and subtracts that va lue from the first rainfall. If there was no rainfall between the last recorded rainfall in the time interval specified by the user in the interevent input cell, then the intercept valu e is substracted from the rain event until the intercept value has been reached. For example, in Figure 6-4, the user has specified the interevent at six (6) hours. Since the design of the model is an op timization based upon the pumping rate to the water plant, the plant becomes a constraint. The system was designed for a forced main pump of 58 hp with va riable speed capacity and us er defined. Therefore, the pumping capacity will affect the utilization and optimization of the model outputs. The uniqueness of this approach is the immedi ate pumping from the co llection and storage systems, a direct transfer of the roof runoff water to the treatment plant. This approach also creates a constraint ba sed upon the capacity of the plan t to process the transferred water. The capacity of the treatment plan t at the site location was a maximum of

PAGE 138

117 4,000,000 gallons-per-day. Located under the heading “Maximum Plant (Gal)” the user can define the constraint of the treatme nt plant at their specific location. Variable Input Screen Output The model outcomes of the variables are shown below the title heading “Model Analyses from 19 years of Data” in table fo rmat listing the supply pe rcentage utilization of the captured roof runoff. Under the table heading “Supply” there are entries for use, such as groundwater, rain partial and reserv oir, with the adjacent column indicating the percentage of utilization based on the input de mand. The partial is a combination of rain and groundwater. The Model Inputs Elements for the Sect or Collection and Piping System The sector collection in pi ping system module consists of a user input page to facilitate the pipe sizing sele ction and to identify the piping constraints. Shown in Figure 6-5 is the screen input page for the piping module. Under the heading, "Characteristics," the user can find the slope of the pipes by changing the top row labeled "Depth in feet" and "Distance in feet" for th e corresponding pipe diameters listed below in inches. The "Max Flow in gallons/15 Min" will change based upon the input of the slope. The "Max Flow in gallons/15 Min" value is calculated from the light blue tinted data for specific pipe diameters in the columns. The calcu lation consists of th e Manning equation, the slope, cross-sectional area, roughness coefficien t, and the hydraulic ra dius, with output of the velocity full in ft/sec. Using the continuity equation flow at full is calculated in cfs,

PAGE 139

118 gpm, and MGD, which are listed in the appropriate columns. The "Max Flow in gallons/15 Min" is the maximum carry ing capacity for that "Pipe Size." The second lower half of the input page a llows the user to se lect the "Pipe Size" from the drop window in the input cell. The user then must input the "Run," the distance in feet for each of the connections or runs for the length of pipe. The user then uses a matrix to c o mplete the pipe configuration for that sp ecific sector. For example, in Figure 6-5 under the heading "Step 1" the "Pipe Size is (8) inches, the "Run" is 100 feet, "Quantity units to connector” is 1, the "QA va riable page" in tinted light blue indicates there are 1,200 homes from the user's previous entry, Home to lateral" is 1, and “Home Connection flow to Lateral" duration is 24 homes. The pr evious inputs listed above calculate the "System Connection Flow" and the value is 14,645,190.52 gallons found in the tinted light blue cell. The same procedure is followed for the adjacent column labeled "Lateral." The user selected "Pipe Size" at 14 inches, followed by the "Run" of 1,200 feet, and the "Quantity units to connectors" at 50. Proceedi ng down to the next tint ed light blue cell, the calculated value is 606,947.69 gallons, whic h represents the maximum flow that the piping can receive. Looking at this stair step matrix below the piping and/or trunks, we find in Figure 6-5 "System Connection Fl ow" the value is 14,645,190.52, under "Step 1" under the "Lateral,” the maximum value is 606,945.69 gallons.

PAGE 140

F F igure 6-5: Pipi ng Routing Inp u u t Page Screen 119

PAGE 141

120 The trunk selection choice under labeled "Tr unk 1" has the input value as one (1) trunk of 48 inch-diameter with a trunk line maximum of 347.584.79 gallons; whereas, the selection "Trunk 2" has a value of 649,169.58 gallons for two (2) trunks of 48 inchdiameter. The preferred choice is “Trunk 2” because the maximum value is greater than the lateral capacity. The stair step matrix in this example has identified the constraints of the piping configuration, which is the "Later al" in this case, the lowest maximum flow compared to "Trunk 2" flows. The ability of the user to change th e various inputs that comprise the stair step matrix allows a multitude of combinations for the outcome of this sector and the ability to co mbine multiple sectors. Model Generated Results as Tables and Graphs After the user inputs have been assigned to the respective cells within the model, the model then begins to process the input va riables and to perform various calculations. The model takes the routing portion in pipi ng constraints in a subroutine and then incorporates the information into the main model. Because of the size of the files and data, the model may not have the capacity to run on a personal computer and may require a server to obtain the results.

PAGE 142

121 Figure 6-6: The Variable Input Sh eet Output Table for the Model

PAGE 143

122 Variable Outcome Output The model outcomes of the variables ar e shown above in Figure 6-6 under the heading "Model Analyses from 19 years of Da ta" in table format listing the ”Supply” and the “Percentage” of utilization of the captu re roof runoff. Under the heading “Supply” there is groundwater, rain, partial, and reserv oir, with the adjacent column indicating the percentage utilization based on the input demand. The part ial indication is a combination of rain and groundwater. This table was placed in the variable input for the ease of the user to see the effect of the changes on the input without referring to another section in the model output.

PAGE 144

123 Figure 6-7: The Model Graphica l Output of the Utilization

PAGE 145

124 Output of the Utilization Re sults as Tables and Graphs The graphical output of the "Model Anal yses from 19 years of Data" found on the variable input page is s hown in Figure 6-5, with supply ta ble in the right upper corner above the graph. In the upper left corner is the table for the sector frequency distribution of the rain in percentage, based upon the row label volumetric parameters. For example, 91.66 percent of the sector rain is between zero (0) and 500,000 gallons. Output of Frequencies of Rain The model output frequency of rain provi des the user a frequency table for the piping configuration limitations posed by the co nfiguration to capture the rain. The rain not captured by the piping configuration is considered excess. Table 6-1 shows the frequencies for the sector piping configuration and count s the excess events under the heading "Count of Excess." This table is usef ul in the design of th e configuration because the user utilizes the information from the frequency and the volume count. This tool allows the configuration for the best manage ment practices and the highest probability of capture for the least amount of dollars spent on the Augmentation system.

PAGE 146

125 Table 6-1: Model Output Frequenc y of the Piping and Rain Events. Row Labels Count of Excess 0.00 735296 118,880.78 109 446,130.78 287 773,380.78 66 1,100,630.78 153 1,427,880.78 35 1,755,130.78 91 2,082,380.78 30 2,409,630.78 71 2,736,880.78 12 3,064,130.78 36 3,391,380.78 7 3,718,630.78 32 4,045,880.78 3 4,373,130.78 15 4,700,380.78 5 5,027,630.78 8 5,682,130.78 6 6,336,630.78 6 6,991,130.78 4 7,645,630.78 2 7,972,880.78 1 9,281,880.78 1 10,590,880.78 1 Grand Total 736277 Table 6-2 provides the user a summary ta ble of roof runoff utilization over the 19 years of rain records. The ta ble provides the total rain amount in gallons for that sector; maximum, mean, standard deviation; and the pe rcent that the rain collector system could capture.

PAGE 147

T 1 v T T able 6-2: S u Table 9 years of r a ariability in T able 6-3: D a Year 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 u mmary of t h 6-3 is a fre q a in records. T rainfall ove r a ta Rain Re c Rain in 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 h e Sector Pi p q uency tabl e T his table is r the perio d c ords Annu a Inches 69.8 46.1 33.4 62.5 64.3 35.6 35.7 51.7 56.7 61.6 51.8 38.6 57 48.9 50.6 63.3 58.8 55.5 64.7 63.4 126 p ing Constr a e of annual r a useful for t h a l Rainfall. a ints. a infall for t h h e user to il l h e given yea r l ustrate nor m r based upo n m al annual n the

PAGE 148

127 The estimated cost for this project is $4,243,618.20. The model includes an interactive amortization sche dule, which shows that the annual cost of $18.00 per connection over the next 30 years. This tr anslates to a cost per homeowner of $1.50 dollars per month of equal paymen ts at an interest rate of three (3) percent. There is the possibility of a variety of funding oppor tunities for the implementation of the Augmentation system, such as direct f unding from the federal government water improvement program, municipal bonds for th e funding of capital. The payback period for the investment based upon the curren t wholesale water rate of $3.10 per 1,000 gallons. The savings per year from the a ugmentation is $276,345.91, thus the system with the four (4) million gallon st orage configuration would re quire a payback period of 15.36 years. This does not take into account incr eased wholesale water rate increases. This scenario does not provide for increased populatio n or connections to th e city service area not in the City of Temple Terrace. The model output estimates and th e amortization are in Appendix II. The Model as a Feasibility Tool for Alternative Sources This model is a tool for analyzing an d evaluating flow, storage and economic considerations of roof runoff and dete rmining the most economical strategy for augmenting a potable water supply using roof runoff. This source strategy may be more viable, considering a recent ruling by the South Florida Water Management District (SFWMD). Water providers can no longer us e traditional resources such as the Everglades for their continued water s upply per F.A.C. Section 3.2.1. (South Florida Water Management District, 2007a, 2007b). This ruling by the SFWMD has water

PAGE 149

128 providers searching for alternat ive supplies, and this model could be used as a tool for a selection of augmentations to water supplies, along with other opti ons. As this strategy revealed, the key to the storage issue is us ing the roof runoff as the events occur and processing the runoff as the demand increases. Current water use, consumption needs and public attitudes, appear to be changing in some parts of the United States due to multi-year droughts combined with increased populations. Communities are begi nning to investigate use of alternative supplies such as roof runoff and reuse in the United States; such is the case in parts of Texas. The review of the literature did find many specific pape rs discussing the eff ects of trace metal concentration in roof runoff, and many re searchers considered roof runoff only as a nonsource pollution–not a potential resource. In the literature wo rldwide, communities are looking for the sustainability of their communities and their re sources. In order to collect and use runoff from roofs, the chem ical composition of the runoff from a variety of roofing materials must be analyzed. DRRH is no longer just for developing countries. Rather, many industrialized countries are im plementing programs that include DRRH in urban centers. While this ar ea of DRRH is emerging in the United States, there are numerous international studies for developi ng countries and some industrial countries. Rain precipitation is a random even tempor al and spatial in nature that may present potential risks interacting w ith different roofing material in the Southeastern United States. At the time of this investigation and research, the review did not find any similar research in the state of Florida. However, the review found some generalized roof runoff research in the United States and around the world.

PAGE 150

129 This situation of water scarcity is not unique to the investigation region of Tampa Bay Florida. This is a global issue, and with the increasing populati on, we need to search for alternatives and efficiencies. Rainwater systems such as a community-based system, as presented in the model, offer the capa bility for storage and using the rainwater immediately, increasing the efficiency and reducing possible eva poration. Florida's latitude creates a unique environment that a llows the applicability of the model in most regions. Many of the world's communities are creating decentralized integrative systems versus large pipe centralized systems. Austra lia developed a strategy to use an integrated urban water management system and takes a comprehensive view of water supply, which includes DRRH, drainage, and sanitation (Coombes, et al., 2002; Mitchell, 2006). Globally and following the Australians’ concept, there are some other locations such as the US Virgin Islands and Taiwan, who have al l incorporated this strategy of DRRH into their construction code. More states and count ries such as Germany are also considering integration and code requirements for new construction (Cheng, et al., 2006; Herrmann & Schmida, 2000). The System Advantages Contra sted to Individual Units The advantage of an Augmentation Model is the economy of scale and the unique strategy of transferring collected water directly to the water treatment plant. The transfer strategy of collecting rain while using minimal size storage tanks to transfer water to the water treatment plant while the community is using the same water reduces the overall costs. The reductions in costs are associated with a reduction in the land cost and storage tank size. This same economy of scale allo ws monitoring of the Augmentation Model to

PAGE 151

130 piggyback on pre-existing monitoring programs already in place at the water plant. This immediate transfer accomplishes several obje ctives, such as removing citizens from water quality monitoring and disinfection of potable water. A strategic objective is removing the issues with individual cisterns and allowing professionals to monitor the treatment process at the water plant. Using th e services of water tr eatment professionals allows a mechanism to ensure the safety of the potable water and compliance to any future regulations. Another advantage of the Augmentati on Model is sectors can be operated individually or combined to meet the needs of the city. This approach also allows for a phased development of sectors coming on line with the water treatment plant. The contrast to the Augmentation system is the individual citizen cistern or system, which must manage water treatment and disi nfection. The costs and economics are not to scale to provide saving of the chemicals trea tment processes, or th e construction of the cistern. This is further complicated by the re sponsibility of the i ndividual to manage the water treatment. Repairs, ma intenance and operation costs wo uld have to be sustained by the individual. There would be additional co sts for the installation of backflow valve devices to prevent contamination to the potable water from the city. In the test case, the individual system could only support the home for only a few months of the year. Then there are costs in time to the individual citizen, who has to spend time managing the water treatment process, instead of work ing in gainful employment or relaxing.

