Research On Lake Ice Phenology Remote Sensing Monitoring And Data Processing System

“Lake Ice Phenology” is a term used to describe the seasonal cycle of lake-ice cover,including the freezing period,thawing period and ice-cover duration.It is one of the key variables of the cryosphere.The development of remote sensing provides new methods for the extraction of lake ice phenology.However,the current lake ice phenology datasets for small lakes are still lacking,and the terabyte data volume poses new challenges for the extraction of lake ice phenology.Currently,optical remote sensing data and passive microwave remote sensing data are mainly used to monitor the lake ice phenology.Both monitoring methods have advantages and disadvantages.Optical remote sensing data will be affected by clouds and polar nights.Although the resolution is high enough to monitor the lake ice phenology of a large number of lakes,the data began in 2000 and it is difficult to trace the freeze-thaw of lakes in historical periods.Passive microwave remote sensing data has a long duration and is not affected by cloud and rain,but its resolution is low,making it difficult to monitor the small lakes.This paper mainly verifies the feasibility of remote sensing data,that is,MODIS optical remote sensing data and passive microwave remote sensing data,to monitor lake freezing and thawing and extract lake ice phenology.Based on the MODIS data to identify lake ice under clear sky conditions,a lake ice discrimination rule under cloud cover was formed to solve the influence of MODIS optical remote sensing data by clouds,and the daily lake ice extent and coverage dataset of the Tibetan Plateau were established.Based on the principle of passive microwave satellite imaging,a mixed pixel decomposition model was established to monitor the freezing and thawing of small and medium-sized lakes,and 178 lake ice bright temperature-lake ice phenology sample sets in the northern hemisphere were prepared.Based on the dataset of lake ice coverage in the Tibet Plateau,a convolutional neural network is used to classify whether lakes can extract lake ice phenology,which lays a foundation for the rapid extraction of lake ice phenology by the threshold method.Based on the passive microwave lake brightness temperature-lake ice phenology sample set,the support vector regression method was used to extract the lake ice phenology.The freezing and complete thawing of the two lake ice phenology parameters all passed the validity test,indicating the use of support vector machine regression.The method can effectively extract lake ice phenology information.Replacing manual classification and visual interpretation methods with two machine learning methods can improve the efficiency of obtaining lake ice phenology.On this basis,in order to cope with the high-volume remote sensing data,the lake ice phenology data processing system was studied.The system is mainly composed of input data acquisition and update module,lake ice phenology extraction module based on passive microwave remote sensing data,and lake ice phenology extraction module based on MODIS optical remote sensing data.The modular design of the three parts can achieve the goal of streamline extraction of lake ice phenology.