Coastal Analyst System from Space Imagery Engine (CASSIE): Shoreline management module

https://doi.org/10.1016/j.envsoft.2021.105033Get rights and content

Highlights

  • CASSIE: new web tool for automatic shoreline detection and analysis from satellite imagery.

  • Open-source, and built with JavaScript using Google Earth Engine API.

  • CASSIE enables access to Sentinel-2 and Landsat satellite imagery datasets.

  • The accuracy of the CASSIE-derived shoreline and rates-of-changes are sub-pixel.

  • Can be applied at any coastal region and requires minimum computer memory and storage.

Abstract

CASSIE is an open-source web tool for automatic shoreline mapping and analysis using satellite imagery (www.cassiengine.com). This tool was built in JavaScript, using Google Earth Engine (GEE) API, and can be applied to any coastal region on Earth where a boundary between water and land exists. CASSIE use the Landsat and Sentinel-2 satellite imagery, freely available from GEE, and implements an automatic shoreline detection using the normalized difference water index (NDWI) together with Otsu image segmentation algorithm. The satellite-derived shorelines are analysed for a set of user-defined cross-shore transects along which several statistical analyses are performed. Comparisons of the CASSIE-derived shorelines and rate of changes with state-of-the art methods show that the products from this tool have sub-pixel accuracy. The new concept of big-data cloud-based computing and storage platform (GEE), together with a user-friendly interface and high level of automation makes of CASSIE a complete tool to support a wide variety of studies and applications where the knowledge of the shoreline behaviour is fundamental.

Introduction

The shoreline position, defined as the interface between land and water surface, is the most commonly used indicator to quantify erosion-accretion changes in coastal areas (Douglas and Crowell, 2000; Boak and Turner, 2005; Hanslow, 2007; Luijendijk et al., 2018; Vousdoukas et al., 2020). Shorelines are inherently dynamic features in constant modification driven by natural processes such as wave-induced runup and setup, tides (astronomic and meteorological), and mean sea level oscillations; or anthropogenic factors such as coastal structures, and side effects of human activities, like dams, or clearing of coastal ecosystems, such as mangrove forests (Mentaschi et al., 2018). It is estimated that 24% of world's sandy coasts are suffering shoreline retreat (rates exceeding 0.5 m/yr), while 28% are accreting and 48% are stable (Luijendijk et al., 2018). Land losses areas due to coastal erosion are significantly higher than the gained land, and according to Mentaschi et al. (2018) anthropogenic factors are the dominant driver underlying this change. Currently the global mean sea level is rising at a rate of 2.9 mm/yr (Nerem et al., 2018) and more frequent extreme events are expected to occur (Mölter et al., 2016; Murakami et al., 2017), increasing the likelihood of shoreline retreat being enhanced during the 21st century (Voudoukas et al., 2020).

Shoreline monitoring is a vital element of planning and management of the coast to mitigate and adapt to impacts of climate change. Quantitative assessment of shoreline trends (erosion or accretion) is considered as a crucial indicator of coastal vulnerability to natural hazards, and one of the most important parameters used to determine coastal vulnerability indices (McLaughlin and Cooper, 2010; Hzami et al., 2021), or to feed shoreline prediction models (e.g., Montaño et al., 2020). Data-scarcity, however, remains an ongoing challenge as long-term coastal monitoring programs, based on in-situ measurements, are limited to only a few sites around the world (e.g., Barnard et al., 2015; Turner et al., 2016). Publicly available Earth Observation Satellite imagery (e.g., NASA′ Landsat missions or ESA′ Sentinel's mission), emerge in this context as potential solution to overcome this problem, providing long-term observations (since 1984 to the present for the case of Landsat missions), moderate spatial resolution (pixel size between 10 and 30 m), short revisit times (between 5 and 16 days), and worldwide coverage (Goward et al., 2006; Bergsma and Almar, 2020).

