The Importance of Scale-Dependent Groundwater Processes in Seasonal Drought Forecasts over the Central United States

Summer warm biases over the central United States have been a persistent issue in many global and regional modeling systems. Soil moisture anomalies coincident with these biases can persist for months. Soil moisture interactions with deep groundwater involve processes at timescales spanning from weeks to years. The effect of soil moisture processes affecting surface energy partitioning has been studied extensively in recent years and has the potential of intensifying the hydrologic cycle. Generally, LSMs have not considered the interactions of groundwater with soil moisture in model-resolved soil layers. Using the default Weather Research and Forecasting (WRF) modeling system, regional climate simulations using a CONUS domain have shown a summer warm bias of greater than 1°C over the entire central U.S. spanning from Texas to Minnesota and eastern Colorado to Indiana, with a core bias of 6+°C located around Iowa.

A simple groundwater module is coupled to the Noah-MP land surface model in WRF. The groundwater module considers vertical processes, such as recharge to the aquifer, and horizontal processes, such as lateral flow to adjacent cells and connectivity to sub-grid rivers. Without groundwater, the free drainage lower boundary condition in the soil model will result in a continuous loss of water throughout the simulations. However, with groundwater, by later summer, water is being transported from the groundwater to the active soil layers and provides a source of moisture. This additional moisture source alleviated the warm bias across the entire central U.S. Comparing simulation results to monthly MODIS evapotranspiration and Stage IV radar/gauge precipitation, the simulations with groundwater have much better performance, especially at the end of the summer season. Another important aspect of the groundwater effect is apparent scale dependence that has arisen when conducting simulations at different resolutions. Results are shown at a range of scales (30m to 1km) to assess the resolution necessary to capture these groundwater-atmosphere interactions, which contribute to reducing the summer warm and dry bias. This study emphasizes the importance of including groundwater in seasonal forecasts for drought prediction and that drought indices should include a groundwater component, especially in regions where groundwater is an active participant in the water cycle.