Research On Snow Cover Monitoring In Northeast China Based On Multi-source Remote Sensing Data

Snow contributes to the balance of radiation energy of the Earth and affects various climatic and hydrological processes.Conventional ground station monitoring can not accurately obtain large-scale monitoring results.Whether the monitoring range of ground station is representative directly affects the monitoring accuracy.With the development of satellite technology,the multi-band sensors in optical remote sensing can achieve large-scale observation and provide accurate snow cover information.Cloud interference and forest occlusion are important problems in obtaining accurate snow cover information in optical remote sensing.This paper mainly combines the characteristics of multi-source remote sensing data with high temporal resolution,high spatial resolution,abundant band information,and the forest transmittance model to reduce the impact of cloud and the interference of forest canopy occlusion,so as to produce more accurate and effective snow cover products.The specific research contents and main innovations are as follows:(1)Snow cover monitoring in Northeast China based on FY-2G geostationary satellite remote sensing data.The main pretreatment methods of FY are as follows: radiation calibration,geometric correction and reflectivity correction.Then,cloud pixels are identified by spectral threshold and neighborhood similarity calculation.Finally,according to the characteristics of snow in different spectral bands,the corresponding threshold is set to extract the fraction of snow cover information.The MODIS daily snow products were used to verify the accuracy of the result.The experiments show: MODIS can accurately monitor snow cover under cloudless conditions;The multi-temporal FY-2G data can get more accurate snow cover information under cloud conditions.(2)Snow cover monitoring in Northeast China based on FY-3B orbit satellite satellite remote sensing data.The land types in Northeast China can be divided into forest areas and non-forest areas.For non-forest areas,Landsat8 OLI data and SNOWMAP algorithm are used to obtain "real" snow cover information.Then,the relationship between snow index and the fraction of snow cover is established by linear regression.Finally,the snow cover products under cloudless conditions are obtained by selecting the regression equation of snow index with higher precision.For forest areas,SCAMOD model is used to calculate forest transmittance,and the fraction of snow cover under forest is expressed as FY-3B visible reflectance and forest transmittance function.The method can provide more accurate products of snow cover under forests,and the snow cover is validated with GF-2 data.(3)Research on the fraction of snow cover algorithm based on FY-2G and FY-3B remote sensing data.The FY-2G data has high temporal resolution.It can provide multi-temporal data for the research area,which can effectively reduce the interference of cloud in optical sensors.At the same time,The FY-3B data has higher spatial resolution and more abundant spectral information,which can provide more accurate snow cover products.Combining the advantages of two kinds of satellite data,this paper proposes a fusion algorithm of snow cover products based on multi-source remote sensing data,which can generate high precision products of daily snow cover.The hidden Markov random field model is used to predict the areas where there are still clouds or data gaps.Finally,Cloudless snow cover products were obtained in Northeast China.In this paper,the fraction of snow cover products are obtained by fusing the FY-2G and FY-3B data.Combined with the data fusion algorithm and the SCAMOD model,the interference of the cloud and the effect of the forest canopy occlusion can be reduced.The experiment provides an effective and accurate monitoring method for snow cover monitoring in Northeast China.