Bathymetry of shallow coastal environment using multi-spectral passive data under rapid sea–level change

Abstract

Remote sensing imagery is by far one of the top-tier apparatuses employed in the estimation of water depth in particularly shallow coastal regions. Contra to traditional models, remote-sensing-directed bathymetry is considered a highly efficient approach, as it entails a significant increase in performance, all the while reducing costs. Fluctuations in the level of the Caspian Sea (CS) are one of the most important characteristics of the CS and Gomishan Lagoon, which is near the CS and is directly affected by its level. The main aim of this work is to first obtain the water depth up to ~7m based on field and Landsat data in 2016 in the southeast of the CS using multi-variable regression. Therefore, we examined the linear model between the visible reflectance and the field data. Finally, the depth was calculated from different band combinations (BCs). In order to evaluate the accuracy of the used model, the depth of Gomishan Lagoon is estimated in 2000. The results show that the BC of 234 (blue, green, red) indicate the best BC in depth estimation. According to the linear model, the depth accuracy was obtained R2 = 0.94, MRE = 0.09 and RMSE = 0.35. The predicted depth for Gomishan Lagoon was in accordance with previous findings as well.