PAGE 152

131 Regulatory and Policy Issues In the test case State of Florida, ther e were no statutes or regulatory policies preventing the implementation of the Augmen tation system. Other states such as the Texas (TNRCC, 2004) have embraced the D RRH concept. Johnson in 2009, states Arizona is actively promoting rainfall catchme nt to be installed for all new construction. However, some states in the we stern part of the United States restrict or forbid rainwater harvesting because they have different wate r laws. According to the Utah Division of Water Rights, (2009), it is illegal to harves t rainwater unless the property owner has the water rights. Therefore, the implementation of the Augmentation Model might have some regulatory or policy issues depending on the location. There are social and policy behaviors that have not been addressed becau se they were outside the scope of this research. The Augmentation System Compared to Aquifer Storage and Recovery This study has provided a model, which is feasible for smaller communities to implement. The Augmentation Model is an in expensive alternative compared to another strategy currently bei ng used such as Aquifer Storage and Recovery (ASR). The alternative is a full-scale aquifer storage and recovery (ASR), which requires that potable water be injected typically into a brackish water aquife r where it forms a bubble within the existing aquifer. This allows for storage and allows the potable water to be withdrawn at will. ASR wells are classi fied as injection wells and are regulated by the Underground Injection Control (UIC) progr am under the federal SDWA, including chapters 528 F.A.C. Injected water must meet drinking water standards prior to pumping the water into the

PAGE 153

132 storage zone. This regulation is justified for protection from contamination of the Floridian aquifer, our majo r potable water resource. Some of the considerations for ASR are site-specific due to the topography and the variable geochemical composition of the aquifer. The topography will affect the well depth required for permitting. The geochemical composition of the aquifer presents a challenge. Several studies have found that there has been l eaching of arsenic and other metals that co-precipitate in the water-ro ck interface ,which re sults in causing the mobilization of metals into the extracted wate rs (Arthur, et al., 2002). This could require additional treatment of water to ensure it meets the MCLs of the water drinking standards. This possible additional treatment adds cost to recovering waters from the ASR. There are also site-specific efficiencies for reclaiming the potable water that were injected into the aquifer, with recovery th at can be 65 to 75 percent of the original volume allotted. For example, the elevation of the site above the aquifer for storage and the TDS of the injected aquifer zone create additional costs to the project. Current cost estimations for ASR are $2,000,000 per-million-gallons-per-day; this includes testing and permitting (Southwest Fl orida Water Management District, 2006). This same estimation can be used to estimat e the cost of adding a dditional wells to the current well field. The captur ing of roof runoff strategy maximizes the throughput of resources by eliminating expensive ASR; curre ntly the largest ASRs found in the state are 2 MGD and on the average permitting and testing requires a five-year lead-time or more. Then there is the problematic questi on as to where to find the additional water source. The Hillsborough River cannot supply th e surface water that is currently needed to meet the permitted demand of municipalities and regional water supplier. It is not

PAGE 154

133 feasible or reasonable to store 458,410,950 gallons at a cost of $ 2M per–million-gallons for a total cost of $229,205,475. The permitting and regulatory issues for this size ASR have not been addressed. The lead times fo r large scale ASR are unknown at this point. In contrast to the ASR, this roof ru noff strategy and model is based upon actual collected data and historical rainfall data records from the years 1979 through 1998. The user inputs the catchment area in square-f eet, number of homes, and 15-minute demand into the model. After the sp ecific location, parameters are entered into th e model. The result is calculated as a percentage of demand for this specific site, and this determines the volume available for augmentation. The mo del calculates at fifteen (15) minute intervals the amount of water th at can be used for potable wa ter from roof runoff. This model indicated in the test case that ove r the 19-year period, 56.93 percent of demand was met.

PAGE 155

134 CHAPTER VII: CONCLUSIONS This research has provided information a nd data to the literature on roof runoff water quality on five different roofing materi als in a non-arid area. In conducting this research, the following c onclusions were reached. None of the water quality samples collect ed from the five roofing materials exceeds the primary or secondary drinking wate r standards. There are preferred roofing materials such as the clay tile, which was found to have absorptive and desorptive properties. The other preferre d roofing material is the painted galvanized roofing material, because of decreased concentrations of zinc when compared to the galvanized material. The water quality and model strategy of using the water pl ant as part of the system avoids most problems associated with individual water supplie s and/ or cisterns. Water quality of the roof runoff was of high quality when compared to stormwater water recovery. The model also illustrated the cost benefits of capturi ng roof runoff for augmenting the potable water supply. The versatil ity of the model allows the analyses of individual sector systems, and co mmunity systems or a combination the unique ability to examine this alternative source strategy paramete rs of pipe routing system, cost analysis and feasibility of a system's cost-effectiven ess. The model provides a management tool for examining alternative best management practices.

PAGE 156

135 CHAPTER VIII: RECOMMENDATIONS The study revealed that this is a viable and feasible method of augmenting an existing water supply with the following recommendations. Continue monitoring the apparatus for changes in water quality and meteorological conditions. Continue to work with the City of Te mple Terrace and the water management district to incorporate the strategy and test a prototype or p ilot study sized system potentially in a new development. Propose a pilot plant to test th e efficiency of the model. Explore further refinements to the model a nd collect additional data to refine the model. Provide additional options and improvements for the users based upon their needs. Future research is needed on the clay til e and other roofing materials to define their water quality aspects.

PAGE 157

136 REFERENCES Aldrich, T. E., & Griffith, J. (2002). Environmental Epidemiology and Risk Assessment Hoboken: John Wiley & Sons, Inc. .274 Alfonso, L., & Raga, G. B. (2002). Estimati ng the impact of natu ral and anthropogenic emissions on cloud chemistry: Part I. Sulfur cycle. Atmospheric Research, 62 (12), 33-55. Arthur, J. D., Dabous, A. A., & Cowart, J. B. (2002). Mobilization of Arsenic and Other Trace Elements during Aquifer Storage and Recovery Southwest Florida Paper presented at the U.S. Geological Su rvey Artifical Recharge Workshop Proceedings Sacramento. Australian Bureau of Statistics. (1994). Environmental issues peoples views and practices Canberra: Australian Bu reau of Statistics. Bachmann, K., Haag, I., & Roder, A. (1993). A field study to determine the chemical content of individual raindrops as a func tion of their size. Atmospheric Environment. Part A. General Topics, 27 (13), 1951-1958. Baez, A. P., Belmont, R. D., Garcia, R. M., Torres, M. C. B., & Padilla, H. G. (2006). Rainwater chemical composition at two sites in Central Mexico. Atmospheric Research, 80 (1), 67-85. Barrie, L. A., Lindberg, S. E., Chan, W. H., Ross, H. B., Arimoto, R., & Church, T. M. (1987). On the concentration of trace metals in precipitation. Atmospheric Environment (1967), 21 (5), 1133-1135. Benjamin, M. M. (2002). Water Chemistry (1 ed.). Boston, Ma.: Mc Graw Hill 668 Box, G. E. P., Hunter, J. S ., & Hunter, W. G. (2005). Statistics for Experimenters (2 ed.). Hoboken: John Wiley & Sons.633. Carnahan, B., Luther, H. A., & Wilkes, J. O. (1969). Applied Numerical Methods New York: John Wiley and Sons Inc.604 Chase, R., Jacobs, R., Aquilano, N.,. (2004). Operations Management for Competitive Advantage (10 ed.). New York: McGraw-Hill /Irwin.765

PAGE 158

137 Cheng, C. L., Liao, M. C., & Lee, M. C. (2006). A quantitative evaluation method for rainwater use guideline. Building Services Engineering Research & Technology, 27 (3), 209 -218. City of Temple Terrace Planning Department. (2006). Roof Runoff Collection Project. In R. Carnahan (Ed.). Temple Terrace. City of Temple Terrace Public Works Department. (2006). Plant Water Stats 2006 City of Temple Terrace: City of Temple Terrace.10 Clesceri, L. S., Greenberg, A. R ., & Eaton, A. D. (Eds.). (1998). Standard Methods for Examination of Water and Wastewater (20 ed.). Washington D.C.: American Public Health Association 1152 Connell, J., & Lea, J. (1992). My country will not be there : Global warming, development and the planning resp onse in small island states. Cities, 9 (4), 295309. Coombes, P. J., Argue, J. R., & Kuczera, G. (2000). Figtree Place: a case study in water sensitive urban development (WSUD). Urban Water, 1 (4), 335-343. Coombes, P. J., Kuczera, G., Kalma, J. D., & Argue, J. R. (2002). An evaluation of the benefits of source control meas ures at the regional scale. Urban Water, 4 (4), 307320. Cooper, H. J., Smith, E. A., & Rubes, M. T. (1998). Relevance of Surface Energy Budget within Florida Sea-Breeze Front to Cro ss-Peninsula Rainwater Runoff Gradient. Journal of Applied Meteorology, 37, (9), 939–950 Corden, J. M., & Millington, W. M. (2001). Th e long-term trends and seasonal variation of the aeroallergen Alternaria in Derby, UK. Aerobiologia, 17 (2), 127-136. Councell, T. B., Duckenfield, K. U., Landa E. R., & Callender, E. (2004). Tire-Wear Particles as a Source of Zi nc to the Environment. Environ. Sci. Technol., 38 (15), 4206-4214. Crabtree, K. D., Ruskin, R. H., Shaw, S. B ., & Rose, J. B. (1996). The detection of Cryptosporidium oocysts and Giardia cysts in cistern water in the U.S. Virgin Islands. Water Research, 30 (1), 208-216. Cunliffe, D. A. (1998). Guidance on the use of rainwater tanks Rundle Mall SA National Environmental Health Forum.30 Dillaha, T. A., & Zolan, W. J. (1985). Rain water catchment water quality in Micronesia. Water Research, 19 (6), 741-746.

