Comparison studies of two downscaled passive microwave snow depth products over the Qinghai-Xizang Plateau based on MODIS fractional snow cover dataset
Snow cover is one of the most active factors in the cryosphere,directly affecting the energy exchange between the atmosphere and the Earth.Snow depth(SD)is an important attribute to describe the temporal and spatial variation of snow cover,and is an important input parameter for models such as basin water balance,and the simulation of snow runoff.Passive microwave remote sensing utilizes the strong correlation between the dif-ferences in snow scattering characteristics and snow depth at different frequency brightness temperatures(BT)for snow depth inversion.It is widely used to study the temporal and spatial variations of snow depth at global or regional scales.However,due to the strong spatial and temporal heterogeneity of snow cover in mountainous ar-eas,the spatial distribution is uneven,the temporal variation is different,so these passive microwave remote sensing snow depth products with coarse spatial resolution are greatly limited.Especially in the mountainous ar-eas of the Qinghai-Xizang Plateau,due to the spatial discontinuity and heterogeneity,these data are insufficient to represent the snow conditions at a regional scale.This study is based on MODIS fractional snow cover datas-et,and uses empirical fusion rules and snow decay curves to perform spatial downscaling on the two sets of snow depth products of"the long-term series of daily snow depth dataset in China".The first set of snow depth prod-uct,referred to as Che_SSMI/S products,is inverted from SMMR,SSMI,and SSMI/S,while the second set,Che_AMSR2 products,is inverted from AMSR-2 brightness temperature.Ultimately,the 500 m downscaled snow depth data(Che_SSMI/S_NSD and Che_AMSR2_NSD)of the Qinghai-Xizang Plateau were obtained.There are significant differences between the two sets of snow depth data after downscaling.Utilizing 6 scenes of Landsat-8 images,we conducted a comparative analysis of the downscaled snow depth data from the two sets,it was found that both sets of downscaled data had a high degree of agreement with the spatial distribution of snow cover in Landsat-8 images.The Che_AMSR2_NSD product was compared with the snow depth data of 29 meteorological stations and found that the data were closer to the measured snow depth data,the correlation coefficient(R)was 0.72 and the root mean square error(RMSE)was 3.21 cm,while the correlation coeffi-cient between Che_SSMI/S_NSD and the measured snow depth is 0.67,the RMSE is 4.44 cm.It is speculated that the reason for this discrepancy may be due to the difference in the accuracy of the two sets of original snow depth products with brightness temperature data from different sensors.In addition,the experimental results show that the downscaled accuracy of passive microwave snow depth products is also affected by factors such as snow depth and snow cover period.The results show that the accuracy of downscaled snow depth is different at different snow depth and different snow cover periods.Specifically,when the snow cover is accumulation peri-od and stable period,the downscaled accuracy of the two sets of snow depth products is higher,while when they are in the snow ablation period,the accuracy gradually decreases.The highest downscaled accuracy for both sets of snow depth products occurs,when the snow depth is less than 10 cm.Conversely,the accuracy of the snow depth is significantly reduced when the snow depth exceeds 30 cm,which may be due to the saturation issues in passive microwave snow depth inversion.Through the comparative evaluation of the two sets of downscaled snow depth products,it is helpful to understand the temporal and spatial distribution of snow depth more compre-hensively over the Qinghai-Xizang Plateau,and provide snow depth data support for its application.
NETL
NSTL
万方数据
