Improving subsurface hydrology in Earth System Models

Abstract:

Hydrologic processes that govern storage and transport of soil water and groundwater can have strong dynamic relationships with biogeochemical and atmospheric processes. This understanding has lead to a push to improve subsurface hydrologic parametrization in Earth System Models. Here we present results related to improving the implementation of soil moisture distribution, groundwater recharge/discharge, and subsurface drainage in the Community Land Model (CLM) which is the land surface model in the Community Earth System Model. First we identified geo-climatically different locations around the world to develop test cases. For each case we compare the vertical soil moisture distribution from the different implementations of 1D Richards equation, considering the boundary conditions, the treatment of the groundwater sink term, the vertical discretization, and the time stepping schemes. Generally, large errors in the hydrologic mass balance within the soil column occur when there is a large vertical gradient in soil moisture or when there is a shallow water table within a soil column. We then test the sensitivity of the algorithmic parameters that control temporal discretization and error tolerance of the adaptive time-stepping scheme to help optimize its computational efficiency. In addition, we vary the spatial discretization of soil layers (i.e. quantity of layers and their thicknesses) to better understand the sensitivity of vertical discretization of soil columns on soil moisture variability in ESMs. We present multivariate and multi-scale evaluation for the different model options and suggest ways to move forward with future model improvements.