Temporal trend analysis of long-term monitoring data at karst springs, 2009

Citation
Temporal trend analysis of long-term monitoring data at karst springs, 2009

Material Information

Title:
Temporal trend analysis of long-term monitoring data at karst springs, 2009
Creator:
Herrington, Chris
Hiers, Scott. E.
Publisher:
City of Austin
Watershed Protection and Development Review Department
Publication Date:
Language:
English

Subjects

Subjects / Keywords:
Backdoor Spring (Travis County, Texas, United States) ( 30.25951, -97.8237 )
Barton Springs (Austin, Texas, United States) ( 30.263819, -97.771395 )
Cold Spring (Travis County, Texas, United States) ( 30.27963, -97.7805 )
Eliza Spring (Travis County, Texas, United States) ( 30.27963, -97.7805 )
Old Mill Spring (Travis County, Texas, United States) ( 30.26359, -97.76808 )
Upper Barton Spring (Travis County, Texas, United States) ( 30.26359, -97.77378 )
Geology ( local )
Genre:
Technical Report
serial ( sobekcm )
Location:
United States
Coordinates:
30.25951 x -97.8237
30.263819 x -97.771395
30.27963 x -97.7805
30.26359 x -97.76808
30.26359 x -97.77378

Notes

General Note:
Water quality is degrading over time at Main Barton Springs, Austin, Texas, as determined by multiple linear regression analyses (COA 2005). Similar techniques are applied to long-term monitoring data from 5 other Barton-related springs: Upper Barton Springs, Old Mill Springs, Eliza Springs, Backdoor Springs and Cold Springs. Categorical variables representing low-flow condition, laboratory method, filter fraction and analytical method were also used in the regression analysis. Varying temporal trends were observed at study springs. Due to variations in contributing watersheds, local influences, and aquifer mixing, observed temporal trends at other Barton complex springs did not always match observed trends at Main Barton Springs noted in previous analyses.
Restriction:
Open Access - Permission by Publisher
General Note:
See Extended description for more information.

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:
K26-04338 ( USFLDC DOI )
k26.4338 ( USFLDC Handle )
11528 ( karstportal - original NodeID )

USFLDC Membership

Aggregations:
Added automatically
Karst Information Portal

Postcard Information

Format:
serial

Downloads

This item has the following downloads:


Full Text
Description
Water quality is degrading over time at Main Barton
Springs, Austin, Texas, as determined by multiple linear
regression analyses (COA 2005). Similar techniques are
applied to long-term monitoring data from 5 other
Barton-related springs: Upper Barton Springs, Old Mill
Springs, Eliza Springs, Backdoor Springs and Cold Springs.
Categorical variables representing low-flow condition,
laboratory method, filter fraction and analytical method were
also used in the regression analysis. Varying temporal trends
were observed at study springs. Due to variations in
contributing watersheds, local influences, and aquifer
mixing, observed temporal trends at other Barton complex
springs did not always match observed trends at Main Barton
Springs noted in previous analyses.



PAGE 1

SR 10-06 Page 1 of 43 February 2010 Temporal Trend Analysis of Long -term Monitoring Data at Karst Springs, 2009 SR-10-06, February 2010 Chris Herrington, P.E. Scott Hiers, P.G. Environmental Resource Management Division Watershed Protection Department City of Austin Water quality is degrading over time at Main Bart on Springs, Austin, Texas, as determined by multiple linear regression analyses (COA 2005). Similar techniques are applied to long-term monitoring data from 5 other Barton-related sp rings: Upper Barton Springs, Old Mill Springs, Eliza Springs, Backdoor Springs and Cold Spri ngs. Categorical variables representing low-flow condition, laboratory method, filter fraction and analytical method were also used in the regression analysis. Varying temporal trends were ob served at study springs. Due to variations in contributing watersheds, local influences, and aquifer mixing, observed temporal trends at other Barton complex springs did not always ma tch observed trends at Main Barton Springs noted in previous analyses. Introduction The City of Austin (COA) in cooperation with th e United States Geological Survey (USGS) have conducted long-term water quality monitoring of sev eral karst springs in the Austin area. Previous COA analyses (COA 2005, COA 2000) have demonstrated the utility of multiple linear regression analyses to incorporate data collect ed by multiple agencies or using different laboratory methods to document long-term chang es in water quality. Previous analysis has shown that water quality is degrading over time at the main Barton Springs. Temporal trends at main Barton Springs and five related springs (Upper Barton, Old Mill/Sunken Garden, Eliza, Backdoor, and Cold Springs) are assessed by sim ilar methods in this report. Additionally, temporal trends in chlorophyll-a measured in Barton Springs Pool at the downstream dam are assessed. Methods Sample collection entities include the City of Austin Water Resource Evaluation Section (WRE), the Austin/Travis County Health and Human Servic es Department and the US Geological Survey (USGS). All data included in the analyses are stored in the City of Austin Field Sampling Database (FSDB) and are available upon request Data collected specifically by WRE is available at www.ci.austin.tx.us/wrequery/ and data collected by the USGS is available at waterdata.usgs.gov/tx/nwis These agencies represent the most comprehensive resources in terms of both number of samples and period of record for these springs. Data from main Barton Springs and five rela ted karst springs (Table 1) are assessed. Chlorophyll-a data from Barton Springs Pool measured at the downstream dam are assessed.

PAGE 2

SR 10-06 Page 2 of 43 February 2010 High Barton Springs was initially considered but did not have sufficient period of record to be included. The general period of record for data assessed in this report is the mid-1990s to the summer of 2009, although the period of record for main Barton Springs is much longer with earliest water quality measurements in the 1960s. Approximately 565 parameters from the five stud y sites were excluded from the analysis because data did not satisfy sample size or period of reco rd requirements or did not have a sufficient number of measurements above detection limits. There were 52 parameters included in the subsequent analysis. Parameters were analyzed for temporal trends only if: Data are available after 2005 (recent data) Data are available before 2007 (at least 2 years of data) There are at least 5 measurements (sufficient number of measurements) There are at least 2 detected measurements (some measurements above detection limits). Table 1. Study spring locations with latitude/long itude and site abbreviations used in the report. COA # Spring Name Abbreviation Latitude Longitude 160 Backdoor Spring BKDR 30.25951-97.82370 183 Upper Barton Spring UBS 30.26359-97.77378 422 Old Mill (Sunken Gardens) Spring OM 30.26359-97.76808 428 Eliza Spring ES 30.26425-97.77006 9 Cold Springs CS 30.27963-97.78050 35 Main Barton Springs MBS 30.26356-97.77128 Parameters normally measured in the field (c onductivity, dissolved oxygen, pH, turbidity and water temperature) that were measured in the la boratory were excluded from the analysis (except for USGS conductivity measurements analyzed in th e lab). These data were excluded in previous analyses at Main Barton Springs as well (COA 2005). All outliers as determined from visual inspection of the graphs by site and parameter we re examined individually. Invalid data were removed prior to analysis when a valid rationale for error, such as laboratory quality control or sampling equipment failure, was documented by project personnel in comments associated with database records. Samples affected by Barton Springs Pool draw down for maintenance were excluded from the analysis, as pool drawdown results in temporarily increased conductivity and turbidity and decreased dissolved oxygen (COA 2000) by draw ing in water from the saline water zone. Drawdown sample dates excluded from the analysis are: August 13, 1998; August 27-28, 1998; September 17-18, 1998. Differences in water quality at main Barton Spri ngs have been documented based on categorizing the data with respect to surface water recharge input to the aquifer from flowing creeks. During recharge conditions, Barton Springs water quality is reflective of the current water quality of creeks within the recharge zone (COA 1997; COA 2000). During non-recharge conditions, Barton Springs discharge is primarily a reflecti on of long-term water quality of the aquifer (COA 2000). Recharge condition was dete rmined using mean daily flow at the Barton Creek at Loop 360 USGS gage (USGS 08155300, available at waterdata.usgs.gov/tx/nwis ). Dates with non-zero mean daily flow at the gage were classified as recharge while dates with zero mean daily flow at the gage were classified as non-recharge conditions.

PAGE 3

SR 10-06 Page 3 of 43 February 2010 Previous dye tracing studies have shown that the areas contributing to Cold Springs include the Eanes Creek watershed, the channel and upla nds of Barton Creek between Loop 360 and Lost Creek Boulevard as well as Williamson Creek betw een US 290 and Brush County Boulevard. This data indicates that creek recharge conditions for Cold Springs may be more appropriately determined by use of the USGS gauge on Barton Creek at Lost Creek Boulevard (USGS 08155240) upstream of the Cold Spring contributing area, rather than Loop 360 on the downstream end of the contributing area to Cold Springs. There is no visual difference in the relationship of conductivity to flow at the two gauges (Figure 1), and there is no statistically significant correlation (Kendalls tau) between Cold Springs conduc tivity and flow at either gauge. The Lost Creek gauge did not yield any days with no flow when Cold Springs was sampled, and thus to be used in this analys is some minimum flow rate would need to be determined to accurately separate high and l ow recharge conditions for Cold Springs. For consistency, recharge condition for Cold Springs was determined by Loop 360 flows as done for the other sites. If there is no relation between any parameter and recharge condition at Cold Springs, the method of backward elimination w ill simply remove recharge condition from the model. Backward elimination regression models start with all candidate variables in the regression equation, and then eliminate non-significant variables programmatically until only the most parsimonious set of significant variables remain. 450 500 550 600 650 700 750 0 50100150200250300350 Flow (ft3/s)Conductivity (uS/cm) Loop 360 Flow Lost Creek Flow Figure 1. Conductivity at Cold Springs versus daily average flow from USGS gauges on Barton Creek at Loop 360 and Lost Creek Boulevard. Storm-influenced samples were excluded from the analysis. Storm designation was determined by examination of bacteria or total suspended solids (TSS) results from Main Barton Springs for each sample date. Antecedent rainfall at the National Weather Service Camp Mabry gauge was used in combination with the mean daily flow at Barton Springs as recorded by the USGS and an examination of water quality data and field st aff notes to determine if samples were storminfluenced.

