An Objective Approach for Correcting Model Bias in the GEOS-5 AGCM

An objective approach is introduced to correct model biases in general circulation models (GCMs). This objective approach utilizes the history of the model’s short-term forecasting errors and their seasonal dependence to modify each model’s tendency term. The fields u,v,T,q,ps are then, in fact, adjusted towards the climate of the analyzed fields in the atmosphere of each model integration. An example is presented for the bias correction in the NASA Goddard Earth Observing System Model, Version 5 (GEOS-5) Atmospheric GCM (AGCM). The bias correction takes advantage of the history of analysis increments (“errors”) from The Modern-Era Retrospective Analysis for Research and Application-2 (MERRA-2). The study shows that such an approach removes most of the model biases relative to MERRA-2, including the eastward shift of tropical precipitation and convection in the Pacific and Atlantic, the weak and disoriented north Pacific mean jet stream, the prominent surface warming bias over NH continents, and the dry biases in the US Great Plains. A number of other aspects of the model simulation (e.g. extratropical transient activities) also improve considerably in the model due to the imposed pre-processed bias corrections. The impacts of this bias correction on the sub-seasonal predictions are also assessed in this study. Because many regional biases in the GEOS-5 model are common amongst other current models, our approaches and findings are applicable to these other models as well.