pg(s) 333-342 The city
of Porrentruy (JU, Switzerland) is vulnerable to flooding from
karst water draining the system of the Beuchire-Creugenat.
Major flood events in 1804 and 1901 led to heavy damages
throughout the city and its vicinity. Furthermore small-scale
flood events have been recorded five times in the last 30 years
- each resulting in substantial costs. The Beuchire-Creugenat
karst system is characterized by a perennial outlet (the
Beuchire spring) and several overflow outlets (among which the
Creugenat temporary outflow is the most significant one) where
the discharge rate often exceeds 15 m3/s. The ratio between
rainfall intensity and discharge rate of the overflow springs
is not closely correlated. Therefore, the discharge rates and
the conditions at which a certain overflow becomes active could
not be assessed without a comprehensive understanding of the
karst system behavior. Thus, the establishment of effective
flood risk management measures remains significant challenge.
In order to assess similar flood events and to determine the
most flooding vulnerable areas, the KARSYS approach has been
applied to the Beuchire-Creugenat karst system. A detailed
geological 3D model of the study area has been built in order
to reproduce the aquifer base geometry, the extension of its
expected saturated part(s) and the position of the main vadose
flowpaths "drainage axes". This approach enabled the catchment
area delineation by combination of subterraneous drainage axes.
The comparison of the discharge time series of the main springs
and the relevant rainfalls (~10-year series) provides
sufficient implications for understanding and consequent
reproducing of threshold functionality of the karst system
exposed to flooding due to rainfall events. A relationship
could be established between rainfall intensity/frequency
(return period) and the corresponding elevation of the
groundwater level within the karst conduits (or respectively,
the relevant spring discharge rates). The known overflow
springs have been added in the 3D model. The areas where (and
when) karst groundwater is expected to reach the ground surface
during extreme high-water events could be identified as
potential overflow springs. Such draining sensitive areas have
been delineated and mapped according to the calculated return
period of multiannual, 30- and 300- years flood events and the
relevant maximum discharge rates at the main outlets have been
13TH SINKHOLE CONFERENCE NCKRI SYMPOSIUM 2 events. A relationship could be established between corresponding elevation of the groundwater level within the karst conduits (or respectively, the relevant spring groundwater is expected to reach the ground surface areas have been delineated and mapped according to the rates at the main outlets have been assessed. Introduction 1 st 1804 in Porrentruy (JU) is a the largest known event Associated discharges reaching 100 m recorded for this karst system. Similar events were recorded in 1901 and 1910 (Figure 2). In addition to 9 th approached 20 m The local authority (administration of the Jura canton) has to plan protective measures to diminish the potential understanding and prediction of such extreme situation Abstract The city of Porrentruy (JU, Switzerland) is vulnerable 1901 led to heavy damages throughout the city and its in substantial costs. by a perennial outlet (the Beuchire spring) and several rate often exceeds 15 m not closely correlated. Therefore, the discharge rates active could not be assessed without a comprehensive understanding of the karst system behavior. Thus, area has been built in order to reproduce the aquifer base geometry, the extension of its expected saturated part(s) axes. This approach enabled the catchment area delineation by combination of subterraneous drainage axes. The comparison of the discharge time series of the consequent reproducing of threshold functionality of MAPPING FLOOD-RELATED HAZARDS IN KARST USING THE KARSYS APPROACH: APPLICATION TO THE BEUCHIRE-CREUGENAT KARST SYSTEM Jonathan Vouillamoz, Arnauld Malard, Eric Weber, Pierre-Yves Jeannin Swiss Institute for Speleology and Karst Studies, Rue de la Serre 68, CH-2301, La Chaux-de-Fonds, firstname.lastname@example.org Gabrielle Schwab Rouge BG Ingnieurs Conseils SA Avenue de Cour 61, CH-1007, Lausanne, Gabrielle.SCHWABROUGE@bg-21.com 333
NCKRI SYMPOSIUM 2 13TH SINKHOLE CONFERENCE m a.s.l. It becomes active only at high water stage. The global discharges in the city of Porrentruy may Further upstream, at an elevation of 465 m a.s.l., lies second temporary outlet of the system and becomes active only during very high water stage (see Figure 4). The discharge series of the Beuchire spring and have been measured at hourly time steps, respectively between 1998 to 2004 and 1998 to 2008. In addition to these three main springs, a series of minor temporary outlets do exist. Unfortunately they are badly documented due to infrequent activity (Les sources, Libecourt, etc.). is then required for assessing the probable occurrence the study region. The assessment was conducted applying the KARSYS approach (Jeannin et al. 2012) expanded with some hydraulic considerations. The aim was: karst water table for various recharge scenarios. Recharge events with return periods ranging To determine and map areas which are the most The locations where karst groundwater is expected to reach the most forward ground surface are the most catchment areas of the underground tributaries, discharge Context The geological context refers to the Tabular Jura which is slightly folded and intersected by numerous strike perennial karst spring emerges in the center of Porrentruy which is composed of alternating units of Upper Jurassic limestone and thin layers of marls. The Malm aquifer is underlied by a thick marl formation (Astartes marls, water exchanges with the lower aquifer are highly likely through discontinuities in the marls. Upstream of the Beuchire spring, the Creugenat 334 Figure 2. The flood event of 1910 in Porrentruy. Inundation is the result of extreme karst aquifer discharge through the Creugenat (Archives de la Bourgeoisie de Porrentruy, JU). Figure 1. The Beuchire karst spring is located in the city of Porrentruy (JU) in Northwest Switzerland.
