Reconstruction Of Monthly Precipitation Data In Tibetan Plateau

The Tibetan Plateau is of great significance to the study of global climate.Precipitation data are also important to hydrology,meteorology and ecology.With the change of research content and scale,the demand for historical precipitation data with high spatial and temporal resolution is more and more urgent.This paper selects six factors that are the normalized vegetation index(NDVI),elevation,slope,longitude,latitude and precipitation in order to obtain the historical precipitation data with high spatial and temporal resolution of the Tibetan Plateau from 1982 to 1997.This research is based on the precipitation data of TRMM 3B43,AVHRR NDVI and SRTM DEM.The factors were input into the random forest model.The historical precipitation model was constructed to obtain the resolution of 0.0833 ° annual precipitation.And then calculate the monthly precipitation according to the proportional coefficient.In order to improve the accuracy,the monthly precipitation data is corrected using the site data.Use the Kriging interpolation to correct the historical monthly precipitation.Finally,the paper selects decision factor R2,root mean square error RMSE,mean absolute error MAE and deviation Bias to evaluate monthly precipitation accuracy.At the same time,compare GPM data and TRMM 3B43 data from April 2014 to March 2015.The paper provides the high space-time resolution precipitation data.Historical precipitation data is helpful for the research of hydrology,land surface process.The conclusions of this paper are as follows:(1)The results show that the method can obtain the historical precipitation data with high spatial and temporal resolution.The accuracy of annual precipitation is that R2 is around 0.6,RMSE is about 200 mm,MAE is about 150 mm,Bias is about 0.3.Only a few years precision fluctuated.The accuracy of monthly precipitation is as follows.The coefficient of determination R2 is between 0.4 and 0.9,with an average value of 0.6767.The results show better in summer and worse in winter.The root mean square error and average absolute error is below 50mm.RMSE mean is 22.66mm.MAE mean is 15.97mm.Deviation Bias is between 0 and 0.1.(2)In space,the annual precipitation over the Tibetan plateau gradually increases from northwest to southeast.In the northwest,the precipitation is less.The precipitation is below 200mm;The precipitation in the central reg:ion is between 200-600mm;the precipitation is more in the southeast.The precipitation is above 600mm.In the southeastern margin,the precipitation is more than 1300mm.(3)Monthly precipitation in the Tibetan plateau has seasonal changes.In spring,precipitation is between 100-200mm in most regions and the plateau rainy season begins to move from the southeast to the northwest.In summer,Precipitation increases significantly.Precipitation is mainly in the range of 300-600mm.In autumn,precipitation decreases significantly.In winter,there is almost no precipitation in most parts of Tibetan plateau.(4)The spatial distribution of GPM data and TRMM 3343 data is consistent.The observations of TRMM 3B43 data in regions with lo,wer precipitation are higher than GPM.In regions with higher precipitation,the observations of GPM are higher than TRMM 3B43.In terms of observation accuracy,the observational accuracy of monthly precipitation GPM data is better than that of TRMM 3B43 data.