The Spatio-temporal Downscaling Of TRMM Precipitation Data In The Middle Reaches Of Yellow River

Spatially-distribution precipitation data is one of the most important parameter and have a foundation impact on hydrogen and water resource analyze,drought and flood forecast and ecological environment management.However,it is difficult to reflect the spatial variation and temporal distribution of rainfall effectively from traditional rainfall observation obtained from rain gauges for its limited quantities and random distributions.The technology of remote sensing has become an important source of spatially-distribution precipitation data for the characteristics of wide coverage and high spatial-temporal continuity.However,when applied to the local regions of small scale,it can't met the requirements of some research work for its coarse resolution.In this study,the spatial-temporal downscaling of TRMM precipitation data is conducted in the Hekouzhen to Tongguan of the Yellow River.The main research contents and conclusions are as follows:(1)The applicability of TRMM precipitation data in the Hekouzhen to Tongguan of Yellow River.The applicability of original TRMM data is tested using in-situ measurements form 23 weather stations in the view of scales ranging from month and quarter to year in study area.The results show that TRMM data are in great agreement with in-situ measurements and can reflect the spatial and temporal distribution and variation of precipitation effectively.The accuracy of TRMM precipitation data in the north is better than the south,and data accuracy of dry season is higher than the rain season.(2)The selection of predictive factors relevant to precipitation in the Hekouzhen to Tongguan of Yellow River.Digital Elevation Model(DEM),longitude,latitude,Normalized Difference Vegetation Index(NDVI)and Enhanced Vegetation Index(EVI)are chosen as preliminary factors to indicating rainfall in study area.In order to improve the prediction performance of rainfall heterogeneity,this study build the Moisture Migration Cost Distance(MMCD)on the basis of analyzing the characteristics of precipitation distribution in the study area,and their correlations with precipitation are analyzed at annual,seasonal and mensal scales.The results indicate a negative correlation between precipitation and MMCD and a positive correlation between precipitation and EVI.Both correlation are firm enough to serve as the basis of construction of the initial spatial downscaling model.(3)The construction of spatial downscaling model at different scales.A multiple linear regression model between TRMM precipitation,MMCD and EVI is established,by applying the downscaling methodology,the TRMM precipitation are downscaled to 1km×1km and the precipitation estimates are subsequently validated by using the observations obtained from 23 rain gauges in the study area.The results showed that: at annual scales,the downscaled results are in great agreement with in-situ measurements with the correlation coefficient R greater than 0.92.At seasonal scales,the downscaling results in high accuracy in spring,autumn and winter,but in summer the results are less satisfactory.At monthly scales,the precipitation in June and August are not good but in the rest months the simulation results are pretty well.(4)The construction of temporal downscaling model at daily scale based on soil moisture content.On the basis of monthly precipitation data that obtained from spatial downscaling and using the temporal distribution pattern revealed by the AMSR2 daily soil moisture content data,the monthly precipitation data at 1 km scale are further downscaled to daily scale.The validation with in-situ measurements indicate thatthere are obvious differences accuracy of simulation results in each site.In general,the downscaled result effectively captured the trends in inter-diurnal variability.