13TH SINKHOLE CONFERENCE NCKRI SYMPOSIUM 2 was applied to estimate the geometry of the aquifer(s) boundaries, to delineate groundwater body(ies), and to assess the functioning of the Beuchire spring and This approach was assessed for low, medium, high and extremely high water conditions. Although the Beuchire spring and the Creugenat Schweizer 1970, Monbaron 1975, etc.) available data a clear solution. None of these studies describes in details the potential catchment boundaries and their possible changes in relation to the water stage. In this context the KARSYS approach (Jeannin et al. 2012) 335 Figure 4. SW-NE profile of the Beuchire catchment and projection of the overflow outlets. The real distance between the Beuchire spring and the Creugenat temporary overflow is 4,3 km. The Creugenat overflow and the Creux-des-Prs temporary overflow are 1.45 km appart. Figure 3. The Creugenat overflow in 1934. First pumping test to dry up the siphon (picture A. Perronne). A 3D model to assess the aquifer geometry geological model focusing on the aquifer basement (i.e. Astartes marls) was established for the area of interest 5) to meet the requirements of a pragmatic issue. This was possible thanks to an extensive compilation of all existing data relative to geological information (borehole which provided a strong basis of documentation. Once the geological model has been established and the data checked, the hydrological features have been temporary ones. Then, the extension of the saturated part of the aquifer was assessed by following the KARSYS approach. This approach assumes that at low water stage, the top of the saturated part of a karst aquifer is close to horizontal and can be represented within the model by a horizontal plane at the main perennial spring elevation (the Beuchire spring in the present case). The portion of the aquifer located underneath that horizontal plane should be close to the volume extension of the karst phreatic zone.
NCKRI SYMPOSIUM 2 13TH SINKHOLE CONFERENCE respective drainage axes were rendered. Drainage axes are recognized as vadose ones if they are located above the saturated zones. They are assumed to be developed at the bottom of the aquifer along the dip of the basement. Phreatic passages located within the saturated zones are drainage axes linking input points into the phreatic zone to the main drainage horizontal and a priori follow the shortest hydraulic distance to the outlet(s). The model result for low water stage is presented in Figure 7. The total groundwater catchment area in low water situations is thus estimated at about 79 km 2 Assessing the system at high water stage (hydraulic gradients within the conduits) In the next step, the hydrology of the system is assessed for high water conditions, i.e. active according to the rise of the groundwater head in the conduits. The discharge data from the Beuchire spring are compared with the head data recorded at the Creugenat This process was applied to the model and gives the following results (Figure 6). threshold which top elevation is at 440 m a.s.l. This fact also predetermined the volume extension and 2 (A) and 6 km 2 (B) and taking into account the assumed the water content could be estimated; respectively as 6.4 Mm and 2.9 Mm at the low water stage. Knowing the location and volume extension of the saturated parts of the aquifer, the catchment area feeding each of these groundwater bodies could be delineated. Then, following the shape of the 336 Figure 6. Model identification of two main groundwater bodies (GWB), A = Beuchire GWB (elevation 423 m a.s.l), B = Bonnefontaine GWB (elevation 438 m a.s.l). Figure 5. Illustration of the 3D geological model of the aquifer basement from the Beuchire karst system intersected by geological cross-sections. This was modeled using GeoModeller (Intrepid Geophysic). The main spring is on the front of the model. Figure 7. Model identification of the main underground vadose (yellow) and phreatic (blue) flowpaths of the Beuchire-Creugenat karst system at low water stage.
