DYNAMICS OF SYSTEMATIC-ERRORS IN THE NMC MEDIUM-RANGE FORECAST MODEL

A simple error vorticity model is used to study processes contributing to the evolution of the 300-hPa systematic nondivergent Row errors in the National Meteorological Center Medium Range Forecast model (MRF) during the 1992-93 winter season. The error model is forced by two sou rce terms representing, respectively, systematic errors in the irrotat ional Row and transient eddy vorticity fluxes. The results indicate th at the model simulates reasonably well the development of many of the large-scale features of the zonally asymmetric part of the MRF systema tic nondivergent Row errors, but it fails to simulate the zonally symm etric portion unless an extra forcing term representing systematic err ors in the irrotational flow analyses is included. Two independent met hods are employed to estimate the systematic errors in the irrotationa l Bow analyses. The two estimates are highly correlated, and both indi cate that the analyzed irrotational Row in the Tropics is too weak. Th e magnitude of the estimated analysis errors is of the same order as t he difference between the divergence in the 10-day forecasts and the a nalysis. Globally, systematic errors in the irrotational flow dominate the evolution of nondivergent Bow errors during the first few days of the model integration. Beyond 5-6 days, the extratropical error evolu tion is determined mainly by the integrated effects of systematic erro rs in the transient eddy vorticity fluxes. In the extratropics, transi ents eddy vorticity flux errors appear to be the major factor in produ cing systematic errors in the zonal mean nondivergent flow. In the Tro pics, the rapid development of the zonal mean easterly wind bias is di rectly related to systematic irrotational Row errors. The authors post ulate that early in the forecasts systematic errors in the irrotationa l Bow associated with deficiencies in parameterized tropical convectio n force extratropical stationary wave errors, which in turn lead to sy stematic changes in the midlatitude storm tracks. The transient eddy f lux errors associated with the altered storm tracks then feedback posi tively on the initial stationary wave errors and, after several days, become the dominant source of systematic rotational flow errors.