Application Of MM5 Adjoint Assimilation System In Model Terrain Modification

Model errors and data errors are the limiting factors to the improvement of the numerical weather prediction (NWP) models. In four-dimension data assimilation, the model is often assumed to be precise, with correction made of initial observations and negligence of the results error caused by the model e-rror. Adjoint-model assimilation system (AMAS) can be used in the correction of initial conditions and model parameter. The Four-Dimensional variational d-ata assimilation (4DVAR) methord, AMAS and the theory of terrain modification based on the AMAS are. introduced. Model terrain adjustment is carried o ut by means of the Pennsylvania State University (PSU)/National Center for A tmospheric Research (NCAR) non-hydrostatic mesoscale model MM5 with its AMAS including non-conventional cloud-derived winds and humidity profile in different samples. Two kinds of consensus terrains based on the average consensus and reliable consensus forecast technique are calculated and compared w ith the uncorrected model in different cases. The results are as followed: 1) T-he corrected terrain coordinating with the intial observations and the model c-an improve the model results in the variation of the rang and intensity of the meso-scale heavy-rain center. 2) Direct-assimilation of the non-conventional data in the modle can improve the model results and cause an indeterminable e-ffect comparing to the corrected model. 3) Both the two consensus terrain can do a positive effect in proving the model results, the reliable consensus terrain will do better than the average consensus terrain to improve the model results in the variation of the intensity of the storm rainfall.