A nonparametric fuzzy shoreline extraction approach from Sentinel-1A by integration of RASAT pan-sharpened imagery

Abstract

Coastal zones are among the most important features of the earth as they have social and economic importance for the environment. Thus, they should be frequently monitored for management issues, for example, shoreline erosion, and any essential precautions for safe management taken. Shoreline extraction is an important tool to determine coastal changes. Modern technologies such as remote sensing give an opportunity to gain up-to-date, accurate, and reliable information with fast data acquisition for coastal regions. In this study, two different coastal areas, one in the northern part of Turkey (Terkos/Istanbul) and the second from the southern part of the country (Kemer/Antalya), were studied. The main objective of this study is to extract shorelines from Sentinel-1A radio detection and ranging satellite data using a non-parametric fuzzy approach, by exploiting Turkish multispectral RASAT satellite images. For this purpose, a whale optimization algorithm was initially employed to create land and water classes from RASAT images. The created classes were used as inputs to determine fuzzy membership values for non-parametric shoreline extractions from Sentinel-1A images. The results are evaluated in comparison with manually digitized reference shorelines by calculating the perpendicular distances between the digitized and remotely sensed shorelines. The calculated average distances from the RASAT-integrated Sentinel-1A (RASAT + Sentinel-1A) for Terkos/Istanbul and Kemer/Antalya were 29.18 m and 16.78 m, respectively. Average differences were also calculated using only RASAT images as 13.37 m and 6.29 m for both areas, respectively.