Faculté des sciences

Evaluating the effect of climate change on groundwater resources : from local to catchment scale

Möck, Christian ; Hunkeler, Daniel (Dir.) ; Brunner, Philip (Codir.)

Thèse de doctorat : Université de Neuchâtel, 2013.

There is strong evidence that climate is changing and will affect the water resources. A major question arising from the evaluation of climate change (CC) impacts on groundwater resources is to what extent groundwater recharge will change. Given that for Switzerland, climate models predict more frequent hot and dry summers in the future while precipitation will tend to increase in winter, a... More

Add to personal list
    There is strong evidence that climate is changing and will affect the water resources. A major question arising from the evaluation of climate change (CC) impacts on groundwater resources is to what extent groundwater recharge will change. Given that for Switzerland, climate models predict more frequent hot and dry summers in the future while precipitation will tend to increase in winter, a special attention was given to possible changes in the seasonal distribution of recharge. However, to provide robust predictions, uncertainty has to be considered in all simulations. Three uncertainty sources can be distinguished: the latter can originate from climate models uncertainty, the unknown evolution of land use and society in general, and the hydrological models themselves. The role of these three types of uncertainty has received a major attention in this study. Three studies were carried out to evaluate the effect of CC on the hydrological system. Two of these studies were dedicated to the topic of groundwater recharge whereas the third was focused on the CC response of an aquifer system.
    The first recharge related study deals with the question of how uncertainty due to climate models interacts with uncertainty associated with different hydrological models. Although different models were used to simulate groundwater recharge in numerous climate impact studies, it is not yet clear whether models of different complexity give similar recharge predictions for a given climate scenario. Therefore, five different commonly used approaches to simulate groundwater recharge were compared under CC.
    In this analysis models with different complexity were applied over a time span of several years and predictive model bias occurs. Using CC data with more extreme weather conditions increases the resulting bias. The potential for model predictive error increases with the difference between the climatic forcing function used in the CC predictions and the climatic forcing function used in calibration period. The difference between the reference recharge and simulated recharge from physical based but homogenous model as well as semi-mechanistic model are smallest whereas the differences increase with the simple models. The differences are due to structural model deficits such as the limitation of reproducing preferential flow. Thus, results of CC impact studies using the soil water balance approach to estimate recharge need to be interpreted with caution, although the majority of CC impact assessment studies are using this approach. Comparison of both uncertainties, i.e. CC and model simplification, indicate that the highest uncertainty is related to CC, but a model simplification can also introduce a significant predictive error.
    The second recharge related study explores how different crops and crop rotations influence CC effects on groundwater recharge. The predicted temperature increase will doubtlessly lead to an increase in evaporation and can be intensified by the presence of crops. To address this question, we relied on lysimeter data to ensure that the models represent previously measured crop specific effects on groundwater recharge appropriately before attempting to simulate future trends. In addition to effects of crop types, effects of soils types were considered. To study the effect of soil types on recharge was possible thanks to the presence of three Swiss dominant soil types in the lysimeter facility. This study attempts to explore the combined effect of CC and changes in land use on groundwater recharge. We address these questions by combining numerical modeling techniques with high quality lysimeter data. The simulated results of the 1D numerical model indicate that for most crops a decreasing trend occurs (between -5 to -60%) due to higher evapotranspiration rates. However, for catch crops (fast-growing crop that is grown between successive plantings of a main crop) such as Phacelia and Temporary grassland, an increasing recharge trend can also be observed (up to 15%). Using these catch crops in a crop sequence can buffer the decreasing trend in future recharge rates, but the buffer capacity depends strongly on the growing season.
    It is very likely that crop parameters such as leaf area index (LAI) and root depth (RD) will change in future due to increasing water stress (reduced water content in the lysimeter). Therefore, an analysis of the sensitivity of LAI and RD on recharge was carried out. It was found that simulated recharge is inversely related to LAI and RD where recharge is more sensitive to a decrease in LAI than to RD. Therefore, recharge estimates based on literature LAI and RD values probably represent an upper boundary on recharge rate changes for the future. However, in all simulations a high predictive uncertainty in results is given due to the variability originating from general circulation model (GCM) and regional climate model (RCM) combinations and stochastic realisations of the future climatic conditions.
    The final study explored how changes in groundwater recharge might influence groundwater levels for a small aquifer used for water supply. The soil-unsaturated zone-groundwater system was considered as a whole using the physically based model HydroGeoSphere (HGS). The model was based on a wide range of field data. The main objective of this part was to evaluate if seasonal shifts of groundwater recharge can lead to lower groundwater levels in late summer and a potential water shortage. Such effects are mainly expected for highly transmissive systems with a low storage capacity that are expected to react rapidly to seasonal variations in recharge. Therefore, a small aquifer consisting of highly permeable glacio-fluvial deposits and used for water supply for a small town was selected.
    The physically based model HydroGeoSphere (HGS) was used to simulate changes in recharge rates and groundwater levels based on 10 GCM (Global Circulation Model) - RCM (Regional Climate Model) combinations for the A1B emission scenario. Future recharge rates were compared to rates observed during historical drought periods. The recharge drought frequency was quantified using a threshold approach. The flow simulations indicate that the strongest effect of CC on recharge occurs in autumn and not in summer, when the temperature changes are the highest. For the winter season, recharge rates increase for almost all climate model chains and periods. In summer and autumn, temporal water stress, which is defined as reduced drinking water supply, can occur but the intensity depends on the chosen climate model chain. The uncertainty which comes from the variability among different model chains is large although all climate model chains show the same trend in the recharge seasonality. An estimation of drought frequency for a “worst-case” scenario indicates an increase in frequency and intensity under predicted CC. For the water supply in Wohlenschwil, water shortage will most likely more frequently occur in summer and autumn whereas no water stress is predicted for all other seasons.
    All studies demonstrated that the uncertainty surrounding projected recharge rates and groundwater levels are relatively large. Some model chains indicate decreasing recharge and groundwater levels until the end of century, while other show increasing trends. For instance for the Wohlenschwil aquifer a change in annual recharge between -16% and 12% was simulated, while the mean of all climate model chains indicate no changes. Therefore, it is quite difficult to state on the magnitude of the change with high confidence. However, not the mean is important, but rather the seasonality. Almost all climate model chains lead to a change in seasonality but with a different magnitude. In addition, the uncertainty linked to the interannual variability of the climate is highly uncertain and can lead to strongly different results and conclusions depending on analyzed equiprobable stochastic realisations. However, the main uncertainty is linked to GCM-RCM combinations. This uncertainty is followed by the uncertainty originated by natural variability of the climate and model simplification. The calibration of the hydrological model is a further uncertainty, but could be reduced by improving the model calibration, if needed.
    Although uncertainty in all predictions makes it difficult to state on the magnitude of the change with high confidence, it becomes obviously that a proper consideration of possible effects of CC on groundwater are needed. Results indicate that groundwater is only slightly effected in northern Switzerland on an annual basis but temporal changes can lead to periods with low recharge rates and groundwater tables and therefore to limit water supply.