Improvement on the Effective Snow Cover Extraction Using Fusion Satellite Images Approach

Satellite remote sensing and optical and synthetic aperture radar (SAR) instruments can provide accurate observations of snow cover area (SCA) over inaccessible areas. Both instruments have the ability to complement each other. Notably, there have been limited attempts at incorporating their measurements. We have developed a new approach to create short-term maps of snowmelt areas through the application of data fusion techniques to Sentinel-2 fractional SCA and Sentinel-1 wet snow mask, facilitated by temperature threshold mask. The method is demonstrated in the 800 km² Latiyan basin in the mountainous region of Central Alborz (Tehran, Iran) over a period of 8 months (2018–2019). Snow indices threshold is applied to extract SCA in Sentinel-2, SCA (II). The typical overestimation of SCA by Sentinel-2 is corrected using a fusion of classification algorithms. Next, Sentinel-1, SCA (I) is presented as a mask to derive the fractional cover of wet snow. Results demonstrate that the Sentinel-1 and Sentinel-2 fusion approach, SCA (I, II), effectively improves the snow mapping model, but there is a shortcoming in the presented snow mapping method, especially in the cold months. Hydrologically, the snowmelt phase occurs based on temperature changes. So, daily temperature is introduced as an effective parameter to improve the model's accuracy. Results show that incorporating in situ and satellite data leads to developments in the model's accuracy. It is notable that the proposed method can be used in any mountainous areas where Sentinel-1 and Sentinel-2 data are routinely available. In sum, better runoff analysis, management of water resources, and flood risk depend on short-term snow mapping studies, as an important input in hydrological models.