To reduce the spatial correlation of representation error in observations and computational complexity, we propose a thinning scheme that can extract typical observations within a certain range. This scheme is applied...To reduce the spatial correlation of representation error in observations and computational complexity, we propose a thinning scheme that can extract typical observations within a certain range. This scheme is applied to the Global/Regional Assimilation and Prediction System(GRAPES) with three-dimensional variation(3 DVAR) to study the effect of the thinning radius on the assimilation results. The assimilation experiments indicate that when the ratio of the model resolution to the observational resolution is 1:3, the simulated results for precipitation are relatively good and have a relatively high equitable threat score(ETS). Moreover, the analysis errors in the temperature and the specific humidity are the smallest, the dependence of the norm gradient vector of the objective function on the number of iterations is slow, gentle, and close to 0, and the minimization results in improved conditions.展开更多
基金Young Meteorological Research of Jiangsu Provincial Meteorological Bureau(Q201611)Key Scientific Research Projects of Jiangsu Provincial Meteorological Bureau(KZ201605)+1 种基金Natural Science Foundation of Jiangsu Province(BK20161074)Beijige Fund of Jiangsu Institute of Meteorological Sciences(BJG201512)
文摘To reduce the spatial correlation of representation error in observations and computational complexity, we propose a thinning scheme that can extract typical observations within a certain range. This scheme is applied to the Global/Regional Assimilation and Prediction System(GRAPES) with three-dimensional variation(3 DVAR) to study the effect of the thinning radius on the assimilation results. The assimilation experiments indicate that when the ratio of the model resolution to the observational resolution is 1:3, the simulated results for precipitation are relatively good and have a relatively high equitable threat score(ETS). Moreover, the analysis errors in the temperature and the specific humidity are the smallest, the dependence of the norm gradient vector of the objective function on the number of iterations is slow, gentle, and close to 0, and the minimization results in improved conditions.
