Effects of sounding temperature assimilation on weather forecasting - Model dependence studies

In comparing various methods for the assimilation of remote sounding information into numerical weather prediction (NWP) models, the problem of model dependence for the different results obtained becomes important. The paper investigates two aspects of the model dependence question: (1) the effect of increasing horizontal resolution within a given model on the assimilation of sounding data, and (2) the effect of using two entirely different models with the same assimilation method and sounding data. Tentative conclusions reached are: first, that model improvement as exemplified by increased resolution, can act in the same direction as judicious 4-D assimilation of remote sounding information, to improve 2-3 day numerical weather forecasts. Second, that the time continuous 4-D methods developed at GLAS have similar beneficial effects when used in the assimilation of remote sounding information into NWP models with very different numerical and physical characteristics.