The potential impact of proper land-surface representation on spring dynamical seasonal predictions

We examine the contribution of the proper representation of land-surface conditions, in particular snow and soil moisture, toward improving spring forecasts within the Dynamical Seasonal Prediction (DSP) framework. We construct a suite (i.e. multi-year) of seasonal ensemble hindcasts similar to the DSP framework with the COLA general circulation model (GCM), but using the best possible land-surface initial conditions. The land-surface conditions are derived by forcing the same land-surface scheme coupled to the GCM with observed/analyzed near surface meteorology in a manner similar to the GEWEX Global Soil Wetness Project. These ensemble hindcast experiments are designed to explore the impact of proper initialization of the continental surfaces toward improving predictability. In addition, upper-limit tests of predictability (as well as predictive skill) are being performed by prescribing land-surface conditions throughout the integrations (which serve as a reflection of perfect knowledge/prediction of the continental conditions). Special attention is given to the proper representation of snow cover.