PAGE 4

SR 10-06 Page 4 of 43 February 2010 Some water quality parameters in Barton Springs are strongly correlated with spring discharge (Senger and Kreitler 1984, COA 1997, COA 2000). Barton Springs mean daily discharge was entered into the multiple linear regression equati ons before the sample collection date to account for variation in concentration with discharge in temporal trend analyses, as in previous City of Austin analyses (COA 2000, COA 2005). Non-lin ear relationships with discharge have been observed for multiple dissolved solids paramete rs (chloride, sulfate, sodium, potassium, strontium, TDS and conductivity) at low dischar ge values (COA 2005, 2006). An overall average critical low flow value of 38 ft3/s was used for these 7 parameters (as the low-flow variable) in the temporal regression equations, following the a pproach previously used based on an identified breakpoint in sample result versus spring di scharge regressions (COA 2005). This improves prediction accuracy by removing the non-linear ity of the discharge coefficient for these parameters, enables prediction of change in valu es at both very low discharge and at long-term average discharge and follows the methods used to previously assess Main Barton Springs temporal trends. Multiple linear regression was used to determine if parameter results were changing over time, the direction of the change and the statisti cal significance of the change, following the methodology of previous studies (COA 2000, COA 2005). Analysis groups (non-recharge, recharge) were analyzed together and accounted for in the model with a recharge group variable. Mean daily flow at Barton Springs as determined by the USGS gauge (08155500) was associated with every sample collection date. The low-flow variable was included for the parameters with non-linear relationships to flow (COA 2005). Ca tegorical method/collector variables were entered into the regression equation, followed by Barton Springs flow and th en date. A backward elimination model was used in PROC REG (SAS 2004) to eliminate non-significant regression coefficients from the model. The full model is thus: Concentration = (Collector or Method or Filter)+(Collector or Method or Fi lter date) + (recharge condition) + (low-flow)+(discharge)+(date) Multiple linear regression results for parameters with censored observations (non-detect) were confirmed using the semi-parametric Cox propor tional hazards regression (Allison 1995). Scatter plots of data were examined for all parameters to assess the validity of the analyticallydetermined trends. Values were normalized to 50 ft3/s prior to plotting to remove flow effects for models with a statistically significant flow relation. Observed trends are compared to previous results from Main Barton Springs (COA 2005). Parameters or conditions with no statistically significant ( 0.05) trends over time are not presented. Date values in SAS are represented as a Julian date, calculated as the number of days from the date 1 January 1960. Results Although there were sufficient data for analysis, no significant temporal trends were evident for 32 parameters at any assessed site(Table 2). Similarly, no trends were observed for these same parameters in the 2005 analysis at main Bart on Springs (COA 2005). There was also no temporal trend in color, measured by the USGS in Pt-Co units, at main Barton Springs. Color was not previously assessed in main Barton Springs in 2 005, and insufficient data for color was available at the other sampling locations. Previously reported increasing trends in iron, l ead and zinc (COA 2005) at main Barton Springs have been confirmed as lab error relating to the acid preservation of sample bottles. After removing compromised sample results, there are no statistically significant trends in iron, lead or

PAGE 5

SR 10-06 Page 5 of 43 February 2010 zinc at any site, although high iron values have been detected in recent samples at Old Mill Springs. A method change in 2006 for nickel from EPA 200.7 to EPA 200.8 visually yields a false temporal trend, although PHREG analys is accounting for both change in method and measurements less than detection limits yields no statistically significant nickel trends over time at any site. Table 2. Parameters without significan t temporal trends at any site assessed. Parameter Aluminum Ammonia Arsenic Atrazine Bis(2-ethylhexyl) phthalate Boron Bromide Cadmium Caffeine Carbayl (Sevin) Chloroform Chlorophyll-a* Chromium Copper Diazinon E. coli bacteria Iron Lead Manganese Nickel Oil and Grease Organic Carbon Orthophosphorus Phosphorus Potassium Silver Simazine Tetrachloroethylene Total Suspended Solids Turbidity Volatile Suspended Solids Zinc *assessed at downstream dam in Barton Springs Pool only, insufficient data for trend assessment at other sites. Temporal trend results are presented in tabular fo rmat showing the estimated coefficient value for the variable in the regression equation (Estimate), the standard error associated with that estimate (StdErr) and the significance of the variable est imate (Pr>F). Only significant (Pr>F values less than 0.05) variables remaining after backward elimination are shown in the tables. Positive values for the date estimate coefficient indicat e significant, increasing temporal trends while negative values for the date estimate indicate significant, decreasing temporal trends.

PAGE 6

SR 10-06 Page 6 of 43 February 2010 Alkalinity The USGS generally measures dissolved alkalinity as CaCO3 (mg/L) in the field. COA measures total alkalinity in the lab (Table 3). Categori cal variables were used to represent collecting entities, filter fractions and laboratories. Table 3. Alkalinity data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 420052007 Dissolved USGS ES 5220032006 Dissolved USGS MBS 14019902009 Dissolved USGS OM 4720032005 Dissolved USGS UBS 4620032005 Total ATCHD MBS 9519921993 Total USGS MBS 3619781992 Total WRE BKDR 4919952009 Total WRE CS 5519942009 Total WRE ES 8919952009 Total WRE MBS 14319912009 Total WRE OM 9819942009 Total WRE UBS 8119972008 Prior to backward elimination, the full model for alkalinity is: Alkalinity = filter + filter*date + collector + collector*date + lab + lab*date + recharge_group + bs_flow + date Alkalinity yields an increasing temporal trend only in main Barton Springs (Table 4), with no other temporal trends evident at other sites. Alkalinity was previously found to be increasing during recharge only conditions. This analysis accounts for recharge conditions as a variable in the model, and thus predicted temporal trends are assessed for both recharge and non-recharge conditions combined and not separately as previ ously done. Increased variability in alkalinity measurements was observed in 2004-2005 at Eliza, Old Mill and Upper Barton Springs. Alkalinity appeared to be increasing at Backdoor Springs thru 2000, but appears to have stabilized since that time. Table 4. Alkalinity as CaCO3 (mg/L) results. Site adj r2 Type Estimate StdErr Pr>F Intercept 189.189 35.717 0.0000 filter . filter*date 0.010 0.002 0.0000 collector . coll*date -0.001 0.001 0.0916 lab . lab*date . recharge . bs_flow . BKDR 0.31 date .

PAGE 7

SR 10-06 Page 7 of 43 February 2010 Table 4 (continued) Site adj r2 Type Estimate StdErr Pr>F Intercept 240.765 3.904 0.0000 filter . filter*date . collector . coll*date . lab . lab*date . recharge 20.402 4.737 0.0001 bs_flow . CS 0.25 date . Intercept 8609.615 2897.590 0.0037 filter -6983.457 2413.266 0.0047 filter*date -0.040 0.013 0.0040 collector . coll*date 0.048 0.016 0.0034 lab -837.791 206.803 0.0001 lab*date 0.052 0.013 0.0001 recharge 12.249 2.876 0.0000 bs_flow 0.307 0.051 0.0000 ES 0.51 date . Intercept 254.874 8.414 0.0000 filter . filter*date . collector -56.096 8.193 0.0000 coll*date 0.005 0.001 0.0000 lab 12.929 2.775 0.0000 lab*date . recharge 11.576 2.204 0.0000 bs_flow 0.164 0.040 0.0001 MBS 0.27 date -0.001 0.001 0.0254 Intercept 217.834 9.029 0.0000 filter . filter*date . collector . coll*date 0.001 0.000 0.0038 lab -443.673 172.630 0.0117 lab*date 0.028 0.011 0.0097 recharge 16.707 4.146 0.0001 bs_flow 0.303 0.074 0.0001 OM 0.26 date .

PAGE 8

SR 10-06 Page 8 of 43 February 2010 Table 4 (continued) Site adj r2 Type Estimate StdErr Pr>F Intercept 16733.624 5164.694 0.0018 filter 13779.622 4303.063 0.0020 filter*date -0.075 0.024 0.0026 collector . coll*date 0.094 0.029 0.0016 lab -1123.044 342.600 0.0016 lab*date 0.070 0.021 0.0013 recharge . bs_flow 0.288 0.139 0.0420 UBS 0.43 date . Barium, Dissolved The majority of available barium data is dissolved (Table 5), collected by the USGS and can be assessed only at main Barton Springs. Two suspect values at MBS from the early 1980s were excluded from the analysis as probable outliers. Table 5. Barium data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 320052007 Dissolved USGS ES 120052005 Dissolved USGS MBS 7419782009 Dissolved USGS OM 120052005 Dissolved USGS UBS 120022002 Dissolved WRE CS 119941994 Dissolved WRE OM 119941994 Total WRE BKDR 219951995 Total WRE CS 219951995 Total WRE MBS 519941995 Total WRE OM 219951995 Prior to backward elimination, the full model for Barium was: Ba, diss = recharge_group + bs_flow + date Dissolved barium is increasing over time in MBS (T able 6). Early total barium concentrations at Backdoor Spring in 1995 were lower than curre nt dissolved values (2005-2007), but there is insufficient data for complete assessment. Despite the increasing trend in barium, all measurements in main Barton Springs are less th an 70 g/L and the estimated lowest observed effect concentration for aquatic organisms is approximately 5,800 g/L (Texas Surface Water Quality Standards, 30 TAC Chapter 307). Table 6. Dissolved Barium (g/L) results. Site Adj r2 Type Estimate Std Error Pr>F intercept 48.981 3.831 0.0000 recharge group 3.952 2.077 0.0655 bs_flow -0.141 0.033 0.0002 MBS 0.69 date 0.001 0.000 0.0004

PAGE 9

SR 10-06 Page 9 of 43 February 2010 Calcium The USGS measures dissolved calcium, while COA has measured both total and dissolved (Table 7). Current COA samples typically measure onl y total calcium. A categorical variable was added to account for differences in filter fraction. Prior to backward elimination, the full model for calcium was: Calcium = filter + filter*date + flow_group + recharge_group + bs_flow + date Table 7. Calcium data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 320052007 Dissolved USGS ES 5120032006 Dissolved USGS MBS 13819782009 Dissolved USGS OM 4820032005 Dissolved USGS UBS 4820022005 Dissolved WRE BKDR 120002000 Dissolved WRE CS 219942000 Dissolved WRE ES 120002000 Dissolved WRE MBS 120002000 Dissolved WRE OM 319942000 Total WRE BKDR 5219952009 Total WRE CS 5519952009 Total WRE ES 6319952009 Total WRE MBS 9319912009 Total WRE OM 7019952009 Total WRE UBS 4419972008 Combining dissolved and total calcium in the same regression model yields only an increasing temporal trend at main Barton Springs. Analyzing dissolved calcium separately yields increasing trends only at main Barton Springs. Analyzing to tal calcium separately yields increasing trends at all sites (Table 8).

