On the accuracy of automated shoreline detection derived from satellite imagery: A case study of the Sand Motor mega-scale nourishment

Gerben Hagenaars, Sierd de Vries, Arjen Luijendijk, Wiebe de Boer, Ad Reniers

Research output: Contribution to journalArticleScientificpeer-review

90 Citations (Scopus)
49 Downloads (Pure)

Abstract

Measured trends and variability in shoreline position are used by coastal managers, scientists and engineers to understand and monitor coastal systems. This paper presents a new and generic method for automated shoreline detection from the largely unexplored collection of publicly available satellite imagery. The position of the obtained Satellite Derived Shoreline (SDS) is tested for accuracy for 143 images against high resolution in-situ data along a coastal stretch near the Sand Motor, a well-documented mega-scale nourishment along the Dutch coast. In this assessment, we quantify the effects of potential inaccuracy drivers such as the presence of clouds and wave-induced foam. The overall aim of this study is to verify whether the SDS is suitable to study structural coastline trends for coastal engineering practice.

In the ideal case of a cloud free satellite image without the presence of waves, with limited morphological changes between the time of image acquisition and the date of the in-situ measurement, the accuracy of the SDS is with subpixel precision (smaller than 10–30 m, depending on the satellite mission) and depends on intertidal beach slope and image pixel resolution. For the highest resolution images we find an average offset of 1 m between the SDS position and the in-situ shoreline in the considered domain. The accuracy deteriorates in the presence of clouds and/or waves on the image, satellite sensor corrections and georeferencing errors. The case study showed that especially the presence of clouds can lead to a considerable seaward offset of the SDS of multiple pixels (e.g. order 200 m). Wave-induced foam results in seaward offsets in the order of 40 m.

These effects can largely be overcome by creating composite images, which results in a continuous dataset with subpixel precision (10–30 m, depending on the satellite mission). This implies that structural trends can be detected for coastlines that have changed with at least the pixel resolution within the considered timespan.

Given the accuracy of composite images along the Sand Motor in combination with the worldwide availability of public satellite imagery covering the last decades, this technique can potentially be applied at other locations with large (structural) coastline trends.
Original languageEnglish
Pages (from-to)113-125
JournalCoastal Engineering
Volume133
DOIs
Publication statusPublished - 2018

Bibliographical note

Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Keywords

  • Automated shoreline detection
  • Satellite imagery
  • Google earth engine
  • NASA
  • ESA
  • Positional accuracy
  • Coastline trends
  • Coastal management
  • Dutch coast
  • Sand Motor

Fingerprint

Dive into the research topics of 'On the accuracy of automated shoreline detection derived from satellite imagery: A case study of the Sand Motor mega-scale nourishment'. Together they form a unique fingerprint.

Cite this