2.2.1 Pre-processing of snow data
It has been demonstrated that an image element can be classified as snow-covered when its snow cover factor (SCF) exceeds 50% (Tang et al. 2013). In this study, the MODIS day-by-day cloud-free snowpack area data was binarised using ENVI 5.3 software. Pixels with values lower than 50% were assigned a value of 0, representing the non-accumulating snow area, while pixels with values equal to 50% were assigned a value of 1, representing the accumulating snow area. This process divided the snowpack data into binary snowpack data of 0 and 1. Next, snow accumulation processes in spring and winter were selected for analysis based on data from the nearest meteorological stations, namely the Shiqu station and Xiaojin station, in the two basins. These processes were chosen to examine the stable snow formation process and final pattern. Specifically, in the Mamukao River basin, the snow accumulation process from 7 March to 20 March 2019 in spring and from 6 November to 22 November 2019 in winter was selected. In the Hanliu River basin, the snow accumulation process from 11 April to 28 April 2019 in spring and from 6 December to 17 December 2019 in winter was chosen (see Fig 2). Finally, the binarised snow data for the selected time periods in both basins were cropped using ArcGIS 10.8 software.