13TH SINKHOLE CONFERENCE NCKRI SYMPOSIUM 2 type), the relation can be simulated using the following equation: (Eq. 1) With Q [m 2 ) [m plotted on the chart (models 1 & 2, Figure 8). Model 1 suggests that the hydraulic connection between when the water elevation in the conduits ranges from and 451 m a.s.l. Therefore, model 2 depicts a hypothetical outlet at is valid for an outlet located at a distance of 4,000 m and diameter of 2.7 m) to reproduce the observed trend. Previous observations led to the hydraulic schemes of the karst system presented in Figure 9 that depends on the groundwater level elevation. This provides a set of displayed on Figure 10 where the gradients correspond to event) at high water stage. Upstream of the Beuchire spring the hydraulic gradient strongly increases until it is close to 0.7%). If the water level still increases the mapped in yellow. Similar scenarios could be established for two larger remains comparable to the value encountered above The graphic indicates that: As the Beuchire spring discharge remains lower indicates that both systems are disconnected by a threshold (a) situated at this elevation. level progressively rises up and shows a level at the Creugenat rises up a second time until it reaches the output elevation (c) at 451 m a.s.l threshold (b) as a first activation of the Creugenat overflow, this analysis indicates that the Creugenat becomes active only when the Beuchire spring indicates that an intermediate overflow (or large a.s.l). This could be a karst conduit or an outlet to the ground surface. 337 Figure 8. Comparison of the hourly pressure data recorded at the Creugenat overflow and the hourly discharge values of the Beuchire spring during flood (grey) and recession (red) events (2002-2004). Model 1 simulates a threshold discharge at an elevation of 437 m a.s.l at a suggested distance of 1,300 m downstream from the Creugenat. Model 2 simulates an ideal function of the spring discharge using a k*S of 28 m 3 /s (conduits diameter of ~2.7 m) and a straight distance of 4,000 m.
NCKRI SYMPOSIUM 2 13TH SINKHOLE CONFERENCE the gradient): its slope is approximately 1% extending the groundwater bodies as pictured in Figure 11. The areas the hydraulic gradient within the conduits does not change between the Beuchire spring and the Creugenat of the more elevated outlets (higher than 500 m a.s.l) and the shape of the versants. The gradient is therefore approximately 1.5%. Surfaces that are here vulnerable bottom of the valley (several outlets were active during 338 Figure 9. Sequential evolution of the hydraulic gradient within the Beuchire-Creugenat karst system (i.e. the conduits) for an average annual flood event reaching the Creugenat overflow. The profile of the conduits is here supposed. Processes are the following: 1. The groundwater level at the Creugenat overflow is independent of the Beuchire spring discharge oscillations; 2. The water level at the Creugenat overflow is controlled by the threshold (a); 3. At 443 m a.s.l. the activation of an additional conduit (or a perched spring) show a lag in the water level elevation rise at the Creugenat overflow; 4. The rise of the groundwater level at the Creugenat overflow depends on the Beuchire spring discharge. At 451 m a.s.l the Creugenat overflow is now flowing! Figure 10. Model prediction of the extension of groundwater bodies (GWB) A and B during a flood event reaching the Creugenat overflow (=multiannual occurrence =case 4 in Figure 9). Water from GWB B overflows over two passes and contributes to the discharge of the Beuchire spring. Figure 11. Model prediction of the saturated groundwater bodies extension in the BeuchireCreugenat karst system during a flood event reaching the Creux-des-Prs overflow (~30-year flood event).
13TH SINKHOLE CONFERENCE NCKRI SYMPOSIUM 2 about 0,5% (in the swiss Jura, according to Bauer et al. 1980, Burger and Pasquier 1984). This value may be Mapping the flooded areas According to the previous model results it is possible to map surface areas which could be affected by the potential In addition to these gradients some further temporary by Bouvier 2006. They were used as controls for the Considering the respective values of the hydraulic gradients during these events, it is possible to estimate the 339 Figure 12. Model predicted storage in the aquifer (i.e. the karst conduits) and its development due to groundwater increase within the karst system for the respective flood events (average annual, 30-year flood events and ~300-year flood events). The associated volumes refer to water potentially involved in the floods (1.7 Mm 3 for an average annual flood event, 4.4 Mm 3 for a 30-year flood event and more than 6 Mm 3 for a 300-year flood event). Figure 13. Flood hazard map of the Beuchire-Creugenat catchment area. Color code refers to the considered occurrence: average annual flood event, 30-year flood event, 300-year flood event). Filled areas have to be considered as potentially exposed to flooding or at least as potentially impacted by an overflow from a temporary outlet. The interpreted drainage axes (both vadose and phreatic) are also displayed on this map.
NCKRI SYMPOSIUM 2 13TH SINKHOLE CONFERENCE deepening of the collapse. The recorded oscillations of the groundwater are consistent with the previous interpretation related on Figure 9. provides new calibration elements that improve the model functionality. Currently, a series of simulations are being conducted using the actual release of SWMM event. This is possible once the catchments of these have been clearly delineated as well as their respective recharges. This assessment will be conducted in the second part of model. In the present paper only the maximum discharge of the respective outlets (permanent and temporary ones) was construction or hydraulic works. Maximum expected discharges at the outlets Discharge of springs usually increases as the level of the groundwater increases within the system until groundwater level still rises in the conduit, the discharge on that principle, it is possible to estimate the maximum expected discharges at the Beuchire spring and the Regarding the Beuchire spring, the maximum discharge (Figure 8). Even if the groundwater level in the karst system still increases the Beuchire spring cannot The maximum discharge rate at the Creugenat outlet can be estimated from the discharge of the stream in the city was estimated that the maximum discharge rate at the Discussion In June 2012 a collapse appeared in the middle of a road, on a straight line between the Beuchire spring and the of 8 m and at that time its bottom was dry. This collapse gave the SISKA the opportunity to install a pressure sensor with the aim to survey an eventual presence and oscillations of the groundwater within the pit. The recorded data (only few weeks available) show periodic rises of the groundwater with a maximal elevation of 340 Figure 14. The new collapse of the community of Courtedoux gave access to a 8 m deep pit. The bottom of the pit was monitored with a pressure sensor. The recent data allowed validation and improvement of the Beuchire-Creugenat karst system model functionality.
13TH SINKHOLE CONFERENCE NCKRI SYMPOSIUM 2 in conduit geometry characterization and to improve the hydrological model. The last could be applied in further hydraulic planning, especially in estimating the groundwater discharge contribution for each basin unit. Applications to assist the design of future construction and hydraulic works could also be envisaged. Acknowledgments The study of the characterization and prediction of the and caving clubs for providing information and data. References Bauer F, Benischke R, Bub F, Burger A, Dombrowski mit natrlichen und knstlichen Tracern im Neuenburger Jura .Schweiz.. Steirische Beitrge du Doubs .France. et du canton du Jura .Suisse.. Burger A, Pasquier F. 1984. Prospection et captage Suisse. 219 p. (Storm Water Model Management, Rossman 2004) to the related storage within the aquifer. recharges will be assessed and extrapolated to estimate the maximum discharges which could be expected within the conduits. These simulations are expected to produce relevant results that will improve the modeling of the karst system. They will also bring quantitative elements to design future construction and hydraulic works. Conclusion geological model depicting the aquifer basement has data and literature documentation. By following the hydraulic principles in karst hydrology, it was possible to sketch the vadose and the phreatic zones as well as aiming distinguishing their recharge contribution at the next stages of the study. By using the available discharge data of the Beuchire spring and head measurements of a thresholds functioning of the system and to approach the geometry of the groundwater hydraulic within the conduits in the high water stage and to delineate where water is susceptible to reach the ground surface and to enhance the risk of inundation. Areas on ground occurrence of the mapped considered events (average event). Furthermore successive activation of the outlets and their associated discharges are now predictable. Recent integration of piezometric data from a borehole in the vicinity of the main temporary outlet (Creugenat) and the more recent instrumentation and observations in the collapse at Courtedoux (which appeared in June 2012) brought new indicators to control and improve the established model. Current simulations using SWMM and based on these new data may provide new elements 341
NCKRI SYMPOSIUM 2 13TH SINKHOLE CONFERENCE l'effondrement de Courtedoux. Institut Suisse de Jeannin PY, Eichenberger U, Sinreich M, Vouillamoz J, Malard A et al. 2012. KARSYS: a pragmatic approach to karst hydrogeological system conceptualisation. Assessment of groundwater reserves and resources in Switzerland. Environmental Earth Sciences. Available from: karst aquifers: A hydrodynamic modeling approach circulation souterraine en terrains calcaires. Actes de Prudhomme LM. 1804. Dictionnaire universel, Rossman LA. 2004. Storm water management model. user's manual version 5.0. United States Environmental Protection Agency. foreland fold and thrust belt [PhD dissertation]. 342