复杂地形下地面观测资料同化II. 模式地形与观测站地形高度差异代表性误差

第I部分研究结果(徐枝芳等,2007)表明模式与实际观测站地形高度差异对地面观测资料同化效果有较大影响。此文在MM5_3DVAR同化系统中利用近地层相似理论将地面观测资料进行直接三维变分同化分析,考虑模式与实际观测站地形高度差异对同化效果的影响,提出在地面观测误差中增加地形代表性误差来解决这个问题。研究结果表明:地面资料同化分析时,在其观测误差中加入一项新的误差——地形代表性误差,能较好地解决地面资料同化分析中模式与观测站地形高度差异问题;地面资料参与同化分析,在观测误差中加入与模式和实际观测站地形高度差异大小相关的地形代表性误差时,地面观测值对分析值的影响随着地形高度差异代表性误差的加入而减小,同时又部分地将地面观测信息通过变分分析融进分析场,使得低层分析更接近真实场,且地面资料利用率更高,24小时降水数值预报(模拟)的效果较好。

A Study of Assimilation of Surface Observational Data in Complex Terrain Part II: Representative Error of the Elevation Difference Between Model Surface and Observation Site

Since the topography distribution of China is very complex,there exists elevation difference between surface observation station measurement and numerical model surface.The surface observation data are currently only used as backup or process analysis,but not used in numerical models sufficiently.Reasonable assimilation schemes of surface observation data will put these data into numerical models sufficiently.In Part I,results have shown that the difference in elevation between observation site and numerical model surface can impact the results of surface observation data assimilation.In this paper,surface observation data are assimilated into meso-scale model MM5 directly, based on the similarity theory,using MM5_3DVAR(MM5 three-dimensional variational) system.A new representative error,i.e.Terrain Error of Representativeness(TER),is introduced to solve the influence of elevation difference between observation site and model surface.TER is added into surface observation error of surface observation data assimilation in MM5_3DVAR system.Some experiments are performed with surface observation data and radiosonde data.Results show that the influence of elevation difference between observation site and model surface can be solved when TER is added.Better results can be obtained when surface observation data are assimilated into a numerical model with TER,which is related to the elevation difference between model surface and observation site.Using this method,surface observation data bring less contribution to the analysis field,while some useful information is put into the analysis field through data assimilation,and the lower analysis field is closer to the actual field.The surface data are used more sufficiently.