The Study Of Downscaling Method For TRMM Products In China Region

Precipitation has an important role in the whole water cycle of the earth,and its amount play an important role in hydrology,climate analysis.With the development of remote sensing technology,the observion of precipitation form satellite remote sensing becomes more familiar.As one of them,the Tropical Rainfall observations(Tropical Rainfall Measuring Mission TRMM)satellite precipitation studies have provided a wide range of continuous precipitation products,which has a high precision,and has been widely applicated.But in a small range of hydrological studies,the spatial resolution of TRMM precipitation products is relatively low,the need for downscaling processing is put forward.To overcome this drawback,this paper attemped a downscaling method which fused multi-source data to improve the resolution of TRMM rainfall products in Chinese region,in order to get a spatial resolution of 1km×1km.In this paper,the study area was in China region(98 °E-135 °E,17 °N-50 °N),based on the technology of remote sensing and geographic information,and the meteorological observation data of the site as the true value,we validated TRMM products from 2001 to 2010;in the relationship of rainfall between geographical factors,topographic factors and vegetation factors,we proposed four statistical downscaling models by fusing multi-source data to improve the resolution of TRMM precipitation products;by comparing the error conditions of the results from four downscaling models,we obtained the optimum downscaling model,in order to obtain the area rainfall of China Region with a higer resolution,and analyze the spatial variability of the precipitation characteristics of China Region;at the same time,we used the rainfall data during 2001 as an example,to improve the accuracy of the downscaling model.The main conclusions of the study are as follows:(1)There is a high relativity and consistency between TRMM 3B43_V7 precipitation products and the observation data form meteorological site.Besides,TRMM 3B43_V7 products did a great job in reflecting the distribution of the precipitation macroeconomic trends and seasonal variations in China Region,which are suitble for China region.The applicability of TRMM product increases with time scales enhanced,which means the applicability in year scale is relatively better than in seasonal scale,the applicability in seasonal scale is better than in monthly scale.Within 12 months,the applicability in spring,winter,autumn is better than summer.In the case of the influences of the terrain and the distribution of the sites,TRMM products performed quite differently.Therefore,it turns out,we need to revise the error.By comparing three correction methods,difference correction method,ration correction method and ratio coefficient correction method,we found the last method gets the highest precision.(2)The downscaling models can perform well,which considered geographical factors,topographic factors and vegetation factors.Slope correction factor is introduced as one factor in the influence of topography on precipitation,the simulation results can clearly describe windward,leeward slope of terrain features.Meanwhile,it was proved that the slope and aspect correction factor play an important role in correcting the error form TRMM products.For improving the spatial resolution of the TRMM product,we also take the advantage of the higher resolution from vegetation factor.(3)By the validation of the errors from four downscaling models,we found the model considering terrain correction factors and NDVI data,is superior to the other three models.With the validation of errors from verification sites,we found the correlation coefficient(R)of the downscaling model ? was higher than 0.96,the average relative error(MRE)and root mean square error(RMSE)in the month and year scale were decreasing.Comparing the spatialization of the sites and the results of the downscaling model IV,we comfirmed that the model IV has an advantage in simulating the spatial and temporal distribution of precipitation and local characteristics.(4)In 2001,for example,the introduction of cloud data to the model IV performed well in improving the accuracy of the original model.By comparing the validation results of the original model and the improved model,we found a uptrend in correlation coefficient(R),while a downtend in the average relative error(MRE)and root mean square error(RMSE).The improved downscaling model not only reflects the spatial distribution of rainfall,but also reflects the seasonal precipitation trends center.As a conclusion,the improved model provides a reference for other precipitation products downscaling,and it has some practical significance.