The Effect of the Groundwater Representation on Land surface-Atmosphere Feedbacks in Regional Climate Simulations

Recent studies indicate the high impact of land surface-atmosphere feedbacks in climate change. In particular, feedbacks such as the soil moisture-temperature and the soil moisture-precipitation feedback contribute to the uncertainty of short-range to medium-range weather forecasts as well as climate change scenarios. Yet, in the applied models, the connection of groundwater dynamics with shallow soil moisture and bedrock-to-atmosphere interactions are rarely considered. Moreover, most bedrock-to-atmosphere modeling studies were performed at the catchment scale or rely on highly parameterized groundwater representations at the continental scale. Therefore, a consistent continental, physics-based modeling approach, which closes the terrestrial hydrologic and energy cycles is needed to better understand the physical processes at the continental scale. Our hypothesis is that groundwater dynamics significantly affect land surface-atmosphere feedbacks at the continental scale and contribute to the land surface energy partitioning. In order to test this hypothesis, we present the development of a fully coupled aquifer-to-atmosphere modeling system over the European CORDEX domain. The integrated Terrestrial Systems Modeling Platform, TerrSysMP, consisting of the three-dimensional subsurface and overland flow model ParFlow, the Community Land Model CLM3.5 and the numerical weather prediction model COSMO of the German Weather Service, is used. The model is set up with a spatial resolution of 0.11° (12.5km) and closes the terrestrial water and energy cycles from aquifers into the atmosphere. The modeling system explicitly simulates river flow and considers three-dimensional groundwater dynamics. We perform simulations of the fully coupled system over the 2003 European heat wave and compare it to a reference simulation, which uses a one-dimensional free drainage approach often applied in land surface schemes of regional climate models. First, we use these simulations to address the uncertainty of regional climate simulations with respect to the groundwater representation during an extreme event. A sensitivity analysis indicates that the groundwater representation has a significant influence on land surface-atmosphere feedbacks. In particular, the simulations with a simplified groundwater representation produce systematically higher daily maximum temperatures than the simulations with 3D groundwater dynamics. Even within one-day forecasts, the varying terrestrial simulations can produce an average difference of up to 1°C for the daily maximum temperature. Furthermore, the initialization of the (unknown) subsurface has a significant impact on the evolution of land surface-atmosphere feedbacks and the respective atmospheric state. Secondly, we compare the simulations to observations and reanalysis products, such as the driving ERA Interim atmospheric reanalysis, the respective ERA Interim Land product and FLUXNET observations. The comparison to the ERA Interim Land product shows large spatial differences for soil moisture and energy fluxes, and reaffirms the need for consistent observations and modeling studies. The attached figure shows a snapshot of soil moisture (coloured) and cloud water content (grey) from fully coupled simulations of terrestrial water and energy cycles from aquifers into the atmosphere over the European CORDEX domain.