PAGE 159

138 E.P.A. (1992). "Specifications and Guidance for Cont aminant-Free Samples Containers" U.S. Government Printing Office.28 E.P.A. (2002a). Code Of Federal Regulation Title 40-Protection of the Environment: Part141 National Primary Drinking Water Regulations United States Government Printing Office.228 E.P.A. (2002b). Code Of Federal Regulation Title 40-Protection of the Environment; Part 143 National Secondary Drinking Water Regulations : United States Government Printing Office.184 E.P.A. (2005). The Drinking Water Contam inant Candidate List : --A Source of Priority Contaminants for the Drinking Water Program : United States Government Printing Office.6 Ebi, K. L., Lewis, N. D., & Corvalan, C. (2006). Climate Variability and Change and Their Potential Health Effects in Small Island States: Information for Adaptation Planning in the Health Sector. Environmental Health Perspectives, 114 (12), 19571963. Eisenberg, J. N. S., Desai, M. A., Levy, K., Bates, S. J., Song, L., Naumoff, K., & Scott, J. C. (2007). Environmental Determinants of Infectious Disease: A Framework for Tracking Causal Links and Gu iding Public Health Research. Environmental Health Perspectives, 115 (8), 1216-1223. Eisenberg, J. N. S., Seto, E. Y. W., John M. Colford, J., Olivieri, A., & Spear, R. C. (1998). An Analysis of the Milwaukee Cryptosporidiosis Outbreak Based on a Dynamic Model of the Infection Process. Epidemiology, 9 (3 ), 255-263. Evans, C. A., Coombes, P. J., & Dunstan, R. H. (2006). Wind, rain and bacteria: The effect of weather on the microbial composition of r oof-harvested rainwater. Water Research, 40 (1), 37-44. Evans, C. A., Coombes, P. J., Dunstan, R. H., & Harrison, T. (2007). Indentifying the major influences on the microbial compositi on of roof harvested rainwater and the implication for water quality. Water Science and Technology, 55 (4), 245-253. Fernald, E. A., & Purdem, E. D. (Eds.). (1998). Water resources Atlas of Florida Tallahassee: Institute of Scie nce and Public Affairs 312 Florida Department of Envi ronmental Protection. (2002). The Florida Stormwater, Erosion and Sedimentation Control Inspector Manual Frigon, N. L., & Mathews, D. (1997). Practical Guide to Experimental Design New York John Wiley & Sons 342

PAGE 160

139 Gaffield, S. J., Goo, R. L., Richards, L. A., & Jackson, R. J. (2003). Public health effects of inadequately managed stormwater runoff. American Journal of Public Health, 93 (9), 1527. Galloway, J. N., Savoie, D. L., Keene, W. C., & Prospero, J. M. (1993). The temporal and spatial variability of scavengi ng ratios for NSS sulfate, nitrate, methanesulfonate and sodium in the At mosphere over the North Atlantic Ocean. Atmospheric Environment. Part A. General Topics, 27 (2), 235-250. Galloway, J. N., Thornton, D. J., Norton, S. A., Volchok, H. L., & McLean, R. A. N. (1982). Trace metals in atmospheric de position: A review and assessment. Atmospheric Environment (1967), 16 (7), 1677-1700. Garcia, R., del Torres Ma, C., Padilla, H., Be lmont, R., Azpra, E., Arcega-Cabrera, F., & Baez, A. (2006). Measurement of chemical elements in rain from Rancho Viejo, a rural wooded area in the St ate of Mexico, Mexico. Atmospheric Environment, 40 (32), 6088-6100. Glavas, S., & Moschonas, N. (2002). Origin of observed acidic-alkal ine rains in a wetonly precipitation study in a Mediterran ean coastal site, Patras, Greece. Atmospheric Environment, 36 (19), 3089-3099. Gobel, P., Dierkes, C., & Coldewey, W. G. (2007). Storm water runoff concentration matrix for urban areas. Journal of Contaminant Hydrology, 91 (1-2), 26-42. Gould, J. E. (1999). Is rainwater safe to drink? A review of recent findings. Paper presented at the 9th International Rainwater Catchment Systems., Petrolina, Brazil. Guentzel, J. L., Landing, W. M., Gill, G. A., & Pollman, C. D. (2001). Processes Influencing Rainfall Deposition of Mercury in Florida. Environ. Sci. Technol., 35 (5), 863-873. Haas, C. N., Crockett, C. S., Rose, J. B., Ge rba, C. P., & Fazil, A. M. (1996). Assessing the risk posed by oocysts in drinking water. American Water Wo rks Association. Journal, 88 (9), 131-136. Haas, C. N., Rose, J. B., & Regli, S. (1991) Risk Assessment and Control of Waterborne Giardiasis. American Journal of Public Health, 81 (6), 709-713. Hee, S. S. Q. (Ed.). (1993). Biological Monitoring New York: Van Nostrand Reinhold 650 Herrmann, T., & Schmida, U. (2000). Rainwate r utilisation in Germany: efficiency, dimensioning, hydraulic and environmental aspects. Urban Water, 1 (4), 307-316.

PAGE 161

140 Herut, B., Starinsky, A., Katz, A., & Rose nfeld, D. (2000). Relationship between the acidity and chemical composition of ra inwater and climatological conditions along a transition zone between large dese rts and Mediterranean climate, Israel. Atmospheric Environment, 34 (8), 1281-1292. Heymann, D. L. (Ed.). (2004). Control of Communicable Disease Manual (18 ed.). Washington, D.C.: American Public Health Association.698 Heyworth, J. S., Maynard, E. J., & Cunliffe D. A. (1998). Who consumes what potable water consuption in South Australia. Water, 25 (1), 9-13. Hutson, S. S., Barber, N. L., Kenny, J. K., Lins ey, K. S., Lumia D. S., & Maupin, M. A. (2004). Estimated Use of Water in the United States in 2000 (Vol. U.S. Geological Survey circular 1268). Reston,Va: US Department of the Interior 42 Hydrologic Evaluation Section. (2002). Saltwater Intrusion and the Minimum Aquifer Level in the Southern Water Use Caution Area : Southwest Florida Water Management District.48 Jensen, J. N. (2003). A Problem-Solving Approach to Aquatic Chemistry New York: John Wiley & Sons, Inc.585 Johnson, K. (2009, June,29). It's Now Le gal to Catch a Raindrop in Colorado New York Times Jones, A. M., & Harrison, R. M. (2004). The effects of meteorological factors on atmospheric bioaerosol concentrations--a review. Science of The Total Environment, 326 (1-3), 151-180. Kachigan, S. K. (1991). Multivariate Statistical Analysis (2 ed.). New York Raduis Press.303 Kaya, G., & Tuncel, G. (1997). Trace elemen t and major ion composition of wet and dry depositon in Ankara, Turkey. Atmospheric Environment, 31 (23), 3985-3998. Kieber, R. J., Hardison, D. R., Whitehead, R. F., & Willey, J. D. (2003). Photochemical Production of Fe(II) in Rainwater. Environ. Sci. Technol., 37 (20), 4610-4616. Kieber, R. J., Skrabal, S. A., Smith, B. J ., & Willey, J. D. (2005). Organic Complexation of Fe(II) and Its Impact on the Redox Cycling of Iron in Rain. Environ. Sci. Technol., 39 (6), 1576-1583. Kieber, R. J., Skrabal, S. A., Smith, C ., & Willey, J. D. (2004). Redox Speciation of Copper in Rainwater: Temporal Vari ability and Atmospheric Deposition. Environ. Sci. Technol., 38 (13), 3587-3594.

PAGE 162

141 Kieber, R. J., Willey, J. D., & Zvalaren, S. D. (2002). Chromium Speciation in Rainwater: Temporal Variabil ity and Atmospheric Deposition. Environ. Sci. Technol., 36 (24), 5321-5327. Leeden, F. v. d., Troise, F. L., & Todd, D. K. (1990). The Water Encyclopedia (2 ed.). Chelsa Lewis Publishers.808 Liaw, C.-H., & Tsai, Y.-L. (2004). Optimum storage volume of rooftop rain water harvesting systems for domestic use. Journal of the American Water Resources Association, 40 (4), 901-912. Louvar, J. F., & Louvar, B. D. (1998). Health and Environmental Risk Analysis:Fundamentals with Applications Upper Saddle Creek Prentice-Hall 678 Loye-Pilot, M. D., & Morel li, J. (1988). Fluctuations of ionic composition of precipitations collected in Corsica relate d to changes in the origins of incoming aerosols. Journal of Aerosol Science, 19 (5), 577-585. Luidold, S., & Antrekowitsch., H. (2 007). Hydrogen as A Reducing Agent: Thermodynamic Possibilities. Journal of the Minerals Metals and Materials Society, 59 (10), 58 62. Lye, D. J. (2002). Health risks associated with consumption of untreated water from household roof catchment systems. Journal of the American Water Resources Association, 38 (5), 1301 -1307. Lyubenova, T. S., Matteucci, F., Costa, A., Dondi, M., & Carda, J. (2009). Ceramic pigments with sphene structure obtained by both sprayand freeze-drying techniques. [doi: DOI : 10.1016/j.powtec.2009.01.020]. Powder Technology, 193 (1), 1-5. Manahan, S. E. (1991). Toxicology Chemistry: a guide to Toxic Substances in Chemistry Chelsea Lewis Publishers Inc. Manahan, S. E. (Ed.). (1990). Environmental Chemistry (4 ed.). Boston Lewis Publisher 612 Manahan, S. E. (Ed.). (1994). Environmental Chemistry (6 ed.). Boca Raton Lewis Publishers 811 Marsalek, J., Brownlee, B., Mayer, T., Lawal, S., & Larkin, G. (1997). Heavy Metals and PAHs in Stormwater Runoff from the Skyway Bridge in Burlington, Ontario. Water Quality Research Journal of Canada, 32 815827.

PAGE 163

142 McMillan, W. G. (1960). Surface Chemistry. Annual Review of Physical Chemistry, 11 (1), 449-476. Mitchell, V. (2006). Applying Integrated Ur ban Water Management Concepts: A Review of Australian Experience. Environmental Management, 37 (5), 589-605. Mudgal, P. K., Bansal, S. P., & Gupta, K. S. (2007). Kinetics of atmospheric oxidation of nitrous acid by oxygen in aqueous medium. Atmospheric Environment, 41 (19), 4097-4105. National Climatic Data Center. (2009). COOPID 84797. Retrieved Fed 20, from NOAA: www.ncdc.noaa.gov/oa/ncdc.html Ness, S. (1994). Surface and Dermal Monitoring for Toxic Exposures New York: John Wiley & Sons Inc. NOAA. (2005). Monthly /Annual Normals for Tampa Retrieved from http://www.srh.noaa.gov/tbw/html/tb w/climate/tpa/TPAANNNorm.htm Olem, H., & Berthouex, P. M. (1989). Acidic deposition and cistern drinking water supplies. Environ. Sci. Technol., 23 (3), 333-340. Parker, S. P., & Corbitt, R. A. (Eds.). (1993). McGrawHill encyclopedia of environmental science & engineering (3 ed.). New York McGraw-Hill.749 Petersen, G., Munthe, J., Pleijel, K., Bloxam, R., & Kumar, A. V. (1998). A comprehensive Eulerian modeling framew ork for airborne mercury species: Development and testing of the Tropospheric Chemistry module (TCM). Atmospheric Environment, 32 (5), 829-843. Railsback, L. B. (1997). Lower pH of acid ra in associated with lightning: evidence from sampling within 14 showers and storms in the Georgia Piedmont in summer 1996. Science of The Total Environment, 198 (3), 233-241. Robertson, L. J., Campbell, A. T., & Smith, H. V. (1992). Survival of Cryptosporidium parvum oocysts under various environmental pressures. Appl. Environ. Microbiol., 58 (11), 3494-3500. Sazakli, E., Alexopoulos, A., & Leotsinidis, M. (2007). Rainwater harvesting, quality assessment and utilization in Kefalonia Island, Greece. Water Research, 41 (9), 2039-2047. Schwarzenbach, R. P., Gschwend, P. M., & Imboden, D. M. (1993). Environmental organic chemistry New York: John Wiley & Sons.681

PAGE 164

143 Seinfeld, J. H. (1975). Air pollution: physical and chemical fundamentals New York: McGraw-Hill.523 Sincero, A., & Sincero, G. (1996). Environmental Engineer ing: A design Appproach Upper Saddle River NJ: Prentice Hall.795 Sirkin, R. M. (1994). Statistics for the Social Sciences (2 ed.). Thousand Oaks Sage Publications.569 Smith, S. K. (2005). Florida Population Growth : Past, Present and Future Gainesville University of Florida: Bureau of Economic and Business Research.40 Restricted Allocation § 3.2.1 (2007a). South Florida Water Manage ment District. (2007b). Restricted Allocation South Florida Water Management District.5 Southwest Florida Water Management Distri ct. (2006). Capital Cost Southwest Florida Water Management District Spinks, A. T., Dunstan, R. H., Harrison, T., Coombes, P., & Kuczera, G. (2006). Thermal inactivation of water-borne pathogenic a nd indicator bacteria at sub-boiling temperatures. Water Research, 40 (6), 1326-1332. Stites, D. P., Stobo, J. D., & Wells, J. V. (Eds.). (1987). Basic & Clinical Immunology (6 ed.). Norwalk Connecticu t: Appleton & Lange.734 Stumm, W., & Morgan, J. J. (1996). Aquatic Chemistry :Chemical Equilbria and Rates in Natural Waters (3 ed.). New York John Wiley & Sons Inc.1022 Tanner, P., & Fai, T. (2000). Small-Scale Hori zontal Variations in Ionic Concentrations of Bulk Deposition from Hong Kong. Water, Air, & Soil Pollution, 122 (3), 433448. Tanner, P., & Wong, A. (2000). Soluble Trace Metals and Major Ionic Species in the Bulk Deposition and Atmosphere of Hong Kong. Water, Air, & Soil Pollution, 122 (3), 261-279. Taylor, P. (2001). Experiences in water resources management: Survey of issues for water resources management in the Pacific subregion [Electronic version] Economic and social commission for Asia and the Pacific 236 -267 Tchobanoglous, G., & Schroeder, E. D. (1985). Water Quality Reading, Ma. : AddisonWesley Publishing Company.768

