The water stored in and flowing through the subsurface is fundamental for sustaining human activities and needs, feeding water and its constituents to surface water bodies and supporting the functioning of their ecosystems. Quantifying the changes that affect the subsurface water is crucial for our understanding of its dynamics and changes driven by climate change and other changes in the landscape, such as in land-use and water-use. It is inherently difficult to directly measure soil moisture and groundwater levels over large spatial scales and long times. Models are therefore needed to capture the soil moisture and groundwater level dynamics over such large spatiotemporal scales. This thesis develops a modeling framework that allows for long-term catchment-scale screening of soil moisture and groundwater level changes. The novelty in this development resides in an explicit link drawn between catchment-scale hydroclimatic and soil hydraulics conditions, using observed runoff data as an approximation of soil water flux and accounting for the effects of snow storage-melting dynamics on that flux. Both past and future relative changes can be assessed by use of this modeling framework, with future change projections based on common climate model outputs. By direct model-observation comparison, the thesis shows that the developed modeling framework can reproduce the temporal variability of large-scale changes in soil water storage, as obtained from the GRACE satellite product, for most of 25 large study catchments around the world. Also compared with locally measured soil water content and groundwater level in 10 U.S. catchments, the modeling approach can reasonably well reproduce relative seasonal fluctuations around long-term average values. The developed modeling framework is further used to project soil moisture changes due to expected future climate change for 81 catchments around the world. The future soil moisture changes depend on the considered radiative forcing scenario (RCP) but are overall large for the occurrence frequency of dry and wet events and the inter-annual variability of seasonal soil moisture. These changes tend to be higher for the dry events and the dry season, respectively, than for the corresponding wet quantities, indicating increased drought risk for some parts of the world.

Groundwater is a sensitive component affected by climate change. Modelling the dynamics of groundwater levels is inherently difficult particularly as the response to climate change. Given this complexity, most of the current studies using long term groundwater time series were conducted by statistical analysis or using over simplified assumptions to represent the physical processes in hydrological system. With the objective of providing an improved physically based groundwater modelling approach to support climate change impact assessment, a dataset of long term time series of groundwater levels from two different soil types (sand and till) were selected from the Tärnsjö area located in southeast of Sweden. The CoupModel was chosen to perform the simulation since it offers a physically based representation on groundwater recharge processes. A two-step strategy for calibration with first short-term calibration followed by long-term testing was adopted. Simulated groundwater levels followed the general patterns of measured groundwater level dynamics; however, auto-correlations and periodicities were observed in residuals for all sites of which two sandy soil sites with deeper groundwater tables maintained strong auto-correlations in long time lags and an extra 15.4-year periodicity. The long memory of the system rendered it more susceptible to climate change. Uncertainty arises if different initial condition had been applied in short term period calibration.

No description available.