Previous studies have investigated the potential of medium spatial resolution satellite images to study shoreline change (Blodget et al., 1991; White and El-Asmar, 1999; Liu and Jezek, 2004; Ekercin, 2007; Wang et al., 2010; Pardo-Pascual et al., 2012; Vos et al., 2019a). Uncertainties associated to the shoreline detection was found to be in the range between ±5 m and ±14 m (e.g., Garcia-Rubio et al., 2015; Wang et al., 2018; Pardo-Pascual et al., 2018; Yadav et al., 2018; Vos et al., 2019b), highlighting that these datasets are suitable to monitor moderate and significant structural changes in the shoreline position.

Mapping and analysing shoreline evolution requires an exhaustive and multi-task process that involves selection and download of satellite imagery (freely available on USGS Earth Explorer portal or ESA Sentinel Data Hub), image pre-processing (e.g., radiometric calibration and atmospheric corrections), manual or automatic shoreline digitalization, and statistical analysis (e.g., using freely available software like Digital Shoreline Analysis System – Thieler et al., 2017). In addition to this, the increasingly quality of satellite imagery (higher spatial and radiometric resolution) demands for higher computing capacity (including storage space and processing power) that are not always available, especially in developing countries.

Crucially, the advent of Google Earth Engine (Gorelick et al., 2017) has facilitated the access to the growing archive of publicly available satellite imagery, processing tools and computing power (Google Cloud) providing the opportunity for global-scale analyses stretching back decades (e.g., Donchyts et al., 2016; Li et al., 2019; Luijendijk et al., 2018; Mentaschi et al., 2018). In addition to the web-platform with code editor, that allow users to access and analyse historical remote sensing datasets, Google Earth Engine (GEE) offers the possibility for developers to create their own tools using the GEE Application Programming Interface (API). Advent of the use of GEE technology, is the recently developed CoastSat software (Vos et al., 2019b). This toolkit was written in Python and enables the user to obtain time-series of shoreline position at any coastline worldwide of publicly available (GEE) satellite imagery. Although Coastsat implement novel shoreline detection techniques with sub-pixel accuracy, its use of GEE platform is restricted to the access to the satellite imagery, while the remaining processes are performed in other platforms. In addition to this Coastsat does not produce statistical analysis on the shoreline dataset, which makes of this task dependent from other software's.

The aim of this work is to present a new webtool built with GEE API - CASSIE (Coastal Analyst System from Space Imagery Engine) – that can both automatically extract shoreline position from satellite imagery and perform statistical analysis, at any coastal area worldwide. In contrast with the existing tools that can only perform one of these tasks (Palomar-Vázquez et al., 2018; Vos et al., 2019b; Jackson et al., 2012; or Thieler et al., 2017), CASSIE combines these two processes into a single tool, and in addition to this makes use of GEE computing power/storage (Google Cloud) that reduces the user requirement infrastructure to internet access only. CASSIE webtool is freely available and can operate in different platforms (e.g., desktop PC, laptop, or smartphone), which makes of this tool highly portable (e.g., can be used in classrooms, workshops, office, or fieldwork). With CASSIE is possible to perform shoreline analysis and export the products (shorelines, baseline, transects, and statistical results) to ESRI shapefile format, which allows the end-user to further explore their datasets in GIS tools. In this work we present a detailed description of CASSIE workflow and implemented algorithms (section 2.1.), a comparison between CASSIE-derived shoreline position with ground-truth (RTK-GPS) shoreline observations, and a comparison between CASSIE-derived shoreline rates of change with existing similar datasets (section 2.2).

Section snippets

CASSIE workflow

The ShoreAnalyst module of CASSIE is a graphical user interface (GUI) built with JavaScript GEE API and can be used to load and visualize satellite imagery, and to conduct automatic shoreline detection and analysis to any location on Earth where a boundary between water and land exist. The workflow of the different processes that involve the user interaction with the GUI and internal processing that leads to CASSIE results is presented in Fig. 1 and described in the following subsections.

