Bathymetry from shallow coastal environment using Landsat 8 (Case Study: Southeastern part of the Caspian Sea)

In order to study in shallow coastal areas, free Landsat 8 image with relatively high radiometric resolution and the presence of two bands, coastal blue and blue is suitable. In this study, in addition to Landsat satellite imagery, hydrographic data has also been used. In order to increase the accuracy and reduce the error, tried as much as possible the passing time of the Landsat 8 sensor close to the acquisition terrain date. The purpose of this study is bathymetry of the southeastern part of the Caspian Sea using the PCA algorithm on the pre-processed visible region. In this study, the FLAASH and Dark Object Subtract (DOS) atmospheric corrections were applied separately to visible bands. The obtained depth results by applying PCA transformation to these two types atmospheric correction is investigated. PCA algorithm is implemented in four different forms. Statistical parameters R^2, RMSE, and NRMSE are calculated between obtained data by PCA and hydrographic data in two types of atmospheric correction. The results show that in both of atmospheric correction, the accuracy of estimated depth by PCA is increased when four or three PCA components are introduced as algorithm inputs compared with only two or one PCA components are used. As well as, the use of four components, the accuracy of DOS in bathymetry with values of R2=0.91, RMSE=0.3,and NRMSE=0.05 has shown a better result in comparison to FLAASH correction's values R2=0.87, RMSE=0.38, and NRMSE=0.06.