PAGE 10

SR 10-06 Page 10 of 43 February 2010 Table 8. Calcium (mg/L) results. Site adj r2 Type Estimate Std Err Pr>F intercept 138.427 52.437 0.0112 filter -103.679 42.922 0.0196 filter*date 0.005 0.001 0.0000 flow group . recharge . bs_flow 0.099 0.041 0.0194 BKDR 0.53 date . intercept 131.632 30.840 0.0001 filter -89.829 26.779 0.0016 filter*date 0.002 0.000 0.0000 flow group -5.567 3.010 0.0707 recharge 7.411 2.888 0.0135 bs_flow 0.174 0.064 0.0089 CS 0.46 date . intercept 85.084 14.618 0.0000 filter -28.608 12.194 0.0213 filter*date 0.002 0.000 0.0000 flow group . recharge . bs_flow 0.119 0.019 0.0000 ES 0.38 date . intercept 242.692 72.600 0.0011 filter -156.896 65.218 0.0175 filter*date 0.010 0.004 0.0210 flow group . recharge . bs_flow 0.099 0.013 0.0000 MBS 0.44 date -0.010 0.005 0.0348 intercept 121.438 17.706 0.0000 filter -59.686 15.142 0.0002 filter*date 0.002 0.000 0.0000 flow group -11.901 2.607 0.0000 recharge . bs_flow 0.071 0.036 0.0556 OM 0.42 date . intercept 88.455 17.842 0.0000 filter -39.408 14.766 0.0103 filter*date 0.003 0.001 0.0001 flow group . recharge -4.599 1.422 0.0022 bs_flow . UBS 0.37 date . Chloride The USGS currently measures dissolved chloride, wh ile COA measures total chloride (Table 9). A categorical variable (0=total, 1=dissolved) w as used to represent the filter types. Although included in all models in the initial list of candi date variables, the non-linear relationship between

PAGE 11

SR 10-06 Page 11 of 43 February 2010 chloride and Main Barton Springs flow at low flows near 40 ft3/s was only evident in scatter plots of chloride versus flow at Old Mill and Eliza but not Cold, Upper Barton or Backdoor springs.. Table 9. Chloride data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 320052007 Dissolved USGS ES 2020052006 Dissolved USGS MBS 3520052009 Dissolved USGS OM 1620052005 Dissolved USGS UBS 1620052005 Total ATCHD MBS 4319911991 Total USGS ES 3120032004 Total USGS MBS 10519782004 Total USGS OM 3220032004 Total USGS UBS 3020032004 Total WRE BKDR 5119952009 Total WRE CS 5719942009 Total WRE ES 6419952009 Total WRE MBS 9619912009 Total WRE OM 7119942009 Total WRE UBS 4519972008 Prior to backward elimination, the full model for chloride was: Chloride = filter + filter*date + collector + collector*date + flow_group + recharge_group + bs_flow + date Increasing trends in chloride over time were ob served at Backdoor Springs and Eliza Springs, with a decreasing trend observed in Upper Bart on Springs. Although increasing in 2005 results, chloride yields a weak decreasing trend in main Barton Springs (Table 10). When total and dissolved chloride are analyzed separately at main Barton Springs, a weak decreasing temporal trend is observed for total chloride (19782009) while an increasing trend is observed for dissolved chloride (2005-2009).

PAGE 12

SR 10-06 Page 12 of 43 February 2010 Table 10. Chloride (mg/L) results. Site adj r2 Type Estimate Std Err Pr>F intercept 10.440 6.036 0.0904 filter . filter*date . collector . coll*date . flow_group 7.219 2.108 0.0013 recharge . bs_flow -0.086 0.033 0.0110 BKDR 0.29 date 0.001 0.000 0.0032 intercept 6410.694 1975.431 0.0021 filter 5344.789 1646.357 0.0021 filter*date 0.030 0.009 0.0015 collector . coll*date -0.035 0.011 0.0020 flow_group 6.698 1.135 0.0000 recharge . bs_flow . CS 0.50 date . intercept 3211.324 1447.575 0.0294 filter 2914.291 1251.789 0.0225 filter*date -0.180 0.076 0.0206 collector -249.692 83.683 0.0038 coll*date 0.016 0.005 0.0034 flow_group -11.569 1.703 0.0000 recharge . bs_flow -0.084 0.025 0.0010 ES 0.82 date 0.201 0.088 0.0250 intercept 50.336 2.077 0.0000 filter . filter*date . collector . coll*date . flow_group -12.365 1.469 0.0000 recharge . bs_flow -0.094 0.021 0.0000 MBS 0.72 date -0.0004 0.000 0.0080

PAGE 13

SR 10-06 Page 13 of 43 February 2010 Table 10 (continued) Site adj r2 Type Estimate Std Err Pr>F intercept 109.087 5.364 0.0000 filter . filter*date . collector . coll*date . flow_group -46.781 8.433 0.0000 recharge . bs_flow -0.211 0.115 0.0719 OM 0.63 date . intercept 9.471 10.128 0.3544 filter . filter*date 0.002 0.001 0.0313 collector . coll*date . flow_group . recharge 2.313 1.259 0.0725 bs_flow 0.132 0.043 0.0034 UBS 0.13 date -0.003 0.001 0.0142 Conductivity Conductivity is measured by multiple different fi eld instruments (Table 11). A categorical variable was used to represent the different field instruments. Prior to backward elimination, the full model for instantaneous conductivity was: Inst. Conductivity = method + method*date + flow_group + recharge_group + bs_flow + date

PAGE 14

SR 10-06 Page 14 of 43 February 2010 Table 11. Instantaneous conductivity data with field instrument. Collector Site Method # First Last USGS ES UNKNOWN 16 2004 2004 USGS MBS UNKNOWN 23 1980 2005 USGS OM UNKNOWN 17 2004 2004 USGS UBS UNKNOWN 16 2004 2004 WRE BKDR HORIBA WATER QUALITY METER 20 1995 2001 WRE BKDR HYDROLAB 1 2002 2002 WRE BKDR Quanta Probe 40 2002 2009 WRE CS CORNING M90 1 1995 1995 WRE CS HORIBA WATER QUALITY METER 20 1995 2001 WRE CS HYDROLAB 11 1991 2002 WRE CS Quanta Probe 22 2002 2009 WRE ES HORIBA WATER QUALITY METER 21 1995 2001 WRE ES HYDROLAB 2 2002 2006 WRE ES Quanta Probe 64 2002 2009 WRE MBS COLE PARMER FIELD STICK 1 2001 2001 WRE MBS COLE PARMER PHCON10 1 2001 2001 WRE MBS HORIBA WATER QUALITY METER 164 1995 2001 WRE MBS HYDROLAB 13 2001 2004 WRE MBS Quanta Probe 147 2000 2009 WRE MBS UNKNOWN 1 1995 1995 WRE MBS YSI Probe 1 2000 2000 WRE OM HORIBA WATER QUALITY METER 29 1995 2001 WRE OM HYDROLAB 2 2002 2006 WRE OM Quanta Probe 31 2002 2009 WRE UBS HORIBA WATER QUALITY METER 23 1997 2001 WRE UBS HYDROLAB 3 2002 2002 WRE UBS Quanta Probe 14 2002 2008 Over the period of record, instantaneous c onductivity appears to be increasing at Backdoor Springs and Cold Springs, but decreasing at ma in Barton Springs, Old Mill Springs and Eliza Springs (Table 12). The decrease at main Bart on Springs is observed in both recharge and nonrecharge conditions, and was previously (COA 2005) found to be increasing only in recharge conditions. When only the most recent COA field instrument is assessed (the Quanta probe, in use from 2000 to present), no trend is evident at main Barton Springs.

PAGE 15

SR 10-06 Page 15 of 43 February 2010 Table 12. Instantaneous conductivity (S/cm) results. Site adj r2 Type Estimate Std Err Pr>F intercept -360.059 357.708 0.32 flow_group . method 463.091 138.309 0.00 method*date -0.029 0.009 0.00 recharge_group 35.587 14.452 0.02 bs_flow . BKDR 0.43 date 0.072 0.025 0.01 intercept 426.024 41.973 0.00 flow_group . method . method*date . recharge_group 34.511 12.994 0.01 bs_flow 0.761 0.189 0.00 CS 0.32 date 0.007 0.002 0.01 intercept 1100.540 93.581 0.00 flow_group -45.477 6.687 0.00 method -159.542 51.745 0.00 method*date 0.011 0.003 0.00 recharge_group 15.764 6.611 0.02 bs_flow . ES 0.73 date -0.030 0.006 0.00 intercept 752.365 26.225 0.00 flow_group -53.744 4.582 0.00 method 17.910 2.673 0.00 method*date . recharge_group 12.074 4.435 0.01 bs_flow . MBS 0.46 date -0.007 0.002 0.00 intercept 1633.875 295.134 0.00 flow_group -219.481 23.976 0.00 method -399.456 181.257 0.03 method*date 0.026 0.011 0.02 recharge_group . bs_flow . OM 0.73 date -0.048 0.019 0.02 intercept 511.647 31.986 0.00 flow_group . method . method*date 0.001 0.000 0.00 recharge_group . bs_flow 0.928 0.365 0.02 UBS 0.48 Date . The main Barton Springs daily average conductivity relationship with flow yielded an unexpected change during the recent drought from the previous pattern established since the installation of the continuous monitoring probe in 2003 (Figure 2) In July 2008, main Barton Springs discharge fell below 28 ft3/s. Conductivity measurements from the recent drought from July 2008 thru early

