Applicability Evaluation Of Atmospheric Correction Method For Poyang Lake

The main goal of lake water color remote sensing is to obtain water out-of-water reflectance through remote sensing images,but only less than 10% of the total radiation received by sensors are water signals.So High precision atmospheric correction is needed to accurately obtain the water body’s out of water reflectance.The atmospheric correction method for class I water body is mature,but for class II water body with complex optical characteristics,it’s atmospheric correction method needs to be improved.At present,there is not a set of general atmospheric correction methods for class II water body,and the methods applicable to different regions and different uses are not the same.Therefore,it is necessary to study the adaptability of class II water atmospheric correction methods applicable to lake water body region.In this paper,Landsat 8 OLI images of Poyang Lake at the end of August 2015 and Sentinel2 MSI images at the end of October 2018 were used as data sources.Combined with the spectral response function,the measured spectral data of corresponding bands were fitted.By comparing and analyzing the correction accuracy of five common atmospheric correction methods(6S,ACOLITE,i COR,FLAASH,QUAC)and one proprietary atmospheric correction method(La SRC and Sen2cor),the applicability of each method in Poyang Lake was evaluated.It can provide data support for subsequent inversion of water quality or water color parameters.To meet the needs of atmospheric correction algorithm for Poyang Lake in China,the following two aspects of work have been completed:1.The Sentinel 2 MSI image and Landsat 8 OLI image of Poyang Lake were corrected by using different atmospheric correction methods,and the water reflectance of each band was obtained.Based on the measured remote sensing reflectance,the error calculation and comparative analysis of the correction results were carried out by comparing the visual effect,the overall accuracy and the single band accuracy.The results show that Sen2 cor method has the highest global correction accuracy for Sentinel 2 MSI images,and i COR method has the highest global correction accuracy for Landsat 8 OLI images.For Sentinel 2 MSI images,each atmospheric correction method has the best performance at 665 nm and 705 nm,Sen2cor has the highest accuracy,followed by ACOLITE and 6S methods,and has poor performance at 842 nm and 865 nm,with the average relative error of more than 50%.In Landsat 8 OLI image,other methods except QUAC algorithm have higher accuracy at 563 nm and 655 nm than the other three bands,with average relative error between 10% and 20%.i COR method has the best performance in these two bands,followed by FLAASH method and La SRC method2.The NDWI index was used to extract the water area of the study area,and then the sediment index(SI)was calculated to classify the suspended sediment concentration of the water after atmospheric correction,which can be divided into four categories: High suspended sediment in the water,suspended sediment,low water suspended sediment water and clean water,and then according to the analysis of the measured water element parameters of the site,the sampling points are divided into four categories,the atmospheric correction results compared with the measured results of suspended sediment,it is concluded that the applicability of the two kinds of different area of the image of atmospheric correction methods.The results show that FLAASH method is the most accurate method for Sentinel 2 MSI and Landsat 8 OLI images in the high suspended sediment concentration area.In the medium suspended sediment concentration region,Sen2 cor method is the most suitable method for Sentinel 2 MSI image,and i COR method for Landsat 8 OLI image.In the region of low suspended sediment concentration,i COR method was the best for Sentinel 2 MSI image correction,while ACOLITE method was the best for Landsat 8 OLI image correction.For unsampled clean water,according to the existing results of domestic and foreign researchers,NIR method has the best correction effect.This article is based on two kinds of commonly used image and application effect of the atmospheric correction algorithm and application effects of the two aspects of suspended sediment concentration classification research,for the domestic inland lakes atmospheric correction to choose suitable algorithm,provides the reference for refining after different degrees of suspended sediment concentration of the atmospheric correction algorithm comparison and provides reference for the optimization of the atmospheric correction algorithm.In the end,the paper summarizes the completed work and puts forward some prospects for the future research of atmospheric correction algorithm for inland lakes,including the acquisition of more measured water spectral information in inland lakes in China,and the division of optical depth and shallow water in the future.