Watching the Grass Grow: Delineating Coastal Vegetation Edges from Satellite Imagery

Muir, F. , Hurst, M. , Richardson-Foulger, L., Rennie, A. and Naylor, L. (2023) Watching the Grass Grow: Delineating Coastal Vegetation Edges from Satellite Imagery. 2023 British Society for Geomorphology Annual Meeting, Edinburgh, UK, 4-6 Sept 2023. (Unpublished)

Abstract

Our coasts are increasingly under threat of climate change related risks such as sea level rise, increased erosion rates, and increased frequency and severity of storms and associated wave action. To identify and adequately support the communities at greatest risk of these impacts, regular and repeatable observations of coastal change are required. Shoreline positions offer a simplistic measure of geomorphic change across the intertidal zone, but they fluctuate significantly in macrotidal areas and are subject to tidal bias. To gain a broader understanding of the interplay of coastal processes, a coupling of change indicators is desirable. A more stable measure of coastal change, and one arguably more relevant to coastal communities and infrastructure, is the vegetation edge. Presented here is a Python toolkit which builds on the shoreline extraction tool CoastSat (https://doi.org/10.1016/j.envsoft.2019.104528), but adapted to automatically identify coastal vegetation edges from satellite imagery. A trained neural network classifies pixels and uses Weighted Peaks to extract sub-pixel contours between vegetation and non-vegetation classes. Sentinel-2 images offer the highest accuracy at the test site of St Andrews (RMSE of 10.4m). By taking advantage of the back catalogue of freely available satellite images, dense vegetation timeseries with an average observation interval of 12 days can be easily built to assess historic trends. Extracted vegetation edges are then compared with other derived coastal characteristics such as the vegetation transition zone, beach width, and dune slope, to infer relationships between these different change indicators and therefore create proxies for predicting different geomorphic regimes.

• Item Type: Conference or Workshop Item
• Additional Information: This work is partially supported by funding from NatureScot and the Dynamic Coast Project, and from the NERC IAPETUS2 Research Experience Placement scheme. Contributions provided by CASE partner JBA Trust and in-kind support provided by JBA Consulting.
• Keywords: satellite imagery, coastal, vegetation, machine learning, image classification, Sentinel-2, Landsat, Planet
• Status: Unpublished
• Refereed: No
• Glasgow Author(s) Enlighten ID: Muir, Freya and Naylor, Dr Larissa and Rennie, Dr Alistair and Hurst, Dr Martin
• Authors: Muir, F., Hurst, M., Richardson-Foulger, L., Rennie, A., and Naylor, L.
• Subjects: G Geography. Anthropology. Recreation > GB Physical geography; G Geography. Anthropology. Recreation > GE Environmental Sciences
• College/School: College of Science and Engineering > School of Geographical and Earth Sciences; College of Science and Engineering
• Copyright