Jens Lange

Karst water resources in a changing world: Review of hydrological modeling approaches

Karst regions represent 7–12% of the Earths continental area, and about one quarter of the global population is completely or partially dependent on drinking water from karst aquifers. Climate simulations project a strong increase in temperature and a decrease of precipitation in many karst regions in the world over the next decades. Despite this potentially bleak future, few studies specifically quantify the impact of climate change on karst water resources. This review provides an introduction to karst, its evolution, and its particular hydrological processes. We explore different conceptual models of karst systems and how they can be translated into numerical models of varying complexity and therefore varying data requirements and depths of process representation. We discuss limitations of current karst models and show that at the present state, we face a challenge in terms of data availability and information content of the available data. We conclude by providing new research directions to develop and evaluate better prediction models to address the most challenging problems of karst water resources management, including opportunities for data collection and for karst model applications at so far unprecedented scales.

A noncalibrated rainfall‐runoff model for large, arid catchments

A distributed, field-based rainfall-runoff model was developed for the 1400-km2 arid catchment of Nahal Zin, Israel. No calibration with measured flow data was performed. The model used rainfall radar input applied over a catchment that was spatially disaggregated into different terrain types according to hydrologically relevant surface characteristics. Hortonian overland flow generation on each type was parameterized independently using values of initial loss and temporal decay of infiltration determined from existing field experiments. Delimited by topography, this catchment wide pattern of rainfall excess was distributed over 850 tributary catchments (model elements). Runoff delivery from the model elements to the adjoining channel segments was timed by applying a mean response function determined in an environmentally similar experimental catchment. Inside the channel network the MVPMC3 method of the Muskingum-Cunge technique was used for streamflow routing, accounting for channel dimensions and roughness. For each channel segment a constant infiltration rate was applied to account for transmission losses and discontinued when the wetting front reached the bottom of the available alluvial storage. Within two model tests, one separate for the routing component (October 1979) and one for the complete model (October 1991), observed hydrographs and reconstructed peak discharges were successfully simulated. The spatially distributed model output showed that during the October 1991 test, tributaries produced preceding peaks that wetted the channel alluvium before the main flood had arrived and transmission losses lost their significance downstream. Total maximum model uncertainty was estimated including the uncertainty ranges of each model parameter. In general, this study shows that field-based data on generation and losses of runoff may be incorporated into a distributed hydrologic model to overcome calibration with the poor data records of arid high-magnitude events.

Modeling spatiotemporal impacts of hydroclimatic extremes on groundwater recharge at a Mediterranean karst aquifer

Karst aquifers provide large parts of the water supply for Mediterranean countries, though climate change is expected to have a significant negative impact on water availability. Recharge is therefore a key variable that has to be known for sustainable groundwater use. In this study, we present a new approach that combines two independent methods for karst recharge estimation. The first method derives spatially distributed information of mean annual recharge patterns through GIS analysis. The second is a process-based karst model that provides spatially lumped but temporally distributed information about recharge. By combining both methods, we add a spatial reference to the lumped simulations of the process-based model. In this way, we are able to provide spatiotemporal information of recharge and subsurface flow dynamics also during varying hydroclimatic conditions. We find that there is a nonlinear relationship between precipitation and recharge rates resulting in strong decreases of recharge following even moderate decreases of precipitation. This is primarily due to almost constant actual evapotranspiration amounts despite varying hydroclimatic conditions. During the driest year in the record, almost the entire precipitation was consumed as actual evapotranspiration and only little diffuse recharge took place at the high altitudes of our study site. During wettest year, recharge constituted a much larger fraction of precipitation and occurred at the entire study site. Our new method and our findings are significant for decision makers in similar regions that want to prepare for possible changes of hydroclimatic conditions in the future.

Identification of a karst system’s intrinsic hydrodynamic parameters: upscaling from single springs to the whole aquifer

For water management purposes, information about an entire aquifer system is generally more important than information about a specific spring. Since a karstic aquifer system might drain to several outlets, conclusions derived from a single spring can be misleading for characterization and modeling. In this study we apply a conceptual model to an Alpine dolomite karst system in Austria. The particular challenge was that several small springs with strongly varying hydrological behavior and diffuse flow into surrounding streams drain this system. Instead of applying the model to a single spring, it was calibrated simultaneously to several observations within the system aiming to identify the karst system’s intrinsic hydrodynamic parameters. Parameter identification is supported by modeling the transport of water isotopes (δ18O). The parameters were transferred to the whole system with a simple upscaling procedure and a sensitivity analysis was performed to unfold influence of isotopic information on parameter sensitivity and simulation uncertainty. The results show that it is possible to identify system intrinsic parameters. But the sensitivity analysis revealed that some are hardly identifiable. Only by considering uncertainty reasonable predictions can be provided for the whole system. Including isotopic information increases the sensitivity of some intrinsic parameters, but it goes along with a sensitivity decrease for others. However, a possible reduction of prediction uncertainty by isotopic information is compensated by deficiencies in the transport modeling routines.

