Estimation of Total Suspended Sediment and Chlorophyll-A Concentration from Landsat 8-Oli: The Effect of Atmospher and Retrieval Algorithm

Abstract

TSS and Chl-a are globally known as a key parameter for regular seawater monitoring. Considering the high temporal and spatial variations of water constituent, the remote sensing technique is an efficient and accurate method for extracting water physical parameters. The accuracy of estimated data derived from remote sensing depends on an accurate atmospheric correction algorithm and physical parameter retrieval algorithms. In this research, the accuracy of the atmospherically corrected product of USGS as well as the developed algorithms for estimating TSS and Chl-a concentration using Landsat 8-OLI data were evaluated. The data used in this study was collected from Poteran’s waters (9 stations) on April 22, 2015 and Gili Iyang’s waters (6 stations) on October 15, 2015. The low correlation between in situ and Landsat Rrs(λ) (R2 = 0.106) indicated that atmospheric correction algorithm performed by USGS has a limitation. The TSS concentration retrieval algorithm produced an acceptable accuracy both over Poteran’s waters (RE of 4.60% and R2 of 0.628) and over Gili Iyang’s waters (RE of 14.82% and R2 of 0.345). Although the R2 lower than 0.5, the relative error was more accurate than the minimum requirement of 30%. Whereas, the Chl-a concentration retrieval algorithm produced an acceptable result over Poteran’s waters (RE of 13.87% and R2 of 0.416) but failed over Gili Iyang’s waters (RE of 99.14% and R2 of 0.090). The low correlation between measured and estimated TSS or Chl-a concentrations were caused not only by the performance of developed TSS and Chl-a estimation retrieval algorithms but also the accuracy of atmospherically corrected reflectance of Landsat product. Keywords?remote