PAGE 165

144 Texas Natural Resource Conservation Commission. (2004). Rainwater Treatment Texas Water Development Board. (2005). The Texas Manual on Rainwater Harvesting (3 ed.). Austin 88 USGS. (2004). Public-supplied population, wa teruse, withdrawals, and transfers in Florida by county ,2000. Retrieved from http://www.fl.water.usgs.gov Utah Division of Water Rights. (2009). Wa ter Rights. Retrieved Feb, 02, 2009, from www.waterrights.utah.gov Viessman, W., & Lewis, G. L. (1996 ). Introduction to Hydrology (4 ed.). Upper Saddle River Pearson Education.760 Viessman, W., & Lewis, G. L. (2003 ). Introduction to Hydrology (5 ed.). Upper Saddle River Pearson Education.612 Viesssman, W., & Lewis, G. L. (2003 ). Introduction to Hydrology (5 ed.). Upper Saddle River Pearson Education.612 Wanielista, M., Kerten, R., & Eaglin, R. (1997). Hydrology: water quantity and quality control (2 ed.). New York John Wiley & Sons Inc. Waterborne Pathogens: manual of water supply practices (1999). Denver American Water Works Association.285 White, I., Falkland, T., Perez, P., Dray, A., Metutera, T., Metai, E., & Overmars, M. (2007). Challenges in freshwater ma nagement in low coral atolls. Journal of Cleaner Production, 15 (16), 1522-1528. Williams, P. T., Radojevic, M., & Clarke, A. G. (1988). Dissolution of trace metals from particles of industrial origin and its in fluence on the composition of rainwater. Atmospheric Environment (1967), 22 (7), 1433-1442. Wyngaarden, J. B., & Smith, L. H. (Eds.). (1985). Cecil Textbook of Medicine (17 ed.). Philadelphia W.B Saunders Company.2436 Yaziz, M. I., Gunting, H., Sapari, N., & Ghazal i, A. W. (1989). Variations in rainwater quality from roof catchments. Water Research, 23 (6), 761-765. Zartman, R. E., R H Ramsey III, & Huang, A. (2001). Variability of total and dissolved elements in stormwater runoff. Journal of Soil and Water Conservation, 56 (3), 263.

PAGE 166

145 Zhu, K., Zhang, L., Hart, W., Liu, M., & Che n, H. (2004). Quality issues in harvested rainwater in arid and semi-arid Lo ess Plateau of northern China. Journal of Arid Environments, 57 (4), 487-505.

PAGE 167

146 APPENDICES

PAGE 168

147 Appendix I: Standards and Analysis This section contains the vari ous outputs of the analysis. Table A-1: pH Descriptive Statistica l Analysis Material Surfaces S1-S5. S1 S2 S3 S4 S5 count29 29 29 29 29 mean0.0237 0.7054 0.8857 -0.4098 -0.3701 sample variance0.4644 0.8876 1.0315 0.6366 0.6404 sample standard deviation0.6814 0.9421 1.0156 0.7979 0.8003 minimum-1.61 -1.08 -1.11 -2.6 -1.78 maximum1.3 2.37 2.64 0.81 1.13 range2.91 3.45 3.75 3.41 2.91 normal curve GOF p-value.7023 .0100 .0376 .0258 .8013 chi-square(df=3)1.41 11.34 8.45 9.28 1.00 E4.83 4.83 4.83 4.83 4.83 O(-0.97)4 4 3 6 6 O(-0.43)5 5 7 3 4 O(+0.00)5 10 8 1 4 O(+0.43)6 2 3 7 4 O(+0.97)3 1 1 9 5 O(inf.)6 7 7 3 6

PAGE 169

148 Appendix I (Continued) Figure A-1: pH Frequency Plot Analysis of Material Surface -3-2-10123 S1 Sample pH -Control pH -3-2-10123 S2 Sample pH -Control pH -3-2-10123 S3 Sample pH -Control pH -3-2-10123 S4 Sample pH -Control pH -3-2-10123 S5 Sample pH -Control pH

PAGE 170

149 Appendix I (Continued) Table A-2: pH Paired T-Test Analysis S1 and Control. 0.000000 hypothesized value 6.380690 mean S1 6.357034 mean ConpH 0.023655 mean difference (S1 ConpH) 0.681438 std. dev. 0.126540 std. error 29 n 28 df 0.19 t .8531 p-value (two-tailed) -0.235550 confidence interval 95.% lower 0.282860 confidence interval 95.% upper 0.259205 half-width Table A-3: pH Paired T-Test Analysis S2 and Control. 0.000000 hypothesized value 7.062414 mean S2 6.357034 mean ConpH 0.705379 mean difference (S2 ConpH) 0.942127 std. dev. 0.174949 std. error 29 n 28 df 4.03 t .0004 p-value (two-tailed) 0.347013 confidence interval 95.% lower 1.063745 confidence interval 95.% upper 0.358366 half-width

PAGE 171

150 Appendix I (Continued) Table A-4: pH Paired T-Test Analysis S3 and Control. 0.000000 hypothesized value 7.242759 mean S3 6.357034 mean ConpH 0.885724 mean difference (S3 ConpH) 1.015612 std. dev. 0.188594 std. error 29 n 28 df 4.70 t .0001 p-value (two-tailed) 0.499406 confidence interval 95.% lower 1.272042 confidence interval 95.% upper 0.386318 half-width Table A-5: pH Paired T-Test Analysis S4 and Control. 0.000000 hypothesized value 5.947241 mean S4 6.357034 mean ConpH -0.409793 mean difference (S4 ConpH) 0.797875 std. dev. 0.148162 std. error 29 n 28 df -2.77 t .0099 p-value (two-tailed) -0.713288 confidence interval 95.% lower -0.106298 confidence interval 95.% upper 0.303495 half-width

PAGE 172

151 Appendix I (Continued) Table A-6: pH Paired T-Test Analysis S5 and Control. 0.000000 hypothesized value 5.986897 mean S5 6.357034 mean ConpH -0.370138 mean difference (S5 ConpH) 0.800264 std. dev. 0.148605 std. error 29 n 28 df -2.49 t .0189 p-value (two-tailed) -0.674542 confidence interval 95.% lower -0.065734 confidence interval 95.% upper 0.304404 half-width Table A-7: pH Wilcoxon Analysis S1-Control. variables:S1 ConpH 234sum of positive ranks 201sum of negative ranks 29 n 217.50 expected value 46.21 standard deviation 0.36 z, corrected for ties .7211 p-value (two-tailed)

PAGE 173

152 Appendix I (Continued) Table A-8: pH Wilcoxon Analysis S2-Control. variables:S2 ConpH 383sum of positive ranks 52sum of negative ranks 29 n 217.50 expected value 46.21 standard deviation 3.58 z, corrected for ties .0003 p-value (two-tailed) Table A-9: pH Wilcoxon Analysis S3-Control. variables:S3 ConpH 391sum of positive ranks 44sum of negative ranks 29 n 217.50 expected value 46.18 standard deviation 3.76 z, corrected for ties .0002 p-value (two-tailed) Table A-10: pH Wilcoxon Analysis S4-Control. variables:S4 ConpH 99sum of positive ranks 307sum of negative ranks 28 n 203.00 expected value 43.78 standard deviation -2.38 z, corrected for ties .0175 p-value (two-tailed)

PAGE 174

153 Appendix I (Continued) Table A-11: pH Wilcoxon Analysis S5-Control. variables:S5 ConpH 121.5sum of positive ranks 313.5sum of negative ranks 29 n 217.50 expected value 46.21 standard deviation -2.08 z, corrected for ties .0378 p-value (two-tailed) Table A-12: TDS Descriptive Statistical Analysis Material Surfaces S1-S5. S1 S2 S3 S4 S5 count29 29 29 29 29 mean-10.637931 0.155172 10.413793 -8.293103 -7.534483 sample variance140.980296 67.805419 82.965517 106.098522 75.891626 sample standard deviation 11.873512 8.234405 9.108541 10.300414 8.711580 minimum-44 -20 -14 -42 -33 maximum0.5 15 33 2 4 range44.5 35 47 44 37 normal curve GOF p-value2.82E-07 .3230 .3230 .0001 .0012 chi-square(df=3)33.28 3.48 3.48 21.28 15.90 E4.83 4.83 4.83 4.83 4.83 O(-0.97)4 5 4 4 3 O(-0.43)2 3 4 1 1 O(+0.00)3 4 8 2 5 O(+0.43)4 7 6 10 10 O(+0.97)16 7 3 11 9 O(inf.)0 3 4 1 1

PAGE 175

154 Appendix I (Continued) Figure A-2: TDS Frequency Plot Analysis of Material -50-40-30-20-1001020304050 S1 Sample TDS -Control TDS -50-40-30-20-1001020304050 S2 Sample TDS -Control TDS -50-40-30-20-1001020304050 S3 Sample TDS -Control TDS -50-40-30-20-1001020304050 S4 Sample TDS -Control TDS -50-40-30-20-1001020304050 S5 Sample TDS -Control TDS

PAGE 176

155 Appendix I (Continued) Table A-13: TDS Paired T-Test Analysis S1 and Control. 0.0000 hypothesized value 13.4655 mean S1 24.1034 mean ConTDS -10.6379 mean difference (S1 ConTDS) 11.8735 std. dev. 2.2049 std. error 29 n 28 df -4.82 t 4.48E-05 p-value (two-tailed) -15.1544 confidence interval 95.% lower -6.1215 confidence interval 95.% upper 4.5164 half-width Table A-14: TDS Paired T-Test Analysis S2 and Control. 0.0000 hypothesized value 24.2586 mean S2 24.1034 mean ConTDS 0.1552 mean difference (S2 ConTDS) 8.2344 std. dev. 1.5291 std. error 29 n 28 df 0.10 t .9199 p-value (two-tailed) -2.9770 confidence interval 95.% lower 3.2874 confidence interval 95.% upper 3.1322 half-width

PAGE 177

156 Appendix I (Continued) Table A-15: TDS Paired T-Test Analysis S3 and Control. 0.000 hypothesized value 34.517 mean S3 24.103 mean ConTDS 10.414 mean difference (S3 ConTDS) 9.109 std. dev. 1.691 std. error 29 n 28 df 6.16 t 1.20E-06 p-value (two-tailed) 6.949 confidence interval 95.% lower 13.878 confidence interval 95.% upper 3.465 half-width Table A-16: TDS Paired T-Test Analysis S4 and Control. 0.0000 hypothesized value 15.8103 mean S4 24.1034 mean ConTDS -8.2931 mean difference (S4 ConTDS) 10.3004 std. dev. 1.9127 std. error 29 n 28 df -4.34 t .0002 p-value (two-tailed) -12.2112 confidence interval 95.% lower -4.3750 confidence interval 95.% upper 3.9181 half-width

PAGE 178

157 Appendix I (Continued) Table A-17: TDS Paired T-Test Analysis S5 and Control. 0.0000 hypothesized value 16.5690 mean S5 24.1034 mean ConTDS -7.5345 mean difference (S5 ConTDS) 8.7116 std. dev. 1.6177 std. error 29 n 28 df -4.66 t .0001 p-value (two-tailed) -10.8482 confidence interval 95.% lower -4.2208 confidence interval 95.% upper 3.3137 half-width Table A-18: TDS Wilcoxon Analysis S1-Control. variables:S1 ConTDS 1sum of positive ranks 434sum of negative ranks 29 n 217.50 expected value 43.51 standard deviation -4.98 z, corrected for ties 6.48E-07 p-value (two-tailed)

