| Snow is an important type of surface cover,which has the functions of collecting,storing,releasing water,and replenishing rivers.It is an important freshwater resource in the arid region of Northwest China.Compared with other ground objects,snow has high reflection characteristics,so optical remote sensing has become an important means of snow change monitoring,and the reconstruction of snow information under clouds and cloud shadows has become the key.Because the Fmask cloud and cloud shadow detection algorithm has the characteristics of high accuracy and fast calculation speed,it has been used by the US Geological Survey for Landsat data business processing.In this research,the Fmask algorithm is introduced into the GEE(Google Earth Engine)cloud computing platform,combined with the ESA Sen2 Cor clear sky snow detection method,to realize the cloud,cloud shadow and snow detection of the Sentinel-2 L1 C cloud cover image.Then improve the SNOWL(Snow Line)algorithm based on the unstable snow area in the study area to reconstruct the snow under the cloud and the cloud shadow.This paper takes the Babao River Basin in the upper reaches of the Heihe River Basin as the research area,selects Sentinel-2 A/B products from November 2016 to March 2021,and uses the improved cloud and cloud shadow reconstruction algorithm to obtain the snow under the clouds and clouds shadow.After that,parameters such as snow cover rate and snow cover days were extracted,and the accuracy of the snow extraction results was verified with GF-2 satellite data,and then the temporal and spatial distribution characteristics of snow cover during the158 days of satellite transit were analyzed.Among them,the 27-day images are affected by clouds of varying degrees,and the remaining 131-day images are clear sky images.The research conclusions are as follows:(1)Compared with the original SNOWL algorithm,the overall accuracy is improved from 66.05% to 84.26%.The study found that the original SNOWL algorithm has a high multi-measurement error,and it is easy to overestimate the snow in the windward and sunny slopes of mountainous areas where the snow melts quickly.By calculating the number of snow days,obtaining the unstable snow area and eliminating this part of the area,the multi-measurement error was reduced from43.81% to 9.22%,which further improved the overall accuracy of the snow reconstruction algorithm.(2)In the past five years,there are significant differences in the snow cover ratio(SCR)of the hydrological years in the Babao River Basin,showing an increasing trend year by year.Among them,the snow cover rate and the number of snowfalls are the most from September 2019 to August 2020,and the average snow cover rate from October to March of the following year can reach 51.67%.The time period of concentrated snowfall is also different in each hydrological year.From September 2017 to August 2018,the snowfall is concentrated in late January to late May;from September 2018 to August 2019,the snowfall is concentrated in midNovember to mid-April;from September 2019 to August 2020 In September,the snowfall is concentrated from mid-October to mid-April,and the start date of snow cover gradually advances.(3)The characteristics of the snowfall season in the study area are not obvious.In 2019,the area's snow coverage rate in October reached 69.21%,but in 2017,the first large-scale snow cover(SCR: 88.49%)appeared in February.There are significant differences in snowfall between hydrological years.From September2016 to August 2017 and from September 2020 to August 2021,only two months of snow cover exceeded 40%,while from September 2019 to 2020 In August,the snow cover rate exceeded 40% for 7 months.(4)From the perspective of the spatial distribution of snow days,about half of the study area is unstable snow area,and most of these areas are valley areas with lower elevations.The areas above 3977 m above sea level are covered by snow at least half of the time each year,and these areas are mainly distributed in the north and southeast of the study area. |
PDF Full Text Request