Comparison of shoreline position with in-situ measurements

The comparison between the CASSIE-derived shorelines (retrieved from Landsat 8 image), using different levels of smoothing, and RTK-GPS in-situ observations show, an overall very good agreement, with the best scores showing coefficient of determination of 0.95 and a RMSE of 8.84 m, using 3 neighbour points and 0.75 standard deviation (Fig. 9). These parameters that resulted in the best scores were set as default in CASSIE smoothing filter. Similar method (1D Gaussian filter) and number of

Conclusions

This work describes and demonstrates capabilities of a new tool (CASSIE) for automatic shoreline mapping and analysis based on satellite imagery. CASSIE was built with JavaScript GEE API, and can be applied to any location on Earth, where a boundary between water and land exists. CASSIE makes use of the Landsat and Sentinel 2 satellite imagery (available on GEE) and of the power computing capacity and storage of Google Cloud. Therefore, it can be run from a standard desktop, laptop or

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors would like to thank the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and Ministério da Ciência, Tecnologia e Inovações (MCTIC) of Brazil for funding this research (project No. 441545/2017-3). The authors would like to thank the National Fund on Climate Change of the Ministry of the Environment of Brazil (Process number 352012) for sharing the field observations of the shoreline position measured by RTK-GPS. The authors would like to thank Franco Prado and

References (54)

  • H. Li et al.

    A google earth engine-enabled software for efficiently generating high-quality user-ready landsat mosaic images

    Environ. Model. Software

    (2019)
  • J.E. Pardo-Pascual et al.

    Automatic extraction of shorelines from Landsat TM and ETM+ multi-temporal images with subpixel precision

    Remote Sens. Environ.

    (2012)
  • E. Vermote et al.

    Preliminary analysis of the performance of the Landsat 8/OLI land surface reflectance product

    Remote Sens. Environ.

    (2016)
  • K. Vos et al.

    Sub-annual to multi-decadal shoreline variability from publicly available satellite imagery

    Coast Eng.

    (2019)
  • K. White et al.

    Monitoring changing position of coastlines using Thematic Mapper imagery, an example from the Nile Delta

    Geomorphology

    (1999)
  • P. Baptista et al.

    Monitoring sandy shores morphologies by DGPS-A practical tool to generate digital elevation models

    J. Coast Res.

    (2008)
  • P.L. Barnard et al.

    Coastal vulnerability across the pacific dominated by El niño/southern oscillation

    Nat. Geosci.

    (2015)
  • E.H. Boak et al.

    Shoreline definition and detection: a review

    J. Coast Res.

    (2005)
  • G.V. Da Silva et al.

    Shoreline change analysis and insight into the sediment transport path along Santa Catarina Island North shore, Brazil

    J. Coast Res.

    (2016)
  • A.T.K. Do et al.

    The estimation and evaluation of shoreline locations, shoreline-change rates, and coastal volume changes derived from landsat images

    J. Coast Res.

    (2018)
  • B.C. Douglas et al.

    Long-term shoreline position prediction and error propagation

    J. Coast Res.

    (2000)
  • G. Donchyts et al.

    Earth's surface water change over the past 30 years

    Nat. Clim. Change

    (2016)
  • S. Ekercin

    Coastline change assessment at the Aegean Sea Coasts in Turkey using multitemporal Landsat imagery

    J. Coast Res.

    (2007)
  • L.S. Esteves et al.

    The problem of critically eroded areas (CEA): an evaluation of Florida beaches

    J. Coast Res.SI

    (1998)
  • S. Goward et al.

    Historical record of Landsat global coverage

    Photogramm. Eng. Rem. Sens.

    (2006)
  • D.J. Hanslow

    Beach erosion trend measurement: a comparison of trend indicators

    J. Coast Res.SI

    (2007)
  • A. Hzami et al.

    Alarming coastal vulnerability of the deltaic and sandy beaches of North Africa

    Sci. Rep.

    (2021)
  • Cited by (40)

    View all citing articles on Scopus
    View full text