PAGE 179

158 Appendix I (Continued) Table A-19: TDS Wilcoxon Analysis S2-Control. variables:S2 ConTDS 194sum of positive ranks 157sum of negative ranks 26 n 175.50 expected value 38.72 standard deviation 0.48 z, corrected for ties .6328 p-value (two-tailed) Table A-20: TDS Wilcoxon Analysis S3-Control. variables:S3 ConTDS 382sum of positive ranks 24sum of negative ranks 28 n 203.00 expected value 42.74 standard deviation 4.19 z, corrected for ties 2.81E-05 p-value (two-tailed) Table A-21: TDS Wilcoxon Analysis S4-Control. variables:S4 ConTDS 6sum of positive ranks 429sum of negative ranks 29 n 217.50 expected value 45.10 standard deviation -4.69 z, corrected for ties 2.73E-06 p-value (two-tailed)

PAGE 180

159 Appendix I (Continued) Table A-22: TDS Wilcoxon Analysis S5-Control. variables:S5 ConTDS 10.5sum of positive ranks 395.5sum of negative ranks 28 n 203.00 expected value 42.56 standard deviation -4.52 z, corrected for ties 6.10E-06 p-value (two-tailed) Table A-23: Zinc Descriptive Statistical Analysis Material Surfaces S1-S5. S1 S2 S3 S4 S5 count30 30 30 30 30 mean-0.099563 -0.113567 -0.054190 -0.035647 1.122243 sample variance0.021732 0.027803 0.022279 0.021365 0.944698 sample standard deviation0.147419 0.166741 0.149261 0.146169 0.971956 minimum-0.5081 -0.5703 -0.4312 -0.4122 -0.0016 maximum0.0149 0.0175 0.3535 0.1975 3.3669 range0.523 0.5878 0.7847 0.6097 3.3685 normal curve GOF p-value6.90E-10 2.83E-08 9.10E-08 .0035 9.37E-07 chi-square(df=3)45.60 38.00 35.60 13.60 30.80 E5.00 5.00 5.00 5.00 5.00 O(-0.97)6 4 3 4 1 O(-0.43)1 3 3 3 15 O(+0.00)1 1 2 2 7 O(+0.43)4 5 17 12 0 O(+0.97)18 17 4 6 2 O(inf.)0 0 1 3 5

PAGE 181

160 Appendix I (Continued) Figure A-3: Zinc Fre quency Plot Analysis of Material Surface -0.6-0.4-0.200.20.40.6 S1 Sample Zn -Control Zn -0.8-0.6-0.4-0.21E-150.20.40.6 S2 Sample Zn -Control Zn -0.6-0.4-0.21E-150.20.40.6 S3 Sample Zn -Control Zn -0.6-0.4-0.2-1E-150.20.40.6 S4 Sample Zn -Control Zn -4-3-2-101234 S5 Sample Zn -Control Zn

PAGE 182

161 Appendix I (Continued) Table A-24: Zinc Paired T-Test Analysis S1 and Control. 0.0000000 hypothesized value 0.0743600 mean S1 0.1739233 mean ConZn -0.0995633 mean difference (S1 ConZn) 0.1474187 std. dev. 0.0269149 std. error 30 n 29 df -3.70 t .0009 p-value (two-tailed) -0.1546104 confidence interval 95.% lower -0.0445163 confidence interval 95.% upper 0.0550471 half-width Table A-25: Zinc Paired T-Test Analysis S2 and Control. 0.0000000 hypothesized value 0.0603567 mean S2 0.1739233 mean ConZn -0.1135667 mean difference (S2 ConZn) 0.1667415 std. dev. 0.0304427 std. error 30 n 29 df -3.73 t .0008 p-value (two-tailed) -0.1758289 confidence interval 95.% lower -0.0513044 confidence interval 95.% upper 0.0622623 half-width

PAGE 183

162 Appendix I (Continued) Table A-26: Zinc Paired T-Test Analysis S3 and Control. 0.0000000 hypothesized value 0.1197333 mean S3 0.1739233 mean ConZn -0.0541900 mean difference (S3 ConZn) 0.1492615 std. dev. 0.0272513 std. error 30 n 29 df -1.99 t .0563 p-value (two-tailed) -0.1099252 confidence interval 95.% lower 0.0015452 confidence interval 95.% upper 0.0557352 half-width Table A-27: Zinc Paired T-Test Analysis S4 and Control. 0.0000000 hypothesized value 0.1382767 mean S4 0.1739233 mean ConZn -0.0356467 mean difference (S4 ConZn) 0.1461687 std. dev. 0.0266866 std. error 30 n 29 df -1.34 t .1920 p-value (two-tailed) -0.0902270 confidence interval 95.% lower 0.0189336 confidence interval 95.% upper 0.0545803 half-width

PAGE 184

163 Appendix I (Continued) Table A-28: Zinc Paired T-Test Analysis S5 and Control. 0.0000000 hypothesized value 1.2961667 mean S5 0.1739233 mean ConZn 1.1222433 mean difference (S5 ConZn) 0.9719555 std. dev. 0.1774540 std. error 30 n 29 df 6.32 t 6.57E-07 p-value (two-tailed) 0.7593092 confidence interval 95.% lower 1.4851775 confidence interval 95.% upper 0.3629342 half-width Table A-29: Zinc Wilcoxon Analysis S1-Control. variables:S1 ConZn 18sum of positive ranks 447sum of negative ranks 30 n 232.50 expected value 48.62 standard deviation -4.41 z 1.02E-05 p-value (two-tailed) Table A-30: Zinc Wilcoxon Analysis S2-Control. variables:S2 ConZn 16sum of positive ranks 449sum of negative ranks 30 n 232.50 expected value 48.62 standard deviation -4.45 z 8.47E-06 p-value (two-tailed)

PAGE 185

164 Appendix I (Continued) Table A-31: Zinc Wilcoxon Analysis S3-Control. variables:S3 ConZn 99sum of positive ranks 366sum of negative ranks 30 n 232.50 expected value 48.62 standard deviation -2.75 z .0060 p-value (two-tailed) Table A-32: Zinc Wilcoxon Analysis S4-Control. variables:S4 ConZn 213sum of positive ranks 252sum of negative ranks 30 n 232.50 expected value 48.62 standard deviation -0.40 z .6884 p-value (two-tailed) Table A-33: Zinc Wilcoxon Analysis S5-Control. variables:S5 ConZn 464sum of positive ranks 1sum of negative ranks 30 n 232.50 expected value 48.62 standard deviation 4.76 z 1.92E-06 p-value (two-tailed)

PAGE 186

165 Appendix I (Continued) Table A-34: Lead Descriptive Statistical Analysis Material Surfaces S1-S5. S1 S2 S3 S4 S5 count30 30 30 30 30 mean-0.000317 -0.000460 -0.000370 -0.000630 -0.000823 sample variance0.000005 0.000003 0.000005 0.000004 0.000005 sample standard deviation0.002306 0.001659 0.002262 0.002009 0.002290 minimum-0.0054 -0.0043 -0.005 -0.0056 -0.007 maximum0.004 0.0027 0.0058 0.0022 0.0028 range0.0094 0.007 0.0108 0.0078 0.0098 normal curve GOF p-value.1870 .0267 .3618 .4936 .2615 chi-square(df=3)4.80 9.20 3.20 2.40 4.00 E5.00 5.00 5.00 5.00 5.00 O(-0.97)6 5 4 4 4 O(-0.43)4 3 5 5 2 O(+0.00)3 4 6 5 8 O(+0.43)4 4 8 4 6 O(+0.97)9 11 3 8 5 O(inf.)4 3 4 4 5

PAGE 187

166 Appendix I (Continued) Figure A-4: Lead Frequency Plot Analysis of Material Surface -0.008-0.00400.0040.008 S1 Sample Pb -Control Pb -0.008-0.00400.0040.008 S2 Sample Pb -Control Pb -0.008-0.00400.0040.008 S3 Sample Pb -Control Pb -0.008-0.00400.0040.008 S4 Sample Pb -Control Pb -0.008-0.00400.0040.008 S5 Sample Pb -Control Pb

PAGE 188

167 Appendix I (Continued) Table A-35: Lead T-Test S1 Analysis and Control. 0.0000000 hypothesized value 0.0027833 mean S1 0.0031000 mean ConPb -0.0003167 mean difference (S1 ConPb) 0.0023059 std. dev. 0.0004210 std. error 30 n 29 df -0.75 t .4580 p-value (two-tailed) -0.0011777 confidence interval 95.% lower 0.0005444 confidence interval 95.% upper 0.0008610 half-width Table A-36: Lead Paired T-Test Analysis S2 and Control. 0.0000000 hypothesized value 0.0026400 mean S2 0.0031000 mean ConPb -0.0004600 mean difference (S2 ConPb) 0.0016587 std. dev. 0.0003028 std. error 30 n 29 df -1.52 t .1396 p-value (two-tailed) -0.0010794 confidence interval 95.% lower 0.0001594 confidence interval 95.% upper 0.0006194 half-width

PAGE 189

168 Appendix I (Continued) Table A-37: Lead Paired T-Test Analysis S3 and Control. 0.0000000 hypothesized value 0.0027300 mean S3 0.0031000 mean ConPb -0.0003700 mean difference (S3 ConPb) 0.0022622 std. dev. 0.0004130 std. error 30 n 29 df -0.90 t .3777 p-value (two-tailed) -0.0012147 confidence interval 95.% lower 0.0004747 confidence interval 95.% upper 0.0008447 half-width Table A-38: Lead Paired T-Test Analysis S4 and Control. 0.0000000 hypothesized value 0.0024700 mean S4 0.0031000 mean ConPb -0.0006300 mean difference (S4 ConPb) 0.0020093 std. dev. 0.0003668 std. error 30 n 29 df -1.72 t .0966 p-value (two-tailed) -0.0013803 confidence interval 95.% lower 0.0001203 confidence interval 95.% upper 0.0007503 half-width

PAGE 190

169 Appendix I (Continued) Table A-39: Lead Paired T-Test S5 Analysis and Control. 0.0000000 hypothesized value 0.0022767 mean S5 0.0031000 mean ConPb -0.0008233 mean difference (S5 ConPb) 0.0022901 std. dev. 0.0004181 std. error 30 n 29 df -1.97 t .0586 p-value (two-tailed) -0.0016785 confidence interval 95.% lower 0.0000318 confidence interval 95.% upper 0.0008551 half-width Table A-40: Lead Wilcoxon Analysis S1-Control. variables:S1 ConPb 206sum of positive ranks 259sum of negative ranks 30 n 232.50 expected value 48.59 standard deviation -0.55 z, corrected for ties .5855 p-value (two-tailed) Table A-41: Lead Wilcoxon Analysis S2-Control. variables:S2 ConPb 185sum of positive ranks 280sum of negative ranks 30 n 232.50 expected value 48.56 standard deviation -0.98 z, corrected for ties .3280 p-value (two-tailed)

PAGE 191

170 Appendix I (Continued) Table A-42: Lead Wilcoxon Analysis S3-Control. variables:S3 ConPb 165sum of positive ranks 270sum of negative ranks 29 n 217.50 expected value 46.21 standard deviation -1.14 z, corrected for ties .2560 p-value (two-tailed) Table A-43: Lead Wilcoxon Analysis S4-Control. variables:S4 ConPb 169sum of positive ranks 296sum of negative ranks 30 n 232.50 expected value 48.37 standard deviation -1.31 z, corrected for ties .1893 p-value (two-tailed) Table A-44: Lead Wilcoxon Analysis S5-Control. variables:S5 ConPb 139.5sum of positive ranks 295.5sum of negative ranks 29 n 217.50 expected value 46.02 standard deviation -1.69 z, corrected for ties .0901 p-value (two-tailed)

