Increase in Near-Surface Temperature Simulation Skill due to Predictive Soil Moisture in a Numerical Seasonal Simulation under Observed SST Forcing

Abstract

Two sets of 12-yr eight-member ensemble integrations were run with the Experimental Climate Prediction Center?s (ECPC) seasonal forecast model (SFM) to investigate the sensitivity of near-surface temperature skill to evolving soil moisture. The first ensemble had evolving soil moisture, which was fully interactive with the atmospheric component of the model. The second ensemble had soil moisture fixed to the monthly climatological value. Several regions showed an increase in skill in the evolving soil moisture ensemble, including northeastern Australia, southeastern Africa, Europe, northern Brazil, Western Australia, northwestern Russia, Argentina, western Canada, and Indo-China. A survey of these regions showed that most had sensitivity to soil moisture following a peak rainy season, and of those, most also had a high soil moisture time-lag correlation (soil moisture memory) at that time. In a few of the regions high year-to-year soil moisture variability was an additional potential source of soil moisture sensitivity. The sensitivity to soil moisture is considered both in terms of actual predictability (anomaly correlation with observations) and theoretical potential predictability. It was found that the regions listed above, with evolving soil moisture, have anomaly correlations that are close to the potential predictability, suggesting that the model estimates near-surface temperature in these regions as well as could be expected. However, it was also found that when comparing the two sets of integrations, improvements in potential predictability of one ensemble over the other did not necessarily give a reasonable estimate to improvements in anomaly correlation.