4.4 Remote sensing image uncertainty
This study utilizes the MODIS day-by-day cloud-free snow cover dataset to mitigate the impact of cloud cover to some extent. However, the spatial resolution of the MODIS data is 500m, which may not be sufficient to detect small-scale areas of snow accumulation in the two watersheds examined in this study. Furthermore, the dataset employs the NDSI normalized snow index with a threshold of 0.4 to extract the snowpack, which has demonstrated reliable accuracy in North America (Hall et al. 1995,Hao et al. 2008a). However, numerous studies have demonstrated that employing an NDSI threshold of 0.4 in China leads to a substantial underestimation of ground snowpack (Hao et al. 2008b,Zhang et al. 2019). Meanwhile, the western Sichuan Plateau is situated in the eastern part of the Tibetan Plateau, characterized by extensive forest coverage. Zhao et al. (year) demonstrated that the accuracy of NDSI in extracting the snowpack in forested areas was inadequate (Zhao et al. 2016). Combined with the aforementioned reasons, the utilization of MODIS day-by-day cloud-free snowpack area data in this study may result in an underestimation of the snowpack area. Consequently, this underestimation may lead to inaccuracies in the classification of snowpack types in certain regions. Therefore, future research conducted in the area should prioritize addressing the concerns of data accuracy and its applicability