As a special geographical unit and the third pole in the world,the Qinghai-Tibet Plateau is the highest plateau in the world.Its surface heat has an important impact on surface and groundwater,surface energy and water balance,carbon exchange and ecosystem diversity.It also has a very sensitive response to climate change,and the freezing-thawing index not only provides indications for climate change,it is also a key parameter in cold area engineering,frozen soil research and other fields.The Qinghai-Tibet Plateau has a vast area and high altitude.There is basically no observation in the no-man's land of the plateau hinterland.Even in places where there are stations,the observation sites are very sparse and discontinuous,which makes it difficult to calculate the surface freezing and thawing index of the surface.This paper hopes to estimate the surface freezing and thawing index of the Qinghai-Tibet Plateau by using MODIS data with high temporal resolution.In this paper,three methods for estimating the daily mean land surface temperature are compared and analyzed.The difference of the fitting results obtained by inputting the different MODIS LST instantaneous value combinations is studied.The result of Sine-linear piecewise function method was used as a reference,while the 10 types of estimation results were evaluated based on the root mean square error and the judgment coefficient.In order to obtain as many daily mean land surface temperatures as possible on the Tibetan Plateau and reduce the impact of missing MODIS observations,the number of instantaneous LSTs on each pixel in the study area is counted and classified according to the LST on the pixel.The combination type is combined with the appropriate estimation method to obtain the daily mean land surface temperature,and the error correction is performed on the various estimation results by the quantile mapping,and the daily mean land surface temperature with better precision of the corrected data is obtained.Then,the time-series interpolation of the daily mean land surface temperature of the whole year is performed,and the interpolated daily mean land surface temperature is subjected to Kriging interpolation.The aim is to combine the limited LST instantaneous observation to obtain the daily mean land surface temperature with better precision as much as possible on the whole plateau surface,thus providing more accurate and complete data input for the calculation of the subsequent freezing-thawing index.Then,by combining the different environmental elements and the daily mean LST after wavelet decomposition as the input variables,the measured ground surface temperature is taken as the output.Through continuous ANFIS training,the optimal input variable combination is used as the predictor,and the estimation of daily mean ground surface temperature is realized.The estimated ground surface temperature data of the Qinghai-Tibet Plateau was verified from time and space by using the ground surface temperature measured by the site.Finally,the cumulative values of temperature and days greater than 0? and less than 0? were calculated by daily mean ground surface temperature of the Qinghai-Tibet Plateau,and the annual surface freezing index and thawing index were obtained,and the spatial variation of the two was analyzed.The research work and main conclusions of the thesis are as follows:(1)For the daily MODIS LST instantaneous observation data on the Tibetan Plateau,the percentage of pixels with complete four observations only accounts for 25.7%of the entire plateau,that is,the percentage of effective pixels under original method(Sin-Linear)is extremely low.By combining different MODIS instantaneous observation combinations with adaptive estimation methods,the daily mean LST of the Tibetan Plateau is estimated,which greatly improves the utilization of MODIS LST observations,and the average percentage of the daily pixels on the plateau can be increased to 53.5%.Through subsequent time and space interpolation,the average proportion of all pixels with daily LST on the Tibetan Plateau increased to 71%.And after the quantile error correction,the fitting error of the estimation results of various Cos-Sin methods can be effectively reduced,so that the accuracy of the daily mean LST after calibration is maintained at about 2?,and the determination coefficient R2 is above 0.8.Through the effective combination of multi-model estimation method and error correction and interpolation complement,the daily mean LST data with less value and better quality on the Qinghai-Tibet Plateau can be obtained,so that the daily mean LST as an intermediate product can be provided for models as more accurate parameter input.(2)After ANFIS training,the optimal input variable for predicting the daily mean ground surface temperature is the combination of daily mean LST after wavelet decomposition and the elevation and vegetation index of environmental factors.Snow depth and latitude and longitude do not affect the soil surface.The estimated ground surface temperature is compared with the measured temperature of the station.The root mean square error is within 2.5?,and the determination coefficient R2 reaches 0.95.Although the factors affecting the ground surface temperature are still uncertain,from the verification results,the ANFIS model based on wavelet decomposition can make up for this defect to a certain extent,making the simulation results more stable.(3)Based on the MODIS data of surface freezing and thawing index estimation method,can effectively get the whole surface of the qinghai-tibet plateau on the domain of surface freezing and melting index,the calculated results and the site observation of surface freezing/melting index correlation in 0.75 above,and the spatial distribution patterns of both with the whole of the qinghai-tibet plateau terrain characteristics can be better.The analysis shows that the spatial distribution of surface freezing/thawing index is negatively correlated,and the distribution of both is affected by the change of altitude.The annual freezing index increases with the increase of altitude,while the thawing index decreases with the increase of altitude. |