PAGE 16

SR 10-06 Page 16 of 43 February 2010 August 2009 (when flows were in the 17 ft3/s range) deviated from the previously established pattern by plotting generally slightly lower that during previous periods of flow less than 30 ft3/s. The recent drought measurements result in a sta tistically significant (Pr>|t| = 0.0001) reduction (0.5 Scm-1/ft3s) in the predicted change of daily average conductivity slope versus daily average Barton Springs discharge at flows less than 38 ft3/s despite a sharp increase in conductivity at flows less than 18 ft3/s. 600 620 640 660 680 700 720 740 102030405060708090100 Barton Springs Flow (cfs)Conductivity (uS/cm) 2003 2008 July 2008 August 2009 Figure 2. Daily average conductivity versus daily average discharge at main Barton Springs. USGS datasonde continuous (15-minute logging intervals) was used to generate daily average conductivity. There were 2,182 daily average conductivity measurements from July 2003 to August 2009 assessed for temporal trends at main Barton Springs only. There was no attempt made to distinguish between different field in struments or deployments. In contrast to instantaneous conductivity over the period of record, the daily average conductivity from continuous monitoring yields an increasing tr end from 2003 to 2009 at main Barton Springs (Table 13). Prior to backward elimination, the full model for continuous conductivity was: Cont. Conductivity = flow_group + recharge_group + bs_flow + date Table 13. Continuous monitoring daily average conductivity (S/cm) results. Site adj r2 Type Estimate Std Err Pr>F intercept 496.733 11.848 0.0000 flow_group recharge_group bs_flow -0.675 0.014 0.0000 MBS 0.64 date 0.012 0.001 0.0000 Dissolved Oxygen

PAGE 17

SR 10-06 Page 17 of 43 February 2010 As with conductivity, instantaneous dissolved oxygen (DO) is measured by multiple field instruments (Table 14). A categorical variable was used to represent the different field instruments. Table 14. DO data. Collector Site Method # First Last USGS BKDR UNKNOWN 2 2005 2006 USGS CS UNKNOWN 2 2005 2006 USGS ES UNKNOWN 33 2003 2006 USGS MBS UNKNOWN 139 1969 2009 USGS OM UNKNOWN 28 2003 2005 USGS UBS UNKNOWN 27 2003 2005 WRE BKDR HORIBA WATER QUALITY METER 2 1996 1996 WRE BKDR HYDROLAB 1 2002 2002 WRE BKDR Quanta Probe 40 2002 2009 WRE BS-D HACH LDO PROBE 1 2006 2006 WRE BS-D HYDROLAB 1 1996 1996 WRE BS-D Quanta Probe 64 2004 2009 WRE CS HORIBA WATER QUALITY METER 5 1995 1996 WRE CS HYDROLAB 10 1991 2002 WRE CS Quanta Probe 22 2002 2009 WRE ES CALCULATION 125 2002 2009 WRE ES HORIBA WATER QUALITY METER 2 1996 1996 WRE ES HYDROLAB 4 2002 2006 WRE ES LAMOTTE TITRATION KIT 1 2007 2007 WRE ES Quanta Probe 63 2002 2009 WRE MBS CALCULATION 85 2002 2009 WRE MBS HORIBA WATER QUALITY METER 21 1995 2000 WRE MBS HYDROLAB 14 2001 2006 WRE MBS LAMOTTE TITRATION KIT 1 2005 2005 WRE MBS Quanta Probe 145 2000 2009 WRE MBS UNKNOWN 1 2006 2006 WRE MBS YSI Probe 1 2000 2000 WRE OM CALCULATION 122 2002 2009 WRE OM HORIBA WATER QUALITY METER 3 1995 1996 WRE OM HYDROLAB 3 2002 2006 WRE OM LAMOTTE TITRATION KIT 6 2005 2007 WRE OM Quanta Probe 31 2002 2009 WRE UBS CALCULATION 125 2002 2008 WRE UBS HYDROLAB 3 2002 2002 WRE UBS LAMOTTE TITRATION KIT 2 2007 2007 WRE UBS Quanta Probe 13 2002 2008 Prior to backward elimination, the full model for instantaneous DO was: DO = method + method*date + recharge_group + bs_flow + date Instantaneously measured DO is decreasing over time at main Barton Springs, Backdoor Springs, and Cold Springs but increasing ove r time at Upper Barton Springs. There is no temporal trend evident at Old Mill or Eliza Springs (Table 15).

PAGE 18

SR 10-06 Page 18 of 43 February 2010 Table 15. Instantaneous DO (mg/L) results. Site adj r2 Type Estimate Std Err Pr>F intercept 11.943 2.282 0.0000 method method*date recharge_group bs_flow BKDR 0.06 date -0.0003 0.0001 0.0699 intercept 9.778 1.122 0.0000 method method*date recharge_group bs_flow CS 0.18 date -0.0002 0.0001 0.0088 intercept 3.739 0.177 0.0000 method method*date recharge_group 0.331 0.116 0.0049 bs_flow 0.029 0.002 0.0000 ES 0.65 date intercept 4.992 0.244 0.0000 method method*date recharge_group 0.222 0.087 0.0113 bs_flow 0.029 0.001 0.0000 MBS 0.71 date -0.0001 0.0000 0.0000 intercept 4.381 0.211 0.0000 method method*date recharge_group bs_flow 0.020 0.003 0.0000 OM 0.25 date intercept -1.419 2.080 0.4962 method method*date recharge_group 0.440 0.150 0.0039 bs_flow 0.039 0.004 0.0000 UBS 0.39 date 0.0003 0.0001 0.0088 USGS continuous (15-minute logging interval) da tasonde DO at main Barton Springs only was used to generate daily average DO measurements and also assessed for temporal trends. There were 1,907 measurements from July 2003 to August 2009. The model did not attempt to account for changes in instrumentation because there was insufficient metadata to differentiate between instrument deployments. Barton Springs continuous DO is also decreasing over time (Table 16). Prior to backward elimination, the full model for continuous DO was: Continuous DO = recharge_group + bs_flow + date

PAGE 19

SR 10-06 Page 19 of 43 February 2010 Table 16. Continuous daily average datasonde DO (mg/L) results. Site adj r2 Type Estimate Std Err Pr>F intercept 12.5308 0.2709 0.0000 recharge_group 0.2482 0.0257 0.0000 bs_flow 0.0235 0.0004 0.0000 MBS 0.83 date -0.0005 0.0000 0.0000 Fecal coliform bacteria Fecal coliform bacteria was assessed only at ma in Barton Springs, where sufficient number of samples exist (Table 17). Fecal coliform is no longer being measured by ATCHHSD in favor of E. coli monitoring. E. coli bacteria yield no statistically signifi cant trends over time at any site. Temporal trends in fecal coliform were assessed at the spring outfall and at the downstream dam of Barton Springs Pool. Table 17. Fecal coliform data. Collector Site Method # First Last ATCHD BS-D SM 9222 D 48720032009 ATCHD MBS SM 9222 D 123619952009 ATCHD MBS UNKNOWN 54919911995 Prior to backward elimination, the full model for fecal coliform was: Fecal = recharge_group + bs_flow + date Fecal coliform may be increasing over time in Ba rton Springs, although there is no statistically significant temporal trend in the swimming pool (T able 18). The rate of increase is small based on the period of record, an d the very low adjusted r2 value indicates the high degree of variability in the measurements. Barton Springs continues to yield high water quality with indicator bacteria concentrations well below the State of Texas standard for safe contact recreation (Figure 3). Table 18. Fecal coliform (col/dL) results. Site Adj r2 Type Estimate Std Err Pr>F intercept 37.8744.9580.0000 recharge_group . bs_flow -0.1860.0730.0113 BSD 0.01 date . intercept -19.1426.8300.0051 recharge_group -4.6581.8830.0135 bs_flow . MBS 0.03 date 0.0030.0000.0000

PAGE 20

SR 10-06 Page 20 of 43 February 2010 0 20 40 60 80 100 120 140 spring outlet downstream damE. coli geometric mean (mpn/dL) Figure 3. E. coli geometric means at Barton Sp rings and the downstream dam of Barton Springs pool versus the Texas 126 mpn/dL contact recreation standard in red. Fluoride Dissolved fluoride is measured by the USGS, wh ile COA measures total fluoride (Table 19). A categorical variable was used to account for the va rying filter fractions for main Barton Springs. However, because of the limited number of measurements at the other springs only total fluoride was assessed at these sites to parsimoniously redu ce the number of variables in the model. Table 19. Fluoride data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 320052007 Dissolved USGS ES 420062006 Dissolved USGS MBS 7919782009 Total WRE BKDR 5019952009 Total WRE CS 5519942009 Total WRE ES 6319952009 Total WRE MBS 9519912009 Total WRE OM 7019942009 Total WRE UBS 4419972008 Prior to backward elimination, the full model for fluoride was: Fluoride = filter + filter*date + recharge_group + bs_flow + date Fluoride may be decreasing from 1978 to 2009 based on regression models combining total and dissolved fluoride (Table 20). When analyzed separately, there is no significant trend in dissolved fluoride at main Barton Springs and total fluoride may be increasing from 1991 to 2009. Total fluoride yields increasing temporal trends at Eliza Springs, Old Mill Springs and Upper Barton Springs.

PAGE 21

SR 10-06 Page 21 of 43 February 2010 Table 20. Fluoride (mg/L) results. Site adj r2 Type Estimate Std Err Pr>F Intercept 3.5954 1.0480 0.0009 Filter -2.9686 0.9380 0.0020 filter*date 0.0002 0.0001 0.0030 recharge_group . bs_flow -0.0018 0.0002 0.0000 MBS (total and diss) 0.50 Date -0.0002 0.0001 0.0041 Intercept 0.3278 0.0201 0.0000 recharge_group . bs_flow -0.0014 0.0003 0.0000 MBS (dissolved only) 0.33 Date . Intercept 0.1668 0.0234 0.0000 recharge_group . bs_flow . BKDR 0.00 Date . Intercept 0.2865 0.0418 0.0000 recharge_group . bs_flow -0.0012 0.0006 0.0578 CS 0.06 Date . Intercept 0.0229 0.1175 0.8459 recharge_group . bs_flow -0.0020 0.0003 0.0000 ES 0.51 Date 0.00002 0.00001 0.0025 Intercept 0.0689 0.0920 0.4568 recharge_group . bs_flow -0.0021 0.0003 0.0000 MBS 0.58 Date 0.00002 0.00001 0.0006 Intercept 0.0505 0.1079 0.6419 recharge_group . bs_flow -0.0022 0.0003 0.0000 OM 0.54 Date 0.00002 0.00001 0.0009 Intercept -0.2193 0.1635 0.1919 recharge_group . bs_flow . UBS 0.17 Date 0.00003 0.00001 0.0191 Hardness as CaCO3 Total hardness is measured by both the USGS and COA (Table 21). A categorical variable was used to represent the different collecting entities.

