|Title||Assessing physical processes in the ECMWF model forecasts using the ARM SGP observations.|
|Series/Collection||ECMWF-ARM Report Series|
|Authors||Cheinet, S, Beljaars, A, Köhler, M, Morcrette, J-J, Viterbo, P|
|Place of publication||Shinfield Park, Reading|
In this study, we compare short-term weather forecasts of the ECMWF model (IFS) to observations at the Southern Great Plains (SGP) site of the Atmospheric Radiation Measurement (ARM) program in July 2003. By using different ARM instruments and complementary satellite and radar network data, a number of systematic deficiencies in the representation of cloud and mixing processes are characterized in the IFS. The IFS correctly predicts the mean flow. However, it misrepresents the intensity of rainy deep convective systems, and their synoptic-scale propagation from the Rockies. It also underestimates the nocturnal Low-Level Jet (LLJ) and the diurnal temperature range. The night-time deficiencies are improved by decreasing the intensity of mixing in the stable boundary layer. The afternoon cold bias in fair-weather days is consistent with a lack of vertical mixing related to an underestimation of the frequency of shallow cumuli, which probably results from the systematic dry bias at low levels. The lower level moisture in the SGP is remotely controlled through the competition between the LLJ advection and the convective processes that reduce the LLJ moisture along the path from the Gulf of Mexico. Sensitivity experiments performed in this study suggest that the model dry bias is caused by a misrepresentation of this mechanism. Implications of our results in terms of parameterization development and evaluation are discussed.