PAGE 192

171 Appendix I (Continued) Table A-45: Cadmium Descriptive Statistic al Analysis Material Surfaces S1-S5. S1 S2 S3 S4 S5 count31 31 31 31 31 mean-0.000010 0.000042 0.000010 0.000003 0.000023 sample variance0.000000 0.000000 0.000000 0.000000 0.000000 sample standard deviation 0.000137 0.000349 0.000108 0.000143 0.000171 minimum-0.0003 -0.0003 -0.0001 -0.0003 -0.0002 maximum0.0004 0.0018 0.0003 0.0003 0.0007 range0.0007 0.0021 0.0004 0.0006 0.0009 normal curve GOF p-value1.92E-05 8.36E-08 4.04E-05 .0019 .0019 chi-square(df=3)24.55 35.77 23.00 14.87 14.87 E5.17 5.17 5.17 5.17 5.17 O(-0.97)3 1 11 5 2 O(-0.43)8 3 0 4 9 O(+0.00)0 17 10 12 10 O(+0.43)14 6 0 0 0 O(+0.97)3 3 7 6 6 O(inf.)3 1 3 4 4

PAGE 193

172 Appendix I (Continued) Figure A-5: Cadmium Frequency Plot Analysis of Material Surface -0.0008-0.0004-2E-180.00040.0008 S1 Sample Cd -Control Cd -0.0008-0.0004-2E-180.00040.0008S2 Sample Cd -Control Cd -0.0008-0.0004-2E-180.00040.0008 S3 Sample Cd -Control Cd -0.0008-0.0004-2E-180.00040.0008 S4 Sample Cd -Control Cd -0.0008-0.0004-2E-180.00040.0008 S5 Sample Cd -Control Cd

PAGE 194

173 Appendix I (Continued) Table A-46: Cadmium Paired T-Te st Analysis S1 and Control. 0.0000000 hypothesized value 0.0001000 mean S1 0.0001097 mean ConCd -0.0000097 mean difference (S1 ConCd) 0.0001375 std. dev. 0.0000247 std. error 31 n 30 df -0.39 t .6979 p-value (two-tailed) -0.0000601 confidence interval 95.% lower 0.0000408 confidence interval 95.% upper 0.0000504 half-width Table A-47: Cadmium Paired T-Te st Analysis S2 and Control. 0.0000000 hypothesized value 0.0001516 mean S2 0.0001097 mean ConCd 0.0000419 mean difference (S2 ConCd) 0.0003491 std. dev. 0.0000627 std. error 31 n 30 df 0.67 t .5087 p-value (two-tailed) -0.0000861 confidence interval 95.% lower 0.0001700 confidence interval 95.% upper 0.0001280 half-width

PAGE 195

174 Appendix I (Continued) Table A-48: Cadmium Paired T-Te st Analysis S3 and Control. 0.0000000 hypothesized value 0.0001194 mean S3 0.0001097 mean ConCd 0.0000097 mean difference (S3 ConCd) 0.0001076 std. dev. 0.0000193 std. error 31 n 30 df 0.50 t .6201 p-value (two-tailed) -0.0000298 confidence interval 95.% lower 0.0000491 confidence interval 95.% upper 0.0000395 half-width Table A-49: Cadmium Paired T-Te st Analysis S4 and Control. 0.0000000 hypothesized value 0.0001129 mean S4 0.0001097 mean ConCd 0.0000032 mean difference (S4 ConCd) 0.0001426 std. dev. 0.0000256 std. error 31 n 30 df 0.13 t .9006 p-value (two-tailed) -0.0000491 confidence interval 95.% lower 0.0000555 confidence interval 95.% upper 0.0000523 half-width

PAGE 196

175 Appendix I (Continued) Table A-50: Cadmium Paired T-Te st Analysis S5 and Control. 0.0000000 hypothesized value 0.0001323 mean S5 0.0001097 mean ConCd 0.0000226 mean difference (S5 ConCd) 0.0001707 std. dev. 0.0000307 std. error 31 n 30 df 0.74 t .4671 p-value (two-tailed) -0.0000400 confidence interval 95.% lower 0.0000852 confidence interval 95.% upper 0.0000626 half-width Table A-51: Cadmium Wilcoxon Analysis S1-Control. variables:S1 ConCd 58.5sum of positive ranks 94.5sum of negative ranks 17 n 76.50 expected value 21.12 standard deviation -0.85 z .3942 p-value (two-tailed) Table A-52: Cadmium Wilcoxon Analysis S2-Control. variables:S2 ConCd 116.5sum of positive ranks 136.5sum of negative ranks 22 n 126.50 expected value 30.80 standard deviation -0.32 z .7454 p-value (two-tailed)

PAGE 197

176 Appendix I (Continued) Table A-53: Cadmium Wilcoxon Analysis S3-Control. variables:S3 ConCd 121sum of positive ranks 110sum of negative ranks 21 n 115.50 expected value 28.77 standard deviation 0.19 z .8484 p-value (two-tailed) Table A-54: Cadmium Wilcoxon Analysis S4-Control. variables:S4 ConCd 92.5sum of positive ranks 97.5sum of negative ranks 19 n 95.00 expected value 24.85 standard deviation -0.10 z .9199 p-value (two-tailed) Table A-55: Cadmium Wilcoxon Analysis S5-Control. variables:S5 ConCd 124sum of positive ranks 107sum of negative ranks 21 n 115.50 expected value 28.77 standard deviation 0.30 z .7677 p-value (two-tailed)

PAGE 198

177 Appendix I (Continued) Table A-56: Nickel Descriptive Statistic al Analysis Material Surfaces S1-S5. S1 S2 S3 S4 S5 count24 24 24 24 24 mean-0.000258 -0.000358 0.001208 -0.000354 -0.000196 sample variance0.000001 0.000001 0.000042 0.000001 0.000001 sample standard deviation 0.000842 0.000879 0.006481 0.001092 0.000981 minimum-0.0026 -0.002 -0.0032 -0.0041 -0.0033 maximum0.0011 0.0012 0.0312 0.0008 0.001 range0.0037 0.0032 0.0344 0.0049 0.0043 normal curve GOF p-value.5724 .8013 1.05E-16 .0117 .0186 chi-square(df=3)2.00 1.00 77.50 11.00 10.00 E4.00 4.00 4.00 4.00 4.00 O(-0.97)2 4 0 3 3 O(-0.43)6 4 2 1 1 O(+0.00)4 3 20 4 4 O(+0.43)4 5 1 8 9 O(+0.97)4 5 0 7 5 O(inf.)4 3 1 1 2

PAGE 199

178 Appendix I (Continued) Figure A-6: Nickel Frequency Plot Analysis of Material Surface -0.005-0.00257E-180.00250.005 S1 Sample Ni -Control Ni -0.005-0.00257E-180.00250.005 S2 Sample Ni -Control Ni -0.005-0.0025-7E-180.00250.005 S3 Sample Ni -Control Ni -0.005-0.0025-7E-180.00250.005 S4 Sample Ni -Control Ni -0.005-0.00251E-170.00250.005 S5 Sample Ni -Control Ni

PAGE 200

179 Appendix I (Continued) Table A-57: Nickel Paired T-Te st Analysis S1 and Control. 0.0000000 hypothesized value 0.0017000 mean S1 0.0019583 mean ConNi -0.0002583 mean difference (S1 ConNi) 0.0008423 std. dev. 0.0001719 std. error 24 n 23 df -1.50 t .1466 p-value (two-tailed) -0.0006140 confidence interval 95.% lower 0.0000973 confidence interval 95.% upper 0.0003557 half-width Table A-58: Nickel Paired T-Te st Analysis S2 and Control. 0.0000000 hypothesized value 0.0016000 mean S2 0.0019583 mean ConNi -0.0003583 mean difference (S2 ConNi) 0.0008792 std. dev. 0.0001795 std. error 24 n 23 df -2.00 t .0578 p-value (two-tailed) -0.0007296 confidence interval 95.% lower 0.0000129 confidence interval 95.% upper 0.0003712 half-width

PAGE 201

180 Appendix I (Continued) Table A-59: Nickel Paired T-Te st Analysis S3 and Control. 0.0000000 hypothesized value 0.0031667 mean S3 0.0019583 mean ConNi 0.0012083 mean difference (S3 ConNi) 0.0064808 std. dev. 0.0013229 std. error 24 n 23 df 0.91 t .3705 p-value (two-tailed) -0.0015283 confidence interval 95.% lower 0.0039449 confidence interval 95.% upper 0.0027366 half-width Table A-60: Nickel Paired T-Te st Analysis S4 and Control. 0.0000000 hypothesized value 0.0016042 mean S4 0.0019583 mean ConNi -0.0003542 mean difference (S4 ConNi) 0.0010919 std. dev. 0.0002229 std. error 24 n 23 df -1.59 t .1257 p-value (two-tailed) -0.0008152 confidence interval 95.% lower 0.0001069 confidence interval 95.% upper 0.0004611 half-width

PAGE 202

181 Appendix I (Continued) Table A-61: Nickel Paired T-Te st Analysis S5 and Control. 0.0000000 hypothesized value 0.0017625 mean S5 0.0019583 mean ConNi -0.0001958 mean difference (S5 ConNi) 0.0009809 std. dev. 0.0002002 std. error 24 n 23 df -0.98 t .3382 p-value (two-tailed) -0.0006100 confidence interval 95.% lower 0.0002184 confidence interval 95.% upper 0.0004142 half-width Table A-62: Nickel Wilcoxon Analysis S1-Control. variables:S1 ConNi 84.5sum of positive ranks 168.5sum of negative ranks 22 n 126.50 expected value 30.80 standard deviation -1.36 z .1727 p-value (two-tailed) Table A-63: Nickel Wilcoxon Analysis S2-Control. variables:S2 ConNi 73.5sum of positive ranks 179.5sum of negative ranks 22 n 126.50 expected value 30.80 standard deviation -1.72 z .0853 p-value (two-tailed)

PAGE 203

182 Appendix I (Continued) Table A-64: Nickel Wilcoxon Analysis S3-Control. variables:S3 ConNi 136sum of positive ranks 117sum of negative ranks 22 n 126.50 expected value 30.80 standard deviation 0.31 z .7578 p-value (two-tailed) Table A-65: Nickel Wilcoxon Analysis S4-Control. variables:S4 ConNi 105.5sum of positive ranks 170.5sum of negative ranks 23 n 138.00 expected value 32.88 standard deviation -0.99 z .3229 p-value (two-tailed) Table A-66: Nickel Wilcoxon Analysis S5-Control. variables:S5 ConNi 71.5sum of positive ranks 81.5sum of negative ranks 17 n 76.50 expected value 21.12 standard deviation -0.24 z .8129 p-value (two-tailed)

PAGE 204

183 Appendix I (Continued) Table A-67: Iron Descriptive Statistica l Analysis Material Surfaces S1-S5. S1 S2 S3 S4 S5 count31 31 31 31 31 mean-0.017303 -0.009474 0.002945 -0.007084 -0.013416 sample variance0.001091 0.001634 0.003030 0.001945 0.001424 sample standard deviation 0.033028 0.040419 0.055047 0.044098 0.037733 minimum-0.1273 -0.1261 -0.1 -0.1294 -0.116 maximum0.0387 0.0526 0.1499 0.0685 0.0411 range0.166 0.1787 0.2499 0.1979 0.1571 normal curve GOF p-value.2196 .0571 .6348 .4120 .1132 chi-square(df=3)4.42 7.52 1.71 2.87 5.97 E5.17 5.17 5.17 5.17 5.17 O(-0.97)4 6 5 3 5 O(-0.43)5 3 6 4 4 O(+0.00)7 3 7 7 4 O(+0.43)2 10 5 6 5 O(+0.97)8 6 3 7 10 O(inf.)5 3 5 4 3

PAGE 205

184 Appendix I (Continued) Figure A-7: Iron Freque ncy Plot Analysis of Material Surface -0.2-0.100.10.2 S1 Sample Fe -Control Fe -0.2-0.100.10.2 S2 Sample Fe -Control Fe -0.2-0.100.10.2 S3 Sample Fe -Control Fe -0.2-0.100.10.2 S4 Sample Fe -Control Fe -0.2-0.100.10.2 S5 Sample Fe -Control Fe