PAGE 22

SR 10-06 Page 22 of 43 February 2010 Table 21. Total hardness data. Filter Collector Site # First Last Total USGS BKDR 4 20052007 Total USGS CS 3 20052007 Total USGS ES 51 20032006 Total USGS MBS 138 19782009 Total USGS OM 48 20032005 Total USGS UBS 48 20022005 Total WRE ES 11 19992009 Total WRE MBS 21 19992009 Total WRE OM 11 19992009 Prior to backward elimination, the full model for hardness was: Hardness = collector + collector*date + recharge_group + bs_flow + date Hardness data yields increasing temporal trends at Eliza, main Barton Springs, and Old Mill Springs over the period of record (Table 22). Fluoride was previously predicted to increase over time at main Barton Springs in the 2005 analysis. Table 22. Hardness (mg/L as CaCO3) results. Site Adj r2 Type Estimate Std Err Pr>F intercept 400.0000 0.0000 collector . collector*date . recharge_group 20.0000 0.0000 bs_flow 0.0000 0.0000 BKDR 1.00 date . intercept 308.0000 . collector . collector*date . recharge_group -5.5000 . bs_flow 0.2500 . CS 0.00 date . intercept 206.0897 56.0365 0.0007 collector . collector*date . recharge_group . bs_flow 0.1789 0.0759 0.0238 ES 0.12 date 0.0059 0.0033 0.0833 intercept 272.6361 7.0003 0.0000 collector . collector*date . recharge_group . bs_flow . MBS 0.24 date 0.0026 0.0005 0.0000

PAGE 23

SR 10-06 Page 23 of 43 February 2010 Table 22. (continued) Site Adj r2 Type Estimate Std Err Pr>F intercept 113.2628 201.8059 0.5784 collector 400.7762 210.6136 0.0658 collector*date -0.0249 0.0129 0.0624 recharge_group . bs_flow . OM 0.09 date 0.0269 0.0123 0.0367 intercept 285.0664 15.7045 0.0000 collector . collector*date . recharge_group . bs_flow 0.5187 0.1742 0.0065 UBS 0.24 date . Magnesium Dissolved and total magnesium have been measured by the USGS and COA (Table 23). Generally consistent analytical methods (EPA 200.7) have been used. A categorical variable was used to account for differences in total and dissolved fractions. Differentiation of filter types also essentially accounts for difference s in collecting entities. Table 23. Magnesium data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 320052007 Dissolved USGS ES 5120032006 Dissolved USGS MBS 13819782009 Dissolved USGS OM 4820032005 Dissolved USGS UBS 4820022005 Dissolved WRE BKDR 120002000 Dissolved WRE CS 219942000 Dissolved WRE ES 120002000 Dissolved WRE MBS 120002000 Dissolved WRE OM 319942000 Total USGS BKDR 220052006 Total USGS CS 220052006 Total WRE BKDR 5219952009 Total WRE CS 5519952009 Total WRE ES 6619952009 Total WRE MBS 10019912009 Total WRE OM 7319952009 Total WRE UBS 4419972008 Prior to backward elimination, the full model for magnesium was: Mg = filter + filter*date + recharge_group + bs_flow + date Magnesium yields increasing temporal trends at main Barton Springs, Eliza Springs and Cold Springs, but no statistically significant temporal trend at the other sites (Table 24). Magnesium previously yielded increasing trends over time at main Barton Springs (COA 2005).

PAGE 24

SR 10-06 Page 24 of 43 February 2010 Table 24. Magnesium (mg/L) results. Site Adj r2 Type Estimate Std Err Pr>F intercept 25.5028 0.7933 0.0000 filter -18.7297 2.3373 0.0000 filter*date 0.0012 0.0001 0.0000 recharge_group 1.8424 0.5063 0.0006 bs_flow . BKDR 0.63 date . intercept 14.2400 1.8984 0.0000 filter -1.0884 0.6061 0.0785 filter*date . recharge_group . bs_flow -0.0184 0.0057 0.0022 CS 0.51 date 0.0006 0.0001 0.0000 intercept 14.7719 1.9213 0.0000 filter . filter*date . recharge_group 1.2912 0.3801 0.0010 bs_flow -0.0350 0.0061 0.0000 ES 0.64 date 0.0005 0.0001 0.0000 intercept 22.4756 0.8318 0.0000 filter -5.7476 1.9507 0.0038 filter*date 0.0004 0.0001 0.0046 recharge_group 0.8288 0.3167 0.0098 bs_flow -0.0411 0.0051 0.0000 MBS 0.59 date 0.0001 0.0001 0.0303 intercept 22.1089 0.3978 0.0000 filter . filter*date . recharge_group 2.2559 0.5196 0.0000 bs_flow . OM 0.17 date . intercept 18.1544 1.9485 0.0000 filter -21.0074 6.2036 0.0014 filter*date 0.0014 0.0004 0.0011 recharge_group . bs_flow 0.0436 0.0212 0.0453 UBS 0.19 date . Nitrate+Nitrite as N Dissolved nitrate plus nitrite as nitrogen (NO3) is measured almost exclusively by the USGS, while total NO3 is the standard parameter for C OA and has also been measured by USGS (Table 25). A categorical variable was used to represent the different filter fractions.

PAGE 25

SR 10-06 Page 25 of 43 February 2010 Table 25. NO3 data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 320052007 Dissolved USGS ES 4320032006 Dissolved USGS MBS 11619902009 Dissolved USGS OM 4120032005 Dissolved USGS UBS 4020032005 Dissolved WRE MBS 220062006 Dissolved WRE UBS 120072007 Total USGS MBS 5719781992 Total WRE BKDR 6219952009 Total WRE CS 5719952009 Total WRE ES 6719952009 Total WRE MBS 37619862009 Total WRE OM 7019952009 Total WRE UBS 4519972008 Prior to backward elimination, the full model for NO3 was: NO3 = filter + filter*date + recharge_group + bs_flow + date NO3 yields increasing temporal trends at ma in Barton Springs and Backdoor Springs. NO3 values at Old Mill Springs yield decreasing temporal trends. No trends are observed at Eliza Springs, Cold Springs or Upper Barton Springs NO3 was previously documented to be increasing in Barton Springs over time in the 2005 analysis (COA 2005). Table 26. NO3 (mg/L) results. Site Adj r2 Type Estimate Std Err Pr>F intercept -0.5615 0.7932 0.4818 filter . filter*date . recharge_group . bs_flow . BKDR 0.15 date 0.0002 0.0000 0.0014 intercept 1.1051 0.1726 0.0000 filter . filter*date . recharge_group 0.5926 0.1158 0.0000 bs_flow -0.0033 0.0017 0.0518 CS 0.67 date . intercept 1.0769 0.0229 0.0000 filter . filter*date . recharge_group 0.2859 0.0293 0.0000 bs_flow . ES 0.51 date .

PAGE 26

SR 10-06 Page 26 of 43 February 2010 Table 26. (continued) Site Adj r2 Type Estimate Std Err Pr>F intercept 1.0892 0.0669 0.0000 filter 0.0624 0.0217 0.0042 filter*date . recharge_group 0.1930 0.0214 0.0000 bs_flow -0.0015 0.0003 0.0000 MBS 0.38 date 0.00001 0.00000 0.0008 intercept 2.0164 0.3631 0.0000 filter . filter*date . recharge_group 0.3540 0.0760 0.0000 bs_flow 0.0052 0.0012 0.0001 OM 0.33 date -0.0001 0.0000 0.0001 intercept 2.1148 0.0480 0.0000 filter . filter*date . recharge_group . bs_flow . UBS 0.00 date . Sample frequency for conventional analytes of 26 scheduled events per year is established by the City of Austin Texas Pollutant Discharge Elim ination System (TPDES) MS4 permit. If reduced budget allowances for sampling mandate reduction or restructuring of sampling schedules, or if there is a desire to refocus sampling resources to new monitoring objectives like recent source water identification efforts, a sample frequency an alysis may inform those decisions. Nitrate is a critical parameter for evaluating eutrophicati on of contributing zone creeks, and may be important in assessing aesthetic impacts to the pool from nuisance algal growth. Temporal trends in nitrate were not evident until the 2005 anal ysis, suggesting that using nitrate may be a conservative estimate of overall trend predic tion sensitivity to sample frequency. An a posteriori sample frequency analysis was conducted on n itrate, systematically sub-sampling the WRE dataset and repeating the trend analysis. WRE sampling frequency may be reduced by up to 75% (only 7 events per year) with no loss of significan t prediction of temporal trends over the period of record, although the relationship to Barton Springs flow becomes non-significant (Table 27). Sampling events may be reduced by 50% with no change in prediction estimates for either temporal trends or relationship to Barton Springs flow.

PAGE 27

SR 10-06 Page 27 of 43 February 2010 Table 27. Nitrate WRE sample frequency reduc tion affect on regression model predictions for main Barton Springs (least significant regression models shown for each alternate sample frequency). Sample Freq adj r2 Type Estimate Std Err Pr>F intercept 1.08921 0.06694 0.0000 filter 0.06238 0.02166 0.0042 filter*date recharge 0.19301 0.02138 0.0000 bs_flow -0.00148 0.00033 0.0000 all samples 0.38 date 0.00001 0.00000 0.0008 intercept 1.04623 0.07221 0.0000 filter 0.05043 0.02275 0.0275 filter*date recharge 0.23376 0.02549 0.0000 bs_flow -0.00091 0.00040 0.0234 50% reduction in frequency 0.42 date 0.00001 0.00000 0.0062 intercept 0.95607 0.07957 0.0000 filter 0.04819 0.02810 0.0880 filter*date recharge 0.27046 0.02671 0.0000 bs_flow 75% reduction in frequency 0.38 date 0.00001 0.00000 0.0205 Non-Carbonate Hardness Filtered non-carbonate hardness is sampled almost exclusively, and only by the USGS (Table 28). Only dissolved non-carbonate hardn ess was included in the analysis. Table 28. Non-carbonate hardness (mg/L) data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 420052007 Dissolved USGS ES 4820032006 Dissolved USGS MBS 10319902009 Dissolved USGS OM 4220032005 Dissolved USGS UBS 4420032005 Total USGS MBS 1019781983 Prior to backward elimination, the full model for non-carbonate hardness was: NCH = recharge_group + bs_flow + date Non-carbonate hardness is increasing over time at main Barton Springs, and may be increasing over time in Eliza Springs (Pr>F = 0.09). Non-carbonate hardness was previously predicted to be increasing over time only in recharge conditions at main Barton Springs (COA 2005), and was not previously assessed at Eliza Springs.

