Remote Sensing Retrieval And Verification Of Snow Parameters On The Tibetan Plateau

Tibetan Plateau,located in the southwest of China,is the highest average altitude in the world.Tibetan Plateau not only forms a unique plateau climate,but also plays an important role in strengthening the monsoon circulation and has a great impact on China's climate.Tibetan Plateau is considered to be a sensitive area of climate change because of its predominance to climate change.As an indicator of climate change,snow plays a more important role in the Tibetan Plateau.Tibetan Plateau is one of the three major snow distribution areas in China.The snow in the area parameters accurate inversion can lay the foundation for snow resource assessment,water resources management and global climate change in China.Therefore,this study taking the snow parameters(snow coverage,snow grain size,snow pollution)of Tibetan Plateau as the research object,based on different satellite remote sensing data and field measured data,combined with the Asymptotic Radiative Transfer model,the snow parameters of Tibetan Plateau are inversed more accurately.The methods are as follows:(1)The Soumi-NPP snow-covered product is used as the research object.the meteorological station data and higher resolution data of Landsat-8 OLI were used to evaluate the accuracy of the Soumi-NPP and MODIS snow cover product,and used MODIS snow-covered products for comparative analysis.(2)Based on Asymptotic Radiative Transfer model(ART),a snow reflectivity spectral library with snow grain size of 10~1200 ?m was constructed.The mixed pixel decomposition of MODIS reflectivity products is carried out to generate Fractional Snow-Cover products(ART_FSC)in accordance with the Tibetan Plateau.UAV Snow cover data(UAV_FSC)and Landsat-8 OLI Snow cover data(OLI_FSC)are used as true values to compare the mapping algorithm with MODIS Global Fractional Snow-Cover Product(MOD_FSC).(3)Three bands and based on the ART model Snow Grain Size and Pollutionalgorithm is used to retrieve the black carbon concentration and snow grain size of snow surface in Tibetan Plateau,and the accuracy of the inversion results is evaluated according to the measured values in the field.The main conclusions are as follows:(1)The total accuracy of the three snow cover products,NPP,MOD10A1 and MYD10A1,are all higher when using the meteorological station to verify the data,but the errors of the three snow cover products are all large,and the omission error of MYD is the largest,which is 64.2%.The snow classification accuracy of the three snow cover products is low when the snow depth is lower than that of 5cm,and the snow classification accuracy of NPP snow range products is the highest which is82.3% when the snow depth is greater than 5 cm,and the accuracy of MOD10A1 and MYD10A1 were 77.1% and 69.4% respectively.The Kappa coefficient of Soumi-NPP snow cover product was the highest with the mean value of 0.707,which was highly consistent with Landsat-8 OLI data.The Kappa coefficients of MOD10A1 and MYD10A1 are 0.476 and 0.557,respectively.The Kappa coefficients of the two MODIS snow products in the moderate consistency with Landsat-8 OLI data.The Kappa coefficients of Soumi-NPP are mostly above 0.6,and the accuracy is relatively stable,while the Kappa coefficient of the MODIS snow cover fluctuates greatly,and the accuracy stability is poor.(2)The snow shape parameters have a great influence on the accuracy of ART_FSC model.When the snow shape parameter is 4.22,the accuracy of ART_FSC is the highest,and the root mean square error and correlation coefficient are the best.When UAV_FSC is used to validate the accuracy of ART_FSC,and the determination coefficient and root mean square error are 0.811 and 0.190,respectively.When OLI_FSC is used to validate the accuracy,the accuracy of ART_FSC is better than that of MOD_FSC,and the correlation coefficient and root mean square error are the best,and the accuracy of ART_FSC is the highest when the underlying surface is grassland,followed by bare land,other land features and forest.(3)The SGSP algorithm can simultaneously retrieve snow black carbon concentration and snow grain size on the Tibetan Plateau,and most of the retrieval results are within a valid range.After verification of the measured data,the retrieval results have better accuracy.The decision coefficient between the retrieval black carbon concentration data and the measured black carbon concentration data is 0.73,and the root mean square error and average absolute error are 420.9 ng·g-1and 310.1ng·g-1.The decision coefficient between the retrieval snow grain size and themeasured snow grain size is 0.72,and the root mean square error and average absolute error are 166.9 ?m and 134.4 ?m,respectively.In general,the SGSP algorithm has higher accuracy in inversion results,and it has a better application range on the Tibetan Plateau.(4)To sum up,compared with MODIS snow cover product,the accuracy of Soumi-NPP snow product has been greatly improved,which provides a better choice for accurately monitoring the snow-covered area of Tibetan Plateau.The spatial heterogeneity of snow in Tibetan Plateau is strong.ART_FSC algorithm is an effective way to improve the accuracy of remote sensing monitoring of snow cover in Tibetan Plateau.The SGSP algorithm has good adaptability to the Tibetan Plateau and can reflect the black carbon concentration and snow grain size of the snow surface in the Tibetan Plateau.