The Influence of Initial Soil Wetness on Medium-Range Surface Weather Forecasts

Abstract

The influence of initial soil wetness on surface weather forecasts was quantitatively assessed through the use of the Center for Ocean?Land?Atmosphere Interactions (COLA) general circulation model with an advanced simple biosphere model. The sensitivity of the COLA GCM to changes in initial soil wetness (ISW) is determined by repeating three 10-day integrations with the same initial and boundary conditions as the control runs except the values of ISW, which are revised at 69 model grid points covering much of the continental United States. It is found that the relationship between the changes in the 5-day mean forecasts of surface air temperature and surface specific humidity and the changes in ISW depends upon vegetation type and the values of ISW, and is approximated by regression equations. With the ISW revised based on these regression equations, the first 5-day mean surface air temperature and mean surface relative humidity forecast errors over the relatively dry western portion of the domain are reduced from 2.9° to 1.1°C and from 15% to 7.6%, respectively. Somewhat smaller surface forecast improvements occur for the following 5 days. The impact on the upper atmosphere is small and is largely confined to lower levels. It is also found that the model soil wetness has strong persistence. Therefore, additional forecast experiments are carried out in which the initial soil wetness for a 10-day integration is revised based on the surface forecast errors for the preceding 5-day mean. This results in a reduction of the first 5-day mean surface air temperature and surface relative humidity forecast errors from 2.4° to 1.3°C and from 15% to 8%, respectively, averaged over the dry region. This study suggests the importance of accurate initial soil wetness for medium-range surface weather forecasts. The regression method developed in this study could be readily used operationally to initialize the soil wetness field for medium-range forecasting.