Multi-tracer experiments to characterise contaminant mitigation capacities for different types of artificial wetlands

Salt tracers (sodium bromide/sodium chloride) and two different fluorescent tracers, uranine (UR) and sulforhodamine-B (SRB), were injected as a pulse into six different surface flow wetlands (SFWs). Salt tracers documented wetland hydraulics. The fluorescent tracers were used as a reference to mimic photolytic decay (UR) and sorption (SRB) of contaminants as illustrated by a comparison to a real herbicide (Isoproturon), which was used as a model for mobile pesticides. Tracer breakthrough curves were used to document residence time distributions, hydraulic efficiencies, peak attenuation and retention capacities of completely different wetland systems. A 530 m2 forest buffer zone showed considerable peak attenuation but limited retention capabilities despite its large area. Approximately 80% of SRB was permanently retained in a re-structured 325 m2 flood detention pond. These two non-steady SFWs indicated long-term tracer washout. The remaining four SFWs displayed constant outflow rates and steady-state flow conditions. Due to photolytic decay in a 330 m2 row of three wetlands, UR was almost entirely degraded, but the SRB breakthrough suggested relatively low sorption. A 65 m2 shallow flow-through wetland yielded negligible photolytic decay but showed considerable sorption losses. Finally two types of vegetated ditches were analysed. In one case, vegetation was removed from a 413 m long ditch immediately prior to tracer injection. A 30% loss by sorption to sediment and plant remnants occurred at the very beginning of the tracer breakthrough. Inside a second ditch, 80 m long and densely vegetated by Phragmites australis, sorption was even higher and yielded eightfold higher specific SRB retention rates. Although the present findings are only valid for low flow conditions, they indicate that a shallow water depth seems to be a key variable which may increase sorption of tracers and therefore contaminants. Large wetlands with deep water bodies may attenuate concentrations efficiently, but unit load reduction was found to be more significant in shallow systems even at much higher flow velocities.

Dissipation of hydrological tracers and the herbicide S-metolachlor in batch and continuous-flow wetlands

Pesticide dissipation in wetland systems with regard to hydrological conditions and operational modes is poorly known. Here, we investigated in artificial wetlands the impact of batch versus continuous-flow modes on the dissipation of the chiral herbicide S-metolachlor (S-MET) and hydrological tracers (bromide, uranine and sulforhodamine B). The wetlands received water contaminated with the commercial formulation Mercantor Gold(®) (960 g L(-1) of S-MET, 87% of the S-enantiomer). The tracer mass budget revealed that plant uptake, sorption, photo- and presumably biodegradation were prominent under batch mode (i.e. characterized by alternating oxic-anoxic conditions), in agreement with large dissipation of S-MET (90%) under batch mode. Degradation was the main dissipation pathway of S-MET in the wetlands. The degradate metolachlor oxanilic acid (MOXA) mainly formed under batch mode, whereas metolachlor ethanesulfonic acid (MESA) prevailed under continuous-flow mode, suggesting distinct degradation pathways in each wetland. R-enantiomer was preferentially degraded under batch mode, which indicated enantioselective biodegradation. The release of MESA and MOXA by the wetlands as well as the potential persistence of S-MET compared to R-MET under both oxic and anoxic conditions may be relevant for groundwater and ecotoxicological risk assessment. This study shows the effect of batch versus continuous modes on pollutant dissipation in wetlands, and that alternate biogeochemical conditions under batch mode enhance S-MET biodegradation.

Photolytic transformation products and biological stability of the hydrological tracer Uranine

Among many fluorescence tracers, Uranine (sodium fluorescein, UR) has most widely been used in hydrological research. Extensive use of UR for tracing experiments or commercial use might cause a potential risk of long-term environmental contamination. As any organic substance released to the environment, also UR is subjected to chemical and physical reactions that can be chemical, biological and photolysis processes. These processes transform the parent compound (PC) and have not been extensively investigated for UR. This study applies two OECDs (301 D and 301 F) tests and a screening water sediment test (WST) to investigate the biodegradability of the PC. Photolysis in water was explored by Xe lamp irradiation. Subsequently, the biodegradability of the photolysis mixtures was examined. The primary elimination of UR was monitored and structures of its transformation products (TPs) were elucidated by HPLC-FLD-MS/MS. UR was found not readily biodegradable, although small degradation rates could be observed in the OECD 301 D and WST. HPLC-FLD analysis showed high primary elimination of the tracer during photolysis. However, the low degree of mineralization found indicates that the UR was not fully degraded, instead transformed to TPs. A total of 5 photo-TPs were identified. According to MS/MS data, chemical structures could be proposed for all identified photo-TPs. Likewise the parent compound it was demonstrated that photo-TPs were largely recalcitrant to microbial degradation. Although we did not find indications for toxicity, target-oriented studies on the environmental impact of these photo-TPs are warranted. Results obtained in this study show that deeper investigations are necessary to fully understand fate and risk connected to the use of UR.