PAGE 206

185 Appendix I (Continued) Table A-68: Iron Paired T-Test Analysis S1 and Control. 0.0000000 hypothesized value 0.0532387 mean S1 0.0705419 mean ConFe -0.0173032 mean difference (S1 ConFe) 0.0330278 std. dev. 0.0059320 std. error 31 n 30 df -2.92 t .0066 p-value (two-tailed) -0.0294179 confidence interval 95.% lower -0.0051885 confidence interval 95.% upper 0.0121147 half-width Table A-69: Iron Paired T-Test Analysis S2 and Control. 0.0000000 hypothesized value 0.0610677 mean S2 0.0705419 mean ConFe -0.0094742 mean difference (S2 ConFe) 0.0404191 std. dev. 0.0072595 std. error 31 n 30 df -1.31 t .2018 p-value (two-tailed) -0.0243000 confidence interval 95.% lower 0.0053516 confidence interval 95.% upper 0.0148258 half-width

PAGE 207

186 Appendix I (Continued) Table A-70: Iron Paired T-Test Analysis S3 and Control. 0.0000000 hypothesized value 0.0734871 mean S3 0.0705419 mean ConFe 0.0029452 mean difference (S3 ConFe) 0.0550470 std. dev. 0.0098867 std. error 31 n 30 df 0.30 t .7678 p-value (two-tailed) -0.0172463 confidence interval 95.% lower 0.0231366 confidence interval 95.% upper 0.0201914 half-width Table A-71: Iron Paired T-Test Analysis S4 and Control. 0.0000000 hypothesized value 0.0634581 mean S4 0.0705419 mean ConFe -0.0070839 mean difference (S4 ConFe) 0.0440980 std. dev. 0.0079202 std. error 31 n 30 df -0.89 t .3782 p-value (two-tailed) -0.0232592 confidence interval 95.% lower 0.0090914 confidence interval 95.% upper 0.0161753 half-width

PAGE 208

187 Appendix I (Continued) Table A-72: Iron Paired T-Test Analysis S5 and Control. 0.0000000 hypothesized value 0.0571258 mean S5 0.0705419 mean ConFe -0.0134161 mean difference (S5 ConFe) 0.0377327 std. dev. 0.0067770 std. error 31 n 30 df -1.98 t .0570 p-value (two-tailed) -0.0272566 confidence interval 95.% lower 0.0004243 confidence interval 95.% upper 0.0138405 half-width Table A-73: Iron Wilcoxon Analysis S1-Control. variables:S1 ConFe 111sum of positive ranks 385sum of negative ranks 31 n 248.00 expected value 51.03 standard deviation -2.68 z .0073 p-value (two-tailed) Table A-74: Iron Wilcoxon Analysis S2-Control. variables:S1 ConFe 111sum of positive ranks 385sum of negative ranks 31 n 248.00 expected value 51.03 standard deviation -2.68 z .0073 p-value (two-tailed)

PAGE 209

188 Appendix I (Continued) Table A-75: Iron Wilcoxon Analysis S3-Control. variables:S3 ConFe 238sum of positive ranks 258sum of negative ranks 31 n 248.00 expected value 51.03 standard deviation -0.20 z .8446 p-value (two-tailed) Table A-76: Iron Wilcoxon Analysis S4-Control. variables:S4 ConFe 226.5sum of positive ranks 269.5sum of negative ranks 31 n 248.00 expected value 51.03 standard deviation -0.42 z .6735 p-value (two-tailed) Table A-77: Iron Wilcoxon Analysis S5-Control. variables:S5 ConFe 170sum of positive ranks 326sum of negative ranks 31 n 248.00 expected value 51.03 standard deviation -1.53 z .1264 p-value (two-tailed)

PAGE 210

189 Appendix I (Continued) Table A-78: Manganese Descriptive Statis tical Analysis Material Surfaces S1-S5. S1 S2 S3 S4 S5 count24 24 24 24 24 mean-0.002908 -0.003513 -0.002925 -0.002146 -0.000842 sample variance0.000108 0.000113 0.000134 0.000119 0.000160 sample standard deviation 0.010384 0.010634 0.011575 0.010909 0.012666 minimum-0.0501 -0.0518 -0.0527 -0.0491 -0.0503 maximum0.0015 0.001 0.0149 0.0163 0.0325 range0.0516 0.0528 0.0676 0.0654 0.0828 normal curve GOF p-value6.99E-15 6.99E-15 7.52E-13 1.14E-14 5.88E-13 chi-square(df=3)69.00 69.00 59.50 68.00 60.00 E4.00 4.00 4.00 4.00 4.00 O(-0.97)1 1 1 1 1 O(-0.43)1 1 2 2 1 O(+0.00)3 3 2 1 3 O(+0.43)19 19 18 19 18 O(+0.97)0 0 0 0 0 O(inf.)0 0 1 1 1

PAGE 211

190 Appendix I (Continued) Figure A-8: Manganese Frequency Plot Analysis of Material Surface -0.06-0.0300.030.06 S1 Sample Mn -Control Mn -0.06-0.0300.030.06 S2 Sample Mn -Control Mn -0.06-0.0300.030.06 S3 Sample Mn -Control Mn -0.06-0.0300.030.06 S4 Sample Mn -Control Mn -0.06-0.0300.030.06 S5 Sample Mn -Control Mn

PAGE 212

191 Appendix I (Continued) Table A-79: Manganese Paired T-Te st Analysis S1 and Control. 0.0000000 hypothesized value 0.0036042 mean S1 0.0065125 mean ConMn -0.0029083 mean difference (S1 ConMn) 0.0103835 std. dev. 0.0021195 std. error 24 n 23 df -1.37 t .1832 p-value (two-tailed) -0.0072929 confidence interval 95.% lower 0.0014762 confidence interval 95.% upper 0.0043846 half-width Table A-80: Manganese Paired T-Te st Analysis S2 and Control. 0.0000000 hypothesized value 0.0030000 mean S2 0.0065125 mean ConMn -0.0035125 mean difference (S2 ConMn) 0.0106340 std. dev. 0.0021706 std. error 24 n 23 df -1.62 t .1193 p-value (two-tailed) -0.0080028 confidence interval 95.% lower 0.0009778 confidence interval 95.% upper 0.0044903 half-width

PAGE 213

192 Appendix I (Continued) Table A-81: Manganese Paired T-Te st Analysis S3 and Control. 0.0000000 hypothesized value 0.0035875 mean S3 0.0065125 mean ConMn -0.0029250 mean difference (S3 ConMn) 0.0115752 std. dev. 0.0023628 std. error 24 n 23 df -1.24 t .2282 p-value (two-tailed) -0.0078128 confidence interval 95.% lower 0.0019628 confidence interval 95.% upper 0.0048878 half-width Table A-82: Manganese Paired T-Te st Analysis S4 and Control. 0.0000000 hypothesized value 0.0043667 mean S4 0.0065125 mean ConMn -0.0021458 mean difference (S4 ConMn) 0.0109091 std. dev. 0.0022268 std. error 24 n 23 df -0.96 t .3453 p-value (two-tailed) -0.0067524 confidence interval 95.% lower 0.0024607 confidence interval 95.% upper 0.0046065 half-width

PAGE 214

193 Appendix I (Continued) Table A-83: Manganese Paired T-Te st Analysis S5 and Control. 0.0000000 hypothesized value 0.0056708 mean S5 0.0065125 mean ConMn -0.0008417 mean difference (S5 ConMn) 0.0126659 std. dev. 0.0025854 std. error 24 n 23 df -0.33 t .7477 p-value (two-tailed) -0.0061900 confidence interval 95.% lower 0.0045067 confidence interval 95.% upper 0.0053484 half-width Table A-84: Manganese Wilc oxon Analysis S1-Control. variables:S1 ConMn 125.5sum of positive ranks 174.5sum of negative ranks 24 n 150.00 expected value 34.87 standard deviation -0.70 z, corrected for ties .4823 p-value (two-tailed) Table A-85: Manganese Wilc oxon Analysis S2-Control. variables:S2 ConMn 81sum of positive ranks 219sum of negative ranks 24 n 150.00 expected value 34.96 standard deviation -1.97 z, corrected for ties .0484 p-value (two-tailed)

PAGE 215

194 Appendix I (Continued) Table A-86: Manganese Wilc oxon Analysis S3-Control. variables:S3 ConMn 88sum of positive ranks 188sum of negative ranks 23 n 138.00 expected value 32.37 standard deviation -1.54 z, corrected for ties .1225 p-value (two-tailed) Table A-87: Manganese Wilc oxon Analysis S4-Control. variables:S4 ConMn 115sum of positive ranks 138sum of negative ranks 22 n 126.50 expected value 30.66 standard deviation -0.38 z, corrected for ties .7076 p-value (two-tailed) Table A-88: Manganese Wilc oxon Analysis S5-Control. variables:S5 ConMn 183sum of positive ranks 117sum of negative ranks 24 n 150.00 expected value 34.96 standard deviation 0.94 z, corrected for ties .3452 p-value (two-tailed)

PAGE 216

195 Appendix I (Continued) Table A-89: Chromium Descriptive Statis tical Analysis Material Surfaces S1-S5. S1 S2 S3 S4 S5 count31 31 31 31 31 mean-0.000142 0.000455 0.000648 0.000090 -0.000058 sample variance0.000000 0.000002 0.000000 0.000000 0.000000 sample standard deviation 0.000280 0.001304 0.000582 0.000393 0.000398 minimum-0.0009 -0.0007 -0.0004 -0.0009 -0.0011 maximum0.0005 0.0072 0.0022 0.001 0.0009 range0.0014 0.0079 0.0026 0.0019 0.002 normal curve GOF p-value.4782 5.76E-18 .4120 .7238 .0239 chi-square(df=3)2.48 83.39 2.87 1.32 9.45 E5.17 5.17 5.17 5.17 5.17 O(-0.97)4 0 5 5 4 O(-0.43)4 2 6 6 2 O(+0.00)6 24 7 4 8 O(+0.43)5 3 2 5 10 O(+0.97)8 1 6 7 4 O(inf.)4 1 5 4 3

PAGE 217

196 Appendix I (Continued) Figure A-9: Chromium Frequency Plot Analysis of Material Surface -0.008-0.00400.0040.008 S1 Sample Cr -Control Cr -0.008-0.00400.0040.008 S2 Sample Cr -Control Cr -0.008-0.00400.0040.008 S3 Sample Cr -Control Cr -0.008-0.00400.0040.008 S4 Sample Cr -Control Cr -0.008-0.00400.0040.008 S5 Sample Cr -Control Cr

PAGE 218

197 Appendix I (Continued) Table A-90: Chromium Paired T-Te st Analysis S1 and Control. 0.0000000 hypothesized value 0.0006129 mean S1 0.0007548 mean ConCr -0.0001419 mean difference (S1 ConCr) 0.0002802 std. dev. 0.0000503 std. error 31 n 30 df -2.82 t .0084 p-value (two-tailed) -0.0002447 confidence interval 95.% lower -0.0000392 confidence interval 95.% upper 0.0001028 half-width Table A-91: Chromium Paired T-Te st Analysis S2 and Control. 0.0000000 hypothesized value 0.0012097 mean S2 0.0007548 mean ConCr 0.0004548 mean difference (S2 ConCr) 0.0013038 std. dev. 0.0002342 std. error 31 n 30 df 1.94 t .0615 p-value (two-tailed) -0.0000234 confidence interval 95.% lower 0.0009331 confidence interval 95.% upper 0.0004782 half-width

PAGE 219

198 Appendix I (Continued) Table A-92: Chromium Paired T-Te st Analysis S3 and Control. 0.0000000 hypothesized value 0.0014032 mean S3 0.0007548 mean ConCr 0.0006484 mean difference (S3 ConCr) 0.0005819 std. dev. 0.0001045 std. error 31 n 30 df 6.20 t 7.89E-07 p-value (two-tailed) 0.0004350 confidence interval 95.% lower 0.0008618 confidence interval 95.% upper 0.0002134 half-width Table A-93: Chromium Paired T-Te st Analysis S4 and Control. 0.0000000 hypothesized value 0.0008452 mean S4 0.0007548 mean ConCr 0.0000903 mean difference (S4 ConCr) 0.0003927 std. dev. 0.0000705 std. error 31 n 30 df 1.28 t .2102 p-value (two-tailed) -0.0000537 confidence interval 95.% lower 0.0002344 confidence interval 95.% upper 0.0001441 half-width