PAGE 28

SR 10-06 Page 28 of 43 February 2010 Table 29. Non-carbonate hardness (mg/L) results. Site Adj r2 Type Estimate Std Err Pr>F intercept 3.200 0.000 recharge_group 34.050 0.000 bs_flow 0.225 0.000 BKDR 1.00 date . intercept 84.600 0.000 recharge_group -36.100 0.000 bs_flow 0.050 0.000 CS 1.00 date . intercept -132.304 114.918 0.2601 recharge_group -18.609 4.108 0.0001 bs_flow . ES 0.43 date 0.012 0.007 0.0981 intercept 29.291 14.067 0.0417 recharge_group -8.053 4.184 0.0591 bs_flow -0.120 0.067 0.0794 MBS 0.11 date 0.002 0.001 0.0089 intercept 75.571 3.590 0.0000 recharge_group -21.571 5.562 0.0008 bs_flow . OM 0.38 date . intercept 62.143 3.084 0.0000 recharge_group -16.543 4.778 0.0022 bs_flow . UBS 0.32 date . pH Water pH is measured by multiple different fiel d instruments (Table 30), by both the USGS and WRE. A categorical variable was used to repr esent the different field instruments, which also differentiated between the USGS and WRE collecting entities. Table 30. pH data. Collector Site Method # First Last ATCHD MBS Quanta Probe 1 2003 2003 USGS BKDR UNKNOWN 3 2005 2007 USGS CS UNKNOWN 3 2005 2007 USGS ES UNKNOWN 54 2003 2006 USGS MBS UNKNOWN 179 1969 2009 USGS OM UNKNOWN 49 2003 2005 USGS UBS UNKNOWN 48 2003 2005 WRE BKDR HACH 3 1990 1995 WRE BKDR HORIBA WATER QUALITY METER 20 1995 2001 WRE BKDR HYDROLAB 2 1994 2002 WRE BKDR Quanta Probe 40 2002 2009 WRE CS HACH 2 1991 1995 WRE CS HORIBA WATER QUALITY METER 20 1995 2001 WRE CS HYDROLAB 14 1991 2002

PAGE 29

SR 10-06 Page 29 of 43 February 2010 Table 30. (continued) Collector Site Method # First Last WRE CS Quanta Probe 22 2002 2009 WRE ES HORIBA WATER QUALITY METER 21 1995 2001 WRE ES HYDROLAB 2 2002 2006 WRE ES Quanta Probe 64 2002 2009 WRE ES SATUROMETER 2 2005 2008 WRE MBS COLE PARMER FIELD STICK 1 2001 2001 WRE MBS COLE PARMER PHCON10 1 2001 2001 WRE MBS HACH 6 1994 1996 WRE MBS HORIBA WATER QUALITY METER 165 1995 2001 WRE MBS HYDROLAB 12 2001 2004 WRE MBS Quanta Probe 147 2000 2009 WRE MBS SATUROMETER 1 2008 2008 WRE MBS YSI Probe 1 2000 2000 WRE OM HACH 1 1994 1994 WRE OM HORIBA WATER QUALITY METER 29 1995 2001 WRE OM HYDROLAB 2 2002 2006 WRE OM Quanta Probe 31 2002 2009 WRE OM SATUROMETER 3 2005 2008 WRE UBS HORIBA WATER QUALITY METER 23 1997 2001 WRE UBS HYDROLAB 3 2002 2002 WRE UBS Quanta Probe 14 2002 2008 WRE UBS SATUROMETER 1 2008 2008 Prior to backward elimination, the full model for pH was: pH = method + method*date + recharge_group + bs_flow + date pH yields decreasing (acidifying) trends over time for all sites except main Barton Springs. Main Barton Springs yields no statistically significan t trends over time, but was predicted to be decreasing only in recharge conditions in previ ous analyses. WRE pH data collected at main Barton Springs using only the Quanta Probe yields no statistically significant trends since 2005. Unadjusted pH measurements in 2008 and 2009 appear to be slightly higher than preceding years but within normal historic ranges. Table 31. pH (standard units) results. Site Adj r2 Type Estimate Std Err Pr>F intercept 10.8132 0.9187 0.0000 method -2.3088 0.5245 0.0001 method*date 0.0001 0.0000 0.0001 recharge_group bs_flow BKDR 0.23 date -0.0002 0.0001 0.0004 intercept 8.8342 0.8931 0.0000 method -1.0203 0.4525 0.0290 method*date 0.0001 0.0000 0.0390 recharge_group bs_flow CS 0.08 date -0.0001 0.0001 0.0843

PAGE 30

SR 10-06 Page 30 of 43 February 2010 Table 31. (continued) Site Adj r2 Type Estimate Std Err Pr>F intercept 9.1692 0.4095 0.0000 method -0.9438 0.2517 0.0003 method*date 0.0001 0.0000 0.0003 recharge_group bs_flow -0.0016 0.0004 0.0005 ES 0.26 date -0.0001 0.0000 0.0000 intercept 7.1008 0.0152 0.0000 method -0.3033 0.0787 0.0001 method*date 0.00002 0.00000 0.0006 recharge_group bs_flow MBS 0.07 date intercept 9.1198 0.6494 0.0000 method -1.2445 0.4349 0.0056 method*date 0.0001 0.0000 0.0057 recharge_group -0.1035 0.0542 0.0602 bs_flow -0.0034 0.0010 0.0014 OM 0.32 date -0.0001 0.0000 0.0130 intercept 10.6938 0.5297 0.0000 method -1.7049 0.3641 0.0000 method*date 0.0001 0.0000 0.0000 recharge_group -0.0859 0.0385 0.0307 bs_flow -0.0041 0.0012 0.0015 UBS 0.55 date -0.0002 0.0000 0.0000 Silica Dissolved silica is measured exclusively by th e USGS (Table 32). Silica was previously estimated to be increasing over time at main Barton Springs. Table 32. Silica data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 320052007 Dissolved USGS ES 5120032006 Dissolved USGS MBS 13819782009 Dissolved USGS OM 4820032005 Dissolved USGS UBS 4820022005 Prior to backward elimination, the full model for dissolved silica was: Si = recharge_group + bs_flow + date Dissolved silica is increasing over time in main Barton Springs (Table 33), in both recharge and non-recharge conditions when assessed separately. Dissolved silica is also increasing over time at Old Mill and Upper Barton Springs, but there is no significant temporal trend observed at Backdoor, Cold Springs or Eliza Springs.

PAGE 31

SR 10-06 Page 31 of 43 February 2010 Table 33. Silica (mg/L) results. Site Adj r2 Type Estimate Std Err Pr>F intercept 11.8200 0.0000 recharge_group 3.5550 0.0000 bs_flow 0.0475 0.0000 BKDR 1.00 date . intercept 6.7700 . recharge_group 3.6550 . bs_flow 0.0125 . CS 1.00 date . intercept 11.5654 0.1867 0.0000 recharge_group . bs_flow 0.0054 0.0022 0.0223 ES 0.15 date . intercept 9.7365 0.3618 0.0000 recharge_group . bs_flow . MBS 0.21 date 0.0001 0.0000 0.0000 intercept -1.6076 5.0025 0.7508 recharge_group 0.2987 0.1371 0.0399 bs_flow . OM 0.22 date 0.0008 0.0003 0.0123 intercept -2.4568 6.8189 0.7218 recharge_group . bs_flow . UBS 0.13 date 0.0009 0.0004 0.0371 Sodium Dissolved sodium is typically collected by the USGS, while total sodium is collected by WRE (Table 34). A categorical variable was added to account for differences in filter fraction, which essentially captured differences in collection entiti es as well. A categorical variable was also added to account for the non-linear response of s odium to Barton Springs discharge at discharge values less than 38 ft3/s. Table 34. Sodium data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 320052007 Dissolved USGS ES 5120032006 Dissolved USGS MBS 13819782009 Dissolved USGS OM 4820032005 Dissolved USGS UBS 4820022005 Dissolved WRE BKDR 120002000 Dissolved WRE CS 219942000 Dissolved WRE ES 120002000 Dissolved WRE MBS 120002000 Dissolved WRE OM 319942000 Total WRE BKDR 5219952009

PAGE 32

SR 10-06 Page 32 of 43 February 2010 Table 34 (continued) Filter Collector Site # First Last Total WRE CS 5519952009 Total WRE ES 6319952009 Total WRE MBS 9419912009 Total WRE OM 7019952009 Total WRE UBS 4519972008 Prior to backward elimination, the full model for sodium was: Na = filter + filter*date + recharge_group + low_flow + bs_flow + date An increasing trend in dissolved sodium was pr eviously observed at main Barton Springs (COA 2005) only during non-recharge conditions. Based on this analysis, sodium is increasing over time only at Cold Springs (Table 35). No signi ficant trend over time was identified at any other site. At main Barton Springs, sodium concentra tions appear to be primarily related to Barton Springs discharge. Table 35. Sodium (mg/L) results. Site Adj r2 Type Estimate Std Err Pr>F intercept 17.3998 0.9840 0.0000 filter -25.1404 3.1508 0.0000 filter*date 0.0015 0.0002 0.0000 flow_group . recharge_group . bs_flow . BKDR 0.55 date . intercept -1.8688 3.0536 0.5434 filter . filter*date . flow_group 2.2844 0.7355 0.0032 recharge_group -2.8257 0.7763 0.0007 bs_flow . CS 0.52 date 0.0010 0.0002 0.0000 intercept 25.7023 3.1721 0.0000 filter . filter*date . flow_group -8.6267 3.6628 0.0208 recharge_group . bs_flow . ES 0.05 date . intercept 28.3821 0.6850 0.0000 filter . filter*date . flow_group -9.4967 1.1224 0.0000 recharge_group . bs_flow -0.0517 0.0160 0.0016 MBS 0.73 date .

