Extraction And Evolution Of Inland Mudflats Based On GEE And Sentinel-2

Inland mudflats are an important part of wetlands,providing a variety of important ecological service functions,and are considered to be a backup resource for land.Due to the combined influence of natural and human factors,the ecological environment of inland mudflats is facing a major threat.Therefore,a timely and accurate grasp of the current status and evolutionary patterns of inland mudflat resources in China is a prerequisite for the sustainable management of inland mudflats.This paper focuses on the study of rapid extraction and evolution of inland mudflats.With 2014 to 2021 as the study years,Sentinel-2 and Landsat8 surface reflectance images covering the study area were selected respectively.Google Earth Engine(GEE)cloud platform was used for image processing,and mudflats extraction was carried out by combining maximum spectral index and maximum variance between categories method.At the same time,support vector machine was used to map the mudflat of East Dongting Lake,and the classification accuracy of the two methods was compared.This paper analyzes the evolution law of the East Dongting Lake mudflats from the Angle of the change of the number of the open air of the tidal mudflats,and quantitatively analyzes the relationship between the mudflats area of the East Dongting Lake and the height of the water level,realizes the function of predicting the mudflats area according to the height of the water level in a short time,and analyzes the causes of the tidal mudflats evolution.The main research conclusions are as follows:(1)The maximum potential area of the mudflat was determined by using the maximum inter-group variance method and the maximum spectral index synthesis method,and the influence of the surrounding paddy fields,ponds and aquaculture ponds on the extraction of mudflats was effectively removed.The remaining land types were water,vegetation and tidal mudflats,which laid the foundation for the accurate classification of the following support vector machine and the maximum inter-group variance method.(2)In the inland mudflats mapping data set based on support vector machine method,the addition of NDVI and NDWI can effectively improve the classification accuracy.The overall classification accuracy of data set "Band843"is 92.7%,and the Kappa coefficient is 0.89.By adding NDVI and NDWI to the data set "Band843",the total classification accuracyc increased to 94.3%and 94.0%,and the Kappa coefficient increased to 0.92 and 0.91,respectively.Both NDVI and NDWI were added to the data set "Band843",which had the most obvious improvement effect.The total classification accuracy increased to 95.3%,and the Kappa coefficient was increased to 0.93.(3)The classification accuracy of inland mudflats mapping based on the maximum inter-class variance method is higher than that of support vector machine,the highest total classification accuracy is 97.0%,and the highest Kappa coefficient is 0.96.The maximum inter-class variance method can greatly save manpower and time cost,and is more suitable for remote sensing image mapping of inland mudflats.It provides a scientific reference for other regions to map inland mudflats.(4)From 2014 to 2021,the maximum area of interannual mudflats and the number of days of complete inundated mudflats in the study area showed an increasing trend.The factors affecting the evolution of mudflats in the study area were divided into natural factors and human factors,and the main factor causing the evolution of mudflats in the study area was sand mining activities in the lake area.Natural factors and human factors jointly promoted the evolution of mudflats in East Dongting Lake.