The Use of Indirect Estimates of Soil Moisture to Initialize Coupled Models and its Impact on Land Surface/Atmosphere Interactions

It is well known that soil moisture is a characteristic of the land surface that strongly affects the partitioning of outgoing radiation into sensible and latent heat which significantly impacts both weather and climate. Detailed land surface schemes are now being coupled to mesoscale atmospheric models in order to represent the effect of soil moisture upon atmospheric simulations. However, there is little direct soil moisture data available to initialize these models on regional to continental scales. As a result, a Soil Hydrology Model (SHM) is currently being used to generate an indirect estimate of the soil moisture conditions over the continental United States at a grid resolution of 36 km on a daily basis since 8 May 1995. The SHM is forced by analyses of atmospheric observations including precipitation and contains detailed information on slope, soil and landcover characteristics.

The purpose of this paper is to evaluate the utility of initializing a detailed coupled model with the soil moisture data produced by the SHM. Our coupled modeling system is composed of the Biosphere-Atmosphere Transfer Scheme (BATS) and the PSU/NCAR mesoscale model version 5 (MM5). In an attempt to improve the land surface hydrology simulated by MM5/BATS, we have added the capability of replacing the BATS soil hydrology subroutines with those of the Simulator for Hydrology and Energy Exchange at the Land Surface (SHEELS). The representation of soil water processes in SHEELS was patterned in many regards (infiltration, runoff, Darcy flow) after those in the SHM. These changes were made to produce better simulations of the strong near-surface moisture and temperature gradients that are often observed, especially in areas of sparse vegetation.

Our results will focus on the impact of using the SHM-derived soil water content fields upon the simulated land surface energy and water budgets produced by MM5/BATS over the continental U.S. on a 36 km grid. The model will be applied on both short-term (0-48hr) and seasonal (120 day) time scales during the Spring-Summer 1999 seasons. The assessment procedure consists of comparing model predicted 2 meter air temperature and specific humidity with all available observations. Components of simulated surface energy budget will be compared with observations from available flux sites such as those from the Energy Balance Bowen Ratio systems deployed at the CART/ARM sites in the Southern Great Plains and the Walker Branch site in Tennessee. In addition, the same assessment procedure will be applied to simulations produced by the NCEP Eta model which also includes a detailed land surface scheme and is initialized with an indirect estimate of soil moisture.