An efficient protocol for accurate and massive shoreline definition from mid-resolution satellite imagery

Abstract

Satellite images may constitute a useful source of information for coastal monitoring as long as it is possible to manage them in an efficient way and to derive precise indicators of the state of the beaches. In the present work, SHOREX system is employed for managing and processing Landsat 8 and Sentinel 2 images to automatically define the instantaneous shoreline position at sub-pixel level. Between the years 2013 and 2017, 91 satellite-derived shorelines (SDS) were assessed by comparing with high-resolution shorelines obtained simultaneously through video-monitoring. The analysis allowed identifying the combination of parameters to perform the extraction algorithm with the highest accuracy. Furthermore, an efficient self-contained workflow is proposed, more robust and independent from inaccuracies in the approximate input line and from multiple morphological and oceanographic issues that may condition the radiometric response near the shore. An iterative procedure ensures firstly a suitable kernel of analysis representing the water-land interface to get, afterward, the definition of the sub-pixel shoreline with high accuracy (below 3 m RMSE).