Research On Precipitable Water Vapor Based On GPS And MODIS Data

Precipitable water vapor(PWV)is of great significance to precipitation and climate,atmospheric correction,InSAR atmospheric effect removal,and severe weather forecast.It is also an important branch of the current GNSS meteorology field.At present,there are many kinds of methods to detect PWV.However,all the methods have its own advantages and limitations.Therefore,it is necessary to use varieties of methods to achieve high temporal resolution,high spatial resolution and long-term stable PWV severed to the deeper reach of PWV.The main work and conclusions are as follows:1)The current research methods and theories to obtain PWV at home and abroad are introduced in detail.The advantages and disadvantages of several methods are compared and analyzed.Put forward an idea that the accuracy and temporal resolution and high spatial resolution of PWV can be improved based on GPS data and MODIS data.2)Research on GPS-PWV:including understanding GPS-PWV obtain process and the algorithms of GAMIT software.Comparing the accuracy between GPS date provided by suominet GPS network and sounding data.It is concluded that the PWV calculated by GPS-PWV and sounding data have a good consistency,the accuracy of GPS in the United States is about 1.2mm,which verifies the GPS-PWV's high precision,consistent with the conclusions put forward by domestic and foreign scholars.3)The method of determining the weighting factor in MODIS three-channel weighted average algorithm is improved by using the combined forecasting model of square error and minimum method of prediction error based on the theories of MODIS-PWV.A set of optimal weight coefficients ?(0.31,0.17,0.52)are determined in the study area.The accuracy of the improved weighted average three-channel ratio method is 33.5%higher than that of the two-channel ratio method,21.7%higher than that of the three-channel ratio method.4)The correlation between GPS-PWV and MODIS-PWV is high.Thus,a linear regression model was established.The RMSE of the linear regression model is ±1.64,and the accuracy of the linear regression model is 49.1%higher than that of the three-channel MODIS model.Therefore,the linear regression model can be used to correct the MODIS-PWV,especially in areas where there is no GPS station.5)A neural network model based on GPS/MODIS data is established.The RMSE of the GPS/MODIS neural network model is ± 1.07.Compared with the three-channel MODIS model,the accuracy of the GPS/MODIS neural network model is improved by 66.7%.