Shoreline Extraction Using Image Processing of Satellite Imageries

Shorelines are one of the most important land surface phenomena with a dynamic nature. Coastal monitoring is an important parameter in sustainable development and environmental protection. Coastal monitoring requires the extraction of shorelines at different times. Nowadays, remote sensing data are considered the most efficient source of information for the study and interpretation of coastal landforms, tidal levels and changes of shoreline and water depth. The main objectives of studying the shoreline from Charak port to Aftab port are to provide a map to calculate shoreline changes, determine erosion and accretion areas and the extent of shoreline advance and retreat to better coastal management. Furthermore, design and construction of suitable coastal facilities, determining the safe margin and ultimately realizing the 14th principle of sustainable development, are other objectives. The case study in this paper is to extract the shoreline of Charak port to Aftab port in the Persian Gulf using five Landsat satellite imageries from 1998 to 2017. For this purpose, in the first stage by applying radiometric, atmospheric and noise reduction corrections and in the second stage by taking training samples of each imagery and also supervised classification methods (SVM and maximum likelihood methods), the boundary between "water" and "land" was specified in each image. Then, the accuracy of each classification method was evaluated using two parameters of overall accuracy and kappa coefficient. The results of image classification methods show that the SVM and maximum likelihood methods have a high accuracy in separating the boundary between water and land. Nevertheless, the SVM method has a higher accuracy than the maximum likelihood method. In these two methods, the minimum and maximum differences in overall accuracy are 0.06 and 0.6%, respectively. Furthermore, the minimum and maximum differences in kappa coefficient are 0.0016 and 0.0117, respectively. Finally, these imageries were transferred to ArcGIS software and shorelines were extracted and beach maps were prepared at a scale of 1: 50,000.