The Impact of Land Surface and Atmospheric Initialization on Seasonal Forecasts with CCSM

Abstract

Abstract Series of forecast experiments for two seasons investigate the impact of specifying realistic initial states of the land in conjunction with the observed states of the ocean and atmosphere while using the National Center for Atmospheric Research (NCAR) Community Climate System Model, version 3 (CCSM3.0). Since direct soil moisture observations adequate for initialization of the land surface do not exist, this study considers proxy data. The authors are able to successfully initialize all components of the CCSM3.0 and produce a good representation of the mean land surface climate in the first season’s forecast. In comparison with a previous set of forecast experiments that had initialized only the observed ocean state, there is firm evidence that this study produces a better representation of the interannual variability of the soil surface. The representation of soil moisture in the fully initialized seasonal forecasts as measured against the reanalysis is improved, due in part to the ability of the CCSM3.0 to persist large-scale anomalies present in the initial soil state. The improvement in the representation of the land surface, in conjunction with the atmospheric initialization, contributes to a skillful seasonal forecast of surface temperature. There is little evidence of an improved forecast of precipitation over land. Results from this study support the use of the CCSM, originally designed for use as a climate model, as a fully initialized seasonal forecast model. The authors suggest that initialization of the land surface state is crucial for skillful seasonal forecasts made with fully coupled models.