Soil Moisture Initialization Effects in the CCCma AGCM3: Relationship of Sub-Seasonal Climate Forecast Error to Uncertainty in Soil Moisture Initializations

Abstract

In the second phase of the Canadian Historical Forecasting Project (HFP2), four different Atmospheric General Circulation Models (AGCMs) were run to produce a series of four-month forecasts over a 33-year period. For the HFP2 project, the land surface wetness state was initialized from model climatology rather than estimates of the true initial soil moisture state. In this study, the impact of soil moisture initialization on the monthly forecasts of air temperature is evaluated using one of these four models, AGCM3 from the Canadian Centre for Climate modelling and Analysis (CCCma), which employs the Canadian Land Surface Scheme (CLASS) to represent land processes. A realistic global estimation of soil moisture was produced by running CLASS offline using a bias-corrected meteorological dataset over the global land surface. The relationship between errors in forecasts of near-surface air temperature over the month and the uncertainty in the surface soil moisture initialization is identified for five northern hemisphere warm-season months during the period 1979–2002. The results demonstrate that soil moisture initialization has a statistically significant impact on monthly air temperature forecasts over many regions. The greatest impact was found over the Sahel region in Africa, India and East Asia, regions in Brazil, and Central North America. The intensity and areal extent of this impact increases over extreme dry and wet soil moisture anomalies.