Sensitivity of summertime weather forecasts to differences in soil moisture and the implication for ensemble weather forecasting

Currently, the land-surface condition is not perturbed in existing operational ensemble forecasts. However, there is much literature suggesting a sensitivity of the weather to nominal changes in land surface. The work described here will test the forecast sensitivity to soil moisture produced by two independent soil moisture analyses as input to weather models. Six summertime case days were selected where the initial conditions were weakly forced with clear skies over the central United States. These case days were chosen because thunderstorms developed within 24 hours of the analysis. For each case day both convectively resolved and parameterized simulations were run with generally large differences in forecast precipitation and 2-m temperatures between forecasts. The differences due to the different soil moisture analyses were comparable to those from different convective parameterizations. For certain synoptic situations, this suggests that soil moisture perturbations will realistically increase the spread of ensemble forecasts.