Monthly Mean Forecast Experiments with the GISS Model

Abstract

The GISS general circulation model is used to compute global monthly mean forecasts for January 1973, 1974 and 1975 from initial conditions on the first day of each month, with ocean surface fluxes based on climatological mean January sea-surface temperatures. Forecasts are evaluated in terms of global and hemispheric energetics, zonally averaged meridional and vertical profiles, forecast error statistics, and monthly mean synoptic fields. Although it generates a realistic mean meridional structure for the month of January, the model does not adequately reproduce the observed interannual variations in the large-scale monthly mean energetics and zonally averaged circulation. The model exhibits no general skill in predicting the monthly mean sea-level pressure field, but it does simulate observed changes in the intensity of the Icelandic low from year to year. For each January the model produces a prognostic monthly mean 500 mb height field that is superior to climatology and persistence. The impact of temporal sea-surface temperature variations on monthly mean global forecasts with the GISS model is investigated by comparing two parallel forecasts for January 1974, one using climatological ocean temperatures for the surface flux computations and the other observed daily ocean temperatures. In the one case studied, the use of daily-updated sea-surface temperatures produced no discernable beneficial effect on the forecasts, and the total impact on the large-scale pressure and temperature fields was small.