A Study On Drought Monitoring Model By Multi-source Remote Sensing Data In Qinghai Province

The precipitation data is the basic data for meteorological hydrology analysis,whose spatial scale and sequence length are of importance for research.At this stage,the data of areal precipitation is mainly obtained by interpolation method.For the complex terrain,vast area and the heterogeneous distribution of meteorological stations in Qinghai province,there will be large error with interpolation method during the process of precipitation analysis in large area.However,the tropical rainfall measuring mission(TRMM)based on radar rainfall measurement could effectively improve the above problem.Firstly,the adaptability of TRMM 3B43 data in Qinghai province was analyzed in this paper.Secondly,based on multisource remote sensing data,the factors of resulting in drought were obtained and the drought monitoring model had been established.The main research contents are shown as follows.(1)The analysis of regional adaptability.With the measured precipitation data of 50 meteorological stations in researched area,the adaptability of TRMM 3B43 data and precipitation process of single station were analyzed with the statistical approach of correlation coefficient and relative deviation on the year,season and month scales,respectively.It was shown that 1)On the whole,the TRMM data has a high precision in Qinghai province with the month scale correlation coefficient of 0.9344 and year scale correlation coefficient of 0.9111.The correlation coefficient of 41 meteorological stations is more than 0.9 and the precipitation process of remote sensing data and measured data has good consistency.2)There exists difference for the fitting precision of precipitation data on time scale.The fitting result in autumn is the best,then the one in spring and summer,and worst the one in winter.3)There exists difference for the fitting precision of precipitation data on spatial scale.The fitting result in the monitoring station of Mangy,Xiaozaohuo and Lenghu of Haixi state is the worst with the correlation coefficient of 0.73 and the deviation of 50%.The fitting result of Wudaoliang station is the best with the correlation coefficient of 0.98.The deviation of Gangcha station is the smallest with the value of 0.08%.(2)Statistical downscaling study.It was found that there exists a good positive association between the remote sensing precipitation data and the normalized difference vegetation index(NDVI).Considering the effect of season and terrain on precipitation,the optimal relationship among TRMM data,geographical factor(namely,longitude and latitude),elevation factor(DEM)and NDVI had been obtained with artificial intelligence software as the core of statistical downscaling model.By downscaling the TRMM 3B43 data from low-resolution(0.05°×0.05°)to high-resolution(0.05°×0.05°and 1km×1km),the fitting equations could all pass the significance testing with the R2 of more than 0.6 except that of January and December.(3)Establishment the TRMM-Z index and TCI index.With the precipitation data month by month of TRMM from 2000 to 2014 and the Z index method of single monitoring,the TRMM-Z index for monitoring of precipitation deficiency had been established.With the above TRMM-Z index,drought monitoring model had been established.Comparing with the SPI value calculating from the measured precipitation data,it was found that the TRMM-Z index has some representation ability of drought.However,there exists some uncertainty of the above calculating results and some phenomena of weakening the drought level on account of the shortness of remote sensing precipitation sequence.Based on the surface temperature data of Qinghai province from MODIS remote sensing information source and its relationship with the maximum and minimum values in the same term,the temperature condition index(TCI)was obtained as the indicator of evaluating drought.(4)By calculating the CI value and Z index and extracting the TCI values of 50 meteorological stations,relevance model was established with the Z index and TCI as the independent variable and CI as the dependent variable to get the parameters of multi-source-remote-sensing integrated drought monitoring model,whose drought indexes were obtained by calculating the TRMM-Z and TCI in the corresponding area.It was found that the correlation coefficient R was 0.5844 in the best fitting case,which could pass the significance testing.In addition,by compared with the instance analysis of typical year,it was proved that integrated drought monitoring model could describe the drought status and further verification was required for this model by comparing other data source.