PAGE 33

SR 10-06 Page 33 of 43 February 2010 Table 35 (continued) Site Adj r2 Type Estimate Std Err Pr>F intercept 67.6063 3.4344 0.0000 filter -28.9521 12.9899 0.0287 filter*date 0.0018 0.0008 0.0275 flow_group -29.4244 4.1389 0.0000 recharge_group . bs_flow -0.1265 0.0580 0.0321 OM 0.75 date . intercept 4.6837 1.4923 0.0028 filter -10.8853 4.5592 0.0208 filter*date 0.0008 0.0003 0.0105 flow_group . recharge_group . bs_flow 0.0591 0.0163 0.0007 UBS 0.27 date . Strontium The majority of strontium sampling has been done in the dissolved fraction by the USGS, although some samples for total strontium have b een collected by WRE (Table 36). A categorical variable was added to represent the differences in collection entity and filter fraction. An additional categorical variable was used to acc ount for differences in the non-linear relationship between strontium and Barton Springs discharge at low flows. Table 36. Strontium data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 320052007 Dissolved USGS ES 4820032005 Dissolved USGS MBS 10619902009 Dissolved USGS OM 4920032005 Dissolved USGS UBS 4820022005 Total WRE BKDR 719952009 Total WRE CS 1019952009 Total WRE ES 920082009 Total WRE MBS 1219952009 Total WRE OM 1019952009 Total WRE UBS 220072008 Prior to backward elimination, the full model for strontium was: Sr = filter + filter*date + recharge_group + low_flow + bs_flow + date Strontium is decreasing over time at Eliza Sp rings (Table 37), but does not yield significant temporal trends at any other site. Although ther e is no trend at main Barton Springs, strontium values in 2008 and 2009 increased sharply due to an inverse relationship with flow thru the ongoing drought but were generally with in historic ranges (Figure 4).

PAGE 34

SR 10-06 Page 34 of 43 February 2010 Table 37. Strontium (g/L) results. Site Adj r2 Type Estimate Std Err Pr>F intercept 36.8589 17.3091 0.0865 filter -960.2194 40.2992 0.0000 filter*date 0.0571 0.0025 0.0000 low_flow -69.9858 10.9160 0.0014 recharge_group 85.1074 8.0943 0.0001 bs_flow 1.9105 0.3070 0.0016 BKDR 1.00 date . intercept 257.8416 25.1291 0.0000 filter -476.5150 89.5497 0.0011 filter*date 0.0310 0.0056 0.0009 low_flow -89.5713 25.6800 0.0102 recharge_group . bs_flow 1.3224 0.6225 0.0713 CS 0.92 date . intercept 6536.6443 1984.4966 0.0027 filter 56665.1595 10803.8069 0.0000 filter*date 3.2587 0.6157 0.0000 low_flow -794.3480 174.1740 0.0001 recharge_group 154.8107 53.5068 0.0073 bs_flow -5.6936 1.7658 0.0032 ES 0.98 date -0.2726 0.1235 0.0357 intercept 3087.9043 61.8599 0.0000 filter . filter*date . low_flow -1311.0988 112.3221 0.0000 recharge_group . bs_flow -10.9881 1.5362 0.0000 MBS 0.94 date . intercept 2043.6302 138.2001 0.0000 filter . filter*date . low_flow -534.8344 153.9209 0.0015 recharge_group 174.8845 79.1284 0.0344 bs_flow -7.3578 2.1308 0.0016 OM 0.82 date . intercept 360.0000 17.3581 0.0000 filter . filter*date . low_flow . recharge_group 89.9231 25.4748 0.0016 bs_flow . UBS 0.30 date .

PAGE 35

SR 10-06 Page 35 of 43 February 2010 0 500 1000 1500 2000 2500 3000 3500 4000 19901991199319951997199920012003200520072009Strontium (ug/L) Dissolved Total Figure 4. Dissolved and total strontium (g/L) over time at main Barton Springs during nonstorm influenced conditions. Sulfate Total and dissolved sulfate have been measured by the USGS, although WRE measures total sulfate (Table 38). Categorical variables were added to represent differences in filter fraction, collecting entity (and thus analytical method) and non-linear response to low flow. Table 38. Sulfate data. Filter Collector Site # First Last Dissolved USGS BKDR 420052007 Dissolved USGS CS 320052007 Dissolved USGS ES 2020052006 Dissolved USGS MBS 3520052009 Dissolved USGS OM 1620052005 Dissolved USGS UBS 1620052005 Total USGS ES 3120032004 Total USGS MBS 10519782004 Total USGS OM 3220032004 Total USGS UBS 3020032004 Total WRE BKDR 5019952009 Total WRE CS 5619942009 Total WRE ES 6319952009 Total WRE MBS 9319912009 Total WRE OM 7019942009 Total WRE UBS 4519972008 Prior to backward elimination, the full model for sulfate was: SO4 = filter + filter*date + collector + collector *date + recharge_group + low_flow + bs_flow + date

PAGE 36

SR 10-06 Page 36 of 43 February 2010 Sulfate yields increasing temporal trends at Backdoor Springs, Cold Springs (Pr>F = 0.06) and Eliza Springs. No temporal trends were obser ved at main Barton Springs, Old Mill or Upper Barton Springs. Sulfate was previously predic ted (COA 2005) to be increasing at main Barton Springs only in recharge conditions, although no trend is now evident in either recharge or nonrecharge conditions when analyzed separately. Table 39. Sulfate (mg/L) results. Site Adj r2 Type Estimate Std Err Pr>F intercept -11.3726 8.5158 0.1883 filter . filter*date . collector . collector*date . low_flow 3.8010 1.8226 0.0426 recharge_group . bs_flow . BKDR 0.22 date 0.0018 0.0005 0.0014 intercept 1001.9665 247.7370 0.0002 filter 1003.7186 248.1876 0.0002 filter*date -0.0597 0.0147 0.0002 collector . collector*date . low_flow 7.0655 2.9393 0.0203 recharge_group . bs_flow 0.1074 0.0453 0.0219 CS 0.63 date 0.0610 0.0146 0.0001 intercept 24.7877 6.7663 0.0004 filter . filter*date . collector . collector*date 0.0002 0.0001 0.0008 low_flow -7.5193 1.2377 0.0000 recharge_group . bs_flow . ES 0.42 date 0.0007 0.0004 0.0723 intercept 37.5113 1.0200 0.0000 filter . filter*date . collector 3.0268 0.9350 0.0015 collector*date . low_flow -7.9334 1.0615 0.0000 recharge_group . bs_flow . MBS 0.32 date .

PAGE 37

SR 10-06 Page 37 of 43 February 2010 Table 39. (continued) Site Adj r2 Type Estimate Std Err Pr>F intercept 78.4022 4.8162 0.0000 filter -32.4049 16.1016 0.0476 filter*date 0.0022 0.0010 0.0298 collector . collector*date . low_flow -34.8385 3.2004 0.0000 recharge_group . bs_flow . OM 0.62 date . intercept 15.6486 4.1039 0.0004 filter . filter*date . collector . collector*date 0.0002 0.0001 0.0610 low_flow . recharge_group . bs_flow 0.1095 0.0446 0.0177 UBS 0.12 date . Total Kjeldahl Nitrogen as N Total TKN was collected by both the USGS and WRE (Table 40), although the USGS now measures dissolved (or filtered) TKN. Only total TKN was included in this analysis. TKN samples include a large proportion of values below detection limits, and trend analysis was confirmed with PHREG. Table 40. TKN data. Filter Collector Site # First Last Total USGS ES 8 20032005 Total USGS MBS 136 19782009 Total USGS OM 9 20032005 Total USGS UBS 8 20032005 Total WRE BKDR 31 19952004 Total WRE CS 35 19942004 Total WRE ES 37 19952008 Total WRE MBS 264 19862006 Total WRE OM 45 19942004 Total WRE UBS 34 19972004 Prior to backward elimination, the full model for TKN was: TKN = collector + collector*date + recharge_group + bs_flow + date TKN is decreasing over time at main Barton Springs (Table 41), with results from multiple linear regression confirmed by PHREG. There is no signi ficant temporal trend observed at any other site by either regression method. Based on the lack of trend in ammonia, it may be inferred that organic nitrogen is decreasing over time at ma in Barton Springs. TKN was observed to be decreasing over time at main Barton Spring in the previous analysis (COA 2005).

PAGE 38

SR 10-06 Page 38 of 43 February 2010 Table 41. TKN (mg/L) results. Multiple Linear Regression PHREG Site Adj r2 Type Estimate Std Err Pr>F Estimate Std Err Pr>x2 intercept 0.1966 0.0272 0.0000 collector . collector*date . recharge_group . bs_flow . BKDR 0.00 date . intercept 0.1512 0.0186 0.0000 collector . collector*date . recharge_group . bs_flow . CS 0.00 date . intercept 0.0693 0.0380 0.0759 collector . collector*date . recharge_group . bs_flow 0.0008 0.0005 0.0991 ES 0.04 date . intercept 0.5319 0.0633 0.0000 . collector -0.2371 0.1353 0.0809 0.6660 1.2578 0.5965 collector*date 0.0000 0.0000 0.0607 0.0000 0.0001 0.9719 recharge_group . -0.1953 0.2230 0.3813 bs_flow . 0.0018 0.0038 0.6259 MBS 0.09 date 0.0000 0.0000 0.0000 -0.0004 0.0001 0.0001 intercept 0.1515 0.0196 0.0000 collector . collector*date . recharge_group . bs_flow . OM 0.00 date . intercept 0.0825 0.0341 0.0232 collector 0.0868 0.0407 0.0429 collector*date . recharge_group . bs_flow . UBS 0.12 date . Water Temperature Instantaneous water temperature was measured by multiple different field instruments (Table 42). A categorical variable was used to account fo r the differences in field instrument.