PAGE 220

199 Appendix I (Continued) Table A-94: Chromium Paired T-Te st Analysis S5 and Control. 0.0000000 hypothesized value 0.0006968 mean S5 0.0007548 mean ConCr -0.0000581 mean difference (S5 ConCr) 0.0003981 std. dev. 0.0000715 std. error 31 n 30 df -0.81 t .4232 p-value (two-tailed) -0.0002041 confidence interval 95.% lower 0.0000880 confidence interval 95.% upper 0.0001460 half-width Table A-95: Chromium Wilc oxon Analysis S1-Control. variables:S1 ConCr 63.5sum of positive ranks 261.5sum of negative ranks 25 n 162.50 expected value 35.13 standard deviation -2.82 z, corrected for ties .0048 p-value (two-tailed) Table A-96: Chromium Wilc oxon Analysis S2-Control. variables:S2 ConCr 312sum of positive ranks 39sum of negative ranks 26 n 175.50 expected value 37.90 standard deviation 3.60 z, corrected for ties .0003 p-value (two-tailed)

PAGE 221

200 Appendix I (Continued) Table A-97: Chromium Wilc oxon Analysis S3-Control. variables:S3 ConCr 394.5sum of positive ranks 11.5sum of negative ranks 28 n 203.00 expected value 43.61 standard deviation 4.39 z, corrected for ties 1.13E-05 p-value (two-tailed) Table A-98: Chromium Wilc oxon Analysis S4-Control. variables:S4 ConCr 251sum of positive ranks 127sum of negative ranks 27 n 189.00 expected value 40.01 standard deviation 1.55 z, corrected for ties .1212 p-value (two-tailed) Table A-99: Chromium Wilc oxon Analysis S5-Control. variables:S5 ConCr 118sum of positive ranks 182sum of negative ranks 24 n 150.00 expected value 33.60 standard deviation -0.95 z, corrected for ties .3409 p-value (two-tailed)

PAGE 222

201 Appendix I (Continued) Table A-100: Copper Descriptive Statistica l Analysis Material Surfaces S1-S5. S1 S2 S3 S4 S5 count31 31 31 31 31 mean-0.000697 -0.000718 0.000284 -0.001119 -0.001168 sample variance0.000004 0.000002 0.000024 0.000007 0.000007 sample standard deviation 0.001884 0.001430 0.004879 0.002703 0.002712 minimum-0.0069 -0.0062 -0.0148 -0.0119 -0.0131 maximum0.0049 0.0014 0.0173 0.0024 0.0013 range0.0118 0.0076 0.0321 0.0143 0.0144 normal curve GOF p-value.0098 .0284 .0028 .0140 .0001 chi-square(df=3)11. 39 9.06 14.10 10.61 21.84 E5.17 5.17 5.17 5.17 5.17 O(-0.97)2 3 2 3 3 O(-0.43)5 4 4 2 0 O(+0.00)7 6 10 7 8 O(+0.43)11 4 10 7 10 O(+0.97)4 11 3 10 10 O(inf.)2 3 2 2 0

PAGE 223

202 Appendix I (Continued) Figure A-10: Copper Fre quency Plot Analysis -0.02-0.0100.010.02 S1 Sample Cu -Control Cu -0.02-0.0100.010.02 S2 Sample Cu -Control Cu -0.02-0.0100.010.02 S3 Sample Cu -Control Cu -0.02-0.0100.010.02 S4 Sample Cu -Control Cu -0.02-0.0100.010.02 S5 Sample Cu -Control Cu

PAGE 224

203 Appendix I (Continued) Table A-101: Copper Paired T-Test Analysis S1 and Control. 0.0000000 hypothesized value 0.0058903 mean S1 0.0065871 mean ConCu -0.0006968 mean difference (S1 ConCu) 0.0018837 std. dev. 0.0003383 std. error 31 n 30 df -2.06 t .0482 p-value (two-tailed) -0.0013877 confidence interval 95.% lower -0.0000058 confidence interval 95.% upper 0.0006909 half-width Table A-102: Copper Paired T-Test Analysis S2 and Control. 0.0000000 hypothesized value 0.0058687 mean S2 0.0065871 mean ConCu -0.0007184 mean difference (S2 ConCu) 0.0014299 std. dev. 0.0002568 std. error 31 n 30 df -2.80 t .0089 p-value (two-tailed) -0.0012429 confidence interval 95.% lower -0.0001939 confidence interval 95.% upper 0.0005245 half-width

PAGE 225

204 Appendix I (Continued) Table A-103: Copper Paired T-Test Analysis S3 and Control. 0.0000000 hypothesized value 0.0068710 mean S3 0.0065871 mean ConCu 0.0002839 mean difference (S3 ConCu) 0.0048785 std. dev. 0.0008762 std. error 31 n 30 df 0.32 t .7482 p-value (two-tailed) -0.0015056 confidence interval 95.% lower 0.0020733 confidence interval 95.% upper 0.0017895 half-width Table A-104: Copper Paired T-Test Analysis S4 and Control. 0.0000000 hypothesized value 0.0054681 mean S4 0.0065871 mean ConCu -0.0011190 mean difference (S4 ConCu) 0.0027034 std. dev. 0.0004855 std. error 31 n 30 df -2.30 t .0283 p-value (two-tailed) -0.0021106 confidence interval 95.% lower -0.0001274 confidence interval 95.% upper 0.0009916 half-width

PAGE 226

205 Appendix I (Continued) Table A-105: Copper Paired T-Test Analysis S5 and Control. Table A-106: Copper Wilcoxon Analysis S1-Control. variables:S1 ConCu 105sum of positive ranks 360sum of negative ranks 30 n 232.50 expected value 48.49 standard deviation -2.63 z, corrected for ties .0086 p-value (two-tailed) Table A-107: Copper Wilcoxon Analysis S2-Control. variables:S2 ConCu 104sum of positive ranks 361sum of negative ranks 30 n 232.50 expected value 48.34 standard deviation -2.66 z, corrected for ties .0079 p-value (two-tailed) 0.0000000 hypothesized value 0.0054194 mean S5 0.0065871 mean ConCu -0.0011677 mean difference (S5 ConCu) 0.0027121 std. dev. 0.0004871 std. error 31 n 30 df -2.40 t .0229 p-value (two-tailed) -0.0021626 confidence interval 95.% lower -0.0001729 confidence interval 95.% upper 0.0009948 half-width

PAGE 227

206 Appendix I (Continued) Table A-108: Copper TDS Anal ysis Wilcoxon S3-Control. variables:S3 ConCu 242sum of positive ranks 193sum of negative ranks 29 n 217.50 expected value 46.18 standard deviation 0.53 z, corrected for ties .5958 p-value (two-tailed) Table A-109: Copper TDS Anal ysis Wilcoxon S4-Control. variables:S4 ConCu 122sum of positive ranks 343sum of negative ranks 30 n 232.50 expected value 48.56 standard deviation -2.28 z, corrected for ties .0229 p-value (two-tailed) Table A-110: Copper TDS Anal ysis Wilcoxon S5-Control. variables:S5 ConCu 87sum of positive ranks 348sum of negative ranks 29 n 217.50 expected value 46.21 standard deviation -2.82 z, corrected for ties .0047 p-value (two-tailed)

PAGE 228

207 Appendix II: Model Development Hydraulic Model Conceptually the hydraulic model consists of two elements, one of which is a collection system and the othe r being a discharge system. The collection system is a gravity flow system and its design is ba sed upon Manning’s equation, while the discharge system is based upon Darcy’s equation. These ca lculations are averag es used prior to developing the Excel model to examin e the potential of such a system. Collection System The model requires that following data be collected or estimated from historical records: Community population = Population /home = Total daily demand = Total number of lots = Lots/ street = Total roof area = (usually not available) Average lot size = Home square footage = Roof area/ home = Rainfall data =

PAGE 229

208 Appendix II (Continued) Rainfall frequency data = Capacity of treatment plant/ system = Topographic Maps = The model requires that follo wing calculations be done: Calculations of availabl e volume to be collected Estimation of the total volume of water that is available /year Estimation of total number of homes assuming 2.5-person occupancy Area of Total Roof Surface, for example: Surface Area roof = No. Homes x 3041 sf/home (Temple Terrace) Per capita consumption rate Total volume of rainfall available from roof catchment Total rainfall received per year Annual volume = (annual rainfall f t/yr) total roof area x eff Annual consumptive demand Cons demand = average daily demand x 365 days/year x pop Percentage of demand Percentage = total volume available Total consumptive demand Assessment of rainfall volume to collect

PAGE 230

209 Appendix II (Continued) The model requires that followi ng constraints be determined: Constraints of volume to collect Capacity of treatment plant Economic capital investment Water quality Rainfall frequency analysis must be done and piping system selection (for example, based upon the data collected, 90% of the time rainfall events in this area are equal to or less than 1 inch-per-hour.) Unit volume = intensity (in/hr) x eff x roof area unit x 7.48 gal/cf Required capacity of latera l pipe to carry water. Qlateral = volumetric rate per uni t x connections per length Selection of pipe diameter Manning’s Equation Estimated Max Flow Rate (Equation B-1) (Equation B-2) (Equation B-3)

PAGE 231

210 Appendix II (Continued) Cost of Pipe in Suburban area $7.00/inch diameter-feet of length For example, assume that the connection included 12 homes and the length of run was 500 feet slope from the map is 7 feet in 500 feet and n = 0.015. The flow from the 12 homes would be: Q = 26.85 gpm x 12 = 322 gpm or 0.72 cfs (Equation B-4) The required pipe diameter is calculated from: Q = 30.86 (D)2.667 ( S)0.5 = 0.54ft = 6.5” or use an 8 inch pipe Cost @ $7.00/ inch-ft = $56/ft x 500 ft = $28,000 The same approach is used to determine main piping that connects the laterals to the storage and pumping stations. Total conne ctions, slope of pipe, and volume to be transported are established to dete rmine the diameter of the pipe. For example, assume the change in elev ation is 45 feet and there are 500 home connected to the main through laterals. The to tal flow for a 1 inch/ hour storm would be 13,425 gallons per minute and would require a 30-inch diameter main 6,500 feet long costing $1,365,000.

PAGE 232

211 Appendix II (Continued) Storage Storage is controlled by the capac ity of the treatment plant. Augmentation Demand = Daily Demand x augmentation fraction Storage = Volume recovered from selected Units Augmentation demand Pumping System Daily demand equals capacity of plant. Pumping Q = Plant capacity in gallons/ minute Total Horsepower required Pump Requirements for Force Main 16 ” (Equation B-5) Power = Q(gpm) x Total Head(static + dynamic) x Pump efficiency Cost Estimated cost of 59hp pump is $ 350,000. Note: Various configurations of the piping network strongly influence the Ca pital Cost of this type of project. NOTE: Variables for these hydraulic calcula tions can be found in Table B-1, below.

PAGE 233

212 Appendix II (Continued) Table B-1: Model Variables for the Hydraulic Calculations.

PAGE 234

213 Appendix II (Continued) Figure B-1: Rain Model Analyses for Demand

PAGE 235

A T A ppendix II T able B-2: M (Continue d M odel Cost S d ) chedule An a 214 a lysis.

PAGE 236

215 Appendix II (Continued) Table B-3: Model Estimated Cost and Payback Period Analysis.

PAGE 237

216 Appendix II (Continued) Table B-4: Model Loan Schedule Analysis.

PAGE 238

ABOUT THE AUTHOR Robert Carnahan Jr. has been the Ch air of the Hillsborough River Board and Technical Advisory Council for the past si x of eight years of the appointed council position. He is currently teaching environmenta l studies and business management as an adjunct professor at Eckerd College. Carnahan was a graduate of Nova South eastern University with a MBA and was inducted into Sigma Beta Delta honor society. He is a graduate of Eckerd College with a BA, with distinction, in management.