The Improvement Of GRAPES-3DVAR Top Level Interpolation Scheme And Its Impact On The Application Of Satellite Data Analysis

GRAPES (Global/regional assimilation and prediction system) model on upper atmosphere Forecast ineffective mainly due to:the three-dimensional variational data assimilation system high-level background field structure program uses flawed extrapolation scheme, the background field and satellite observations in the upper atmosphere have large discrepancies, the satellite high-level channel data were shut down in the assimilation process, thus affecting the assimilation model,cause the high-level observation data assimilation inefficiency and forecast ineffective, then resulting in GRAPES model in the end.To address the problem, we started a theoretical analysis, and design research programs:First, starting one-dimensional variable subsystem simulation test, executing quantitative analysis of high-level extrapolation error and scope of the program; Secondly, the improvement program, using a one-dimensional simulation analysis to prove the feasibility of the improve scheme; Finally, using the GRAPES system to do cycle assimilation Experiment, analysis improvement program assimilation effect. Concluded that:1. The one-dimensional variational simulation experiments show that:the background field profile constructed by extrapolation scheme with the real situation is very different, there is a huge bias the inversion results not only in high-rise, and there are also varying degrees of impact in low-level; when high-level satellite channel is closed, you can ease the high-level background and observation gap and Eliminate a series of questions, the inversion results may be revised, verify the current variational scheme the senior satellite channel data were rejected.2. The improve scheme Simulation results show that:the upper atmosphere climate data structure high-level temperature and humidity background field profile, can Simulate the actual situation of the upper atmosphere very well; using the smooth correction climate value as a high-level background field in one-dimensional variational data assimilation experiment, the result shows that the inversion results compared with the extrapolation scheme has improved significantly.3. Using the improve scheme for the GRAPES-3DVar, doing a ten-day cycle assimilation experiment, the analysis shows:climate value improvement program to adjust the high-level temperature and humidity background field, also can effectively improve the accuracy of the analysis field, geopotential height have a good improvement in the middle and lower level, correction effect of other variables is also very obvious. Proved climate value improvement scheme to replace the current GRAPES model assimilation system high-level interpolation scheme is feasible.