PAGE 39

SR 10-06 Page 39 of 43 February 2010 Table 42. Temperature data. Collector Site Method # First Last USGS BKDR UNKNOWN 320052007 USGS CS UNKNOWN 320052007 USGS ES UNKNOWN 4420032006 USGS MBS UNKNOWN 16019692009 USGS OM UNKNOWN 3920032005 USGS UBS UNKNOWN 3820032005 WRE BKDR HORIBA WATER QUALITY METER 2019952001 WRE BKDR HYDROLAB 219942002 WRE BKDR Quanta Probe 4020022009 WRE BKDR THERMOMETER (ALCOHOL) 319901995 WRE CS HORIBA WATER QUALITY METER 2019952001 WRE CS HYDROLAB 1019912002 WRE CS Quanta Probe 2220022009 WRE CS THERMOMETER (ALCOHOL) 119951995 WRE ES HORIBA WATER QUALITY METER 2119952001 WRE ES HYDROLAB 420022006 WRE ES Quanta Probe 6420022009 WRE ES SATUROMETER 12820022009 WRE MBS COLE PARMER FIELD STICK 120012001 WRE MBS COLE PARMER PHCON10 120012001 WRE MBS HORIBA WATER QUALITY METER 16519952001 WRE MBS HYDROLAB 1420012006 WRE MBS Quanta Probe 14720002009 WRE MBS SATUROMETER 10120022009 WRE MBS THERMOMETER (ALCOHOL) 419941995 WRE MBS UNKNOWN 120062006 WRE MBS YSI Probe 120002000 WRE OM HORIBA WATER QUALITY METER 2919952001 WRE OM HYDROLAB 320022006 WRE OM Quanta Probe 3120022009 WRE OM SATUROMETER 12820022009 WRE OM THERMOMETER (ALCOHOL) 119941994 WRE UBS HORIBA WATER QUALITY METER 2319972001 WRE UBS HYDROLAB 320022002 WRE UBS Quanta Probe 1420022008 WRE UBS SATUROMETER 12820022008 Prior to backward elimination, the full model for instantaneous temperature was: Temp = method + method*date + recharge_group + bs_flow + date Instantaneous temperature is increasing over time at main Barton Springs, Backdoor Springs and Upper Barton Springs (Table 43). Trends in temper ature are most likely related to general trends in ambient air temperature in Austin. Although based on a limited number of measurements (n=68), instantaneous water temperature in th e pool at the downstream dam is increasing over time as well (p=0.0278).

PAGE 40

SR 10-06 Page 40 of 43 February 2010 Table 43. Instantaneous temperature (C) results. Site Adj r2 Type Estimate Std Err Pr>F intercept 17.8601 0.7657 0.0000 method 1.3823 0.4371 0.0026 method*date -0.0001 0.0000 0.0029 recharge_group . bs_flow . BKDR 0.16 date 0.0002 0.0000 0.0028 intercept 20.1420 0.2350 0.0000 method . method*date . recharge_group . bs_flow . CS 0.00 date . intercept 21.1513 0.1505 0.0000 method -0.8837 0.3800 0.0210 method*date 0.0001 0.0000 0.0256 recharge_group 0.2657 0.1225 0.0311 bs_flow . ES 0.05 date . intercept 20.4371 0.2513 0.0000 method . method*date 0.0000 0.0000 0.0456 recharge_group 0.4465 0.0717 0.0000 bs_flow -0.0039 0.0011 0.0004 MBS 0.23 date 0.0000 0.0000 0.0032 intercept 19.1184 0.6506 0.0000 method . method*date . recharge_group 0.7739 0.3999 0.0545 bs_flow 0.0178 0.0064 0.0060 OM 0.03 date . intercept 17.1438 1.6047 0.0000 method 2.2321 0.5690 0.0001 method*date -0.0001 0.0000 0.0002 recharge_group 0.2416 0.0864 0.0058 bs_flow 0.0055 0.0024 0.0229 UBS 0.17 date 0.0002 0.0001 0.0216 Daily average temperature trends were assessed at main Barton Springs using USGS datasonde data from 2003 to 2009, based on continuous monito ring (15-minute logging interval). There were 2,196 days with daily average temperature estimates in that time period. Daily average temperature is increasing over time at main Barton Springs from 2003 to 2009. Prior to backward elimination, the full model for daily average temperature was: Temp = recharge_group + bs_flow + date

PAGE 41

SR 10-06 Page 41 of 43 February 2010 Table 44. Daily average temperature (C) results. Site Adj r2 Type Estimate Std Err Pr>F intercept 19.58600.28100.0000 recharge_group 0.13830.02640.0000 bs_flow -0.00310.00040.0000 MBS 0.14 date 0.00010.00000.0000 Conclusions Barton Springs continues to maintain high water quality, although as exemplified by the decreases in DO and on-going increases in conductivity and nitrate, Barton Springs water quality is degrading over time. Trends in DO and nutrient s are of particular concern due to the potential for impact on both the endangered salamander and aesthetic impairments in the swimming pool. Trend analyses for parameters yielding significan t change over time are summarized (Table 45). The majority (62%) of parameters with sufficient data y ield no trend over time at any site (Table 2). There are few consistent temporal trend patterns between the springs assessed excepting the parameters with no temporal trends at any site. Calcium may be increasing at all sites, and pH may be decreasing at all sites except main Barton Springs. The extreme recent drought may have had unexpected effects on some parameters like conductivity. The relationship of these paramete rs to Barton Springs discharge needs to be examined in more detail, and may be inform ative in the on-going source water identification efforts. Temporal trends at main Barton Springs most fr equently m atch temporal trends at Backdoor and Upper Barton Springs. Among the minor springs, Backdoor Springs appears to most closely match temporal trends at Cold Springs, while Eliza and Old Mill springs are most closely matched. There may be improvement in some parameters at main Ba rton Springs (chloride, fluoride, pH, potassium, sodium and sulfate). WRE sampling frequency may be reduced by up to 75% (only 7 events per year) with no loss of significant prediction of temporal trends for nitrate over the period of record, although the relationship to Barton Springs flow becomes non-si gnificant (Table 27). Sampling events may be reduced by 50% with no change in prediction estimates for either nitrate temporal trends or relationship to Barton Springs flow.

PAGE 42

SR 10-06 Page 42 of 43 February 2010 Table 45. Summary of trend analyses for parame ters yielding significant trends over time. Param MBS 2005 Result MBS BKDR CS ES OM UBS same as 2005 analysis ALKALINITY (AS CACO3) increasing1 increasing no trend no trend no trend no trend no trend CALCIUM increasing increasing increasing5 increas ing5 increasing5 increasing5 increasing5 CONDUCTIVITY increasing1 increasing increasing increasing decreasing decreasing no trend DISSOLVED OXYGEN decreasing decreasing dec reasing decreasing no trend no trend increasing FECAL COLIFORM increasing1 increasing n/a n/a n/a n/a n/a HARDNESS (AS CACO3) increasing increasing no trend no tr end increasing increasing no trend MAGNESIUM increasing increasing increasing in creasing increasing no trend no trend NITRATE/NITRITE AS N increasing increasing increasing no trend no trend decreasing no trend NON-CARB. HARDNESS increasing1 increasing no trend no trend increasing no trend no trend SILICA increasing increasing no trend no tr end no trend increasing increasing STRONTIUM no trend no trend no trend no trend decreasing no trend no trend TKN AS N decreasing decreasing no trend no trend no trend no trend no trend WATER TEMPERATURE increasing increasing increasing no trend no trend no trend increasing changes from 2005 analysis BARIUM no trend increasing no trend no trend no trend no trend no trend CHLORIDE increasing decreasing increasing no trend increasing no trend decreasing FLUORIDE no trend decreasing no trend no trend in creasing increasing increasing PH decreasing1 no trend decreas ing decreasing decreasi ng decreasing decreasing POTASSIUM increasing3 no trend no trend no trend no trend no trend no trend SODIUM increasing3,4 no trend no trend increasing no trend no trend no trend SULFATE increasing1 no trend increasing incr easing increasing no trend no trend 1. in recharge conditions only 2. increasing tre nd previously 3. dissolved fraction only 4. in non-recharge conditions only 5. total fraction only Acknowledgements This report was prepared by Chris Herrington, PE, and Scott Hiers, PG, of the Water Resource Evaluation Section, Environmental Resource Management Division, Watershed Protection Department, City of Austin. References City of Austin (COA). 1997. The Barton Creek Report. City of Austin Environmental and Conservation Services Department. City of Austin (COA). 2000. Update of Barton Springs Water Quality Data Analysis-Austin, Texas. City of Austin Watershed Prot ection Department, Environmental Resources Management Division. SR-00-03. City of Austin (COA). 2005. Update of Bart on Springs Temporal Trend Analysis-2005. Water Resource Evaluation Section, Environmen tal Resource Management Division, Watershed Protection & Development Review Department, City of Austin. SR-05-09. August 2005.

PAGE 43

SR 10-06 Page 43 of 43 February 2010 City of Austin (COA). 2006. Effects of low spring discharge on water quality at Barton, Eliza and Old Mill springs, Austin, Texas. City of Austin Watershed Protection Department. SR-06-06.


printinsert_linkshareget_appmore_horiz

Download Options

close
Choose Size
Choose file type
Cite this item close

APA

Cras ut cursus ante, a fringilla nunc. Mauris lorem nunc, cursus sit amet enim ac, vehicula vestibulum mi. Mauris viverra nisl vel enim faucibus porta. Praesent sit amet ornare diam, non finibus nulla.

MLA

Cras efficitur magna et sapien varius, luctus ullamcorper dolor convallis. Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Fusce sit amet justo ut erat laoreet congue sed a ante.

CHICAGO

Phasellus ornare in augue eu imperdiet. Donec malesuada sapien ante, at vehicula orci tempor molestie. Proin vitae urna elit. Pellentesque vitae nisi et diam euismod malesuada aliquet non erat.

WIKIPEDIA

Nunc fringilla dolor ut dictum placerat. Proin ac neque rutrum, consectetur ligula id, laoreet ligula. Nulla lorem massa, consectetur vitae consequat in, lobortis at dolor. Nunc sed leo odio.