The importance of single events in arid zone rainfall-runoff modelling

Abstract Two high magnitude rainstorm floods are simulated by a distributed rainfall-runoff model in the 1400 km2 arid catchment of Nahal Zin, Israel. Only the processes dominating arid zone flood generation (generation and spatial concentration of Hortonian overland flow on the terrain and transmission losses into the dry channel alluvium) are described. No calibration with measured flow data is performed. During one event (October 1991) almost the entire catchment was covered by high intensity rainfall as detected by rainfall radar. During another (October 1979) only one ground station in the uppermost headwaters recorded heavy precipitation, while the majority of the catchment remained dry. For this event the distributed model serves as ‘runoff-rainfall model’ to reconstruct characteristics of the rainfall. Different event characteristics directly affect parameter sensitivity and model uncertainty. Maximum model uncertainty of the diminished October 1979 peak is governed by transmission loss parameters and exceeds 300 %. During 1991 only 90 % is determined for this value and infiltration characteristics of the terrain are more relevant. Also a paleoenvironmental scenario on the hydrological effects of widespread loess deposition is highly event dependent. It may be concluded that the separate analysis of single events is crucial for the understanding of high magnitude floods in arid catchments.

Fluorescent tracers to evaluate pesticide dissipation and transformation in agricultural soils

This study evaluates the mobility and dissipation of two organic fluorescent tracers (uranine, UR and sulforhodamine-B, SRB) in soil from an agricultural field. Two plot experiments were conducted for 2.5months in 2012 and 2016 to compare the behavior of reactive fluorescent tracers (UR and SRB) to the chloroacetanilide herbicide S-metolachlor (S-MET) and bromide (BR), used as a traditional conservative tracer. SRB in top soil closely mimicked the gradual recession of S-MET, while BR overrated both top soil mobility and slow leaching of S-MET in the soil column. In contrast, UR quickly receded in the soil and was entirely dissipated at the end of the study periods. Instead, a strong fluorescent signal that was stable against acidification, and non-traceable in background samples, gradually developed at an excitation wavelength of 510nm in samples from the uppermost soil layer starting 40 (2012) and 22 (2016) days after tracer application. We hypothesize that (bio-)chemical transformation of UR accelerated tracer loss with concomitant formation of the specific transformation product TP510. By LC-MS/MS analysis we propose a probable molecular structure of TP510 and sulfonation as one likely transformation process. Overall, we anticipate our results to be a starting point to use fluorescent tracers in longer term (>2months) agricultural soil studies as a proxy for S-MET and possibly also other organic pesticides, as they are non-conservative in unsaturated soil and may follow similar dissipation and transformation patterns. At the same time their analysis is less costly and they pose smaller environmental risks.

Patterns in the linkage of water quantity and quality during low-flows

This study investigates 72 catchments across the federal state of Baden-Wuerttemberg, Germany, for changes in water quality during low-flow events. Data from the states water quality monitoring network provided seven water quality parameters (water temperature, electrical conductivity, concentrations of chloride, sodium, sulfate, nitrate and phosphate), which were statistically related to streamflow variability. Water temperatures increased during low-flow in summer but decreased during low-flow in winter. Nitrate concentrations revealed high spatial heterogeneity with about one third of the stations showing decreasing values during low discharge. For most other parameters concentrations increased during low-flow. Despite consistent trend directions, the magnitudes of changes with streamflow differed markedly across the state. Both multiple linear regression and a multiple analysis of variances were applied to explain these differences with the help of catchment characteristics. Results indicated that for sulfate and conductivity, geology of the catchments was the most important control, whereas for chloride, sodium and nitrate sewage treatment plants had largest influence. For phosphate no clear control could be identified. Independent from the applied method, land use was a less important control on river water quality during low-flow than geology or inflow from sewage treatment plants. These results show that the effects of diffuse and point sources, as well as those of natural and anthropogenic sources differ for different water quality parameters. Overall, a high diversity of potential water quality deterioration signals needs to be considered when the ecological status of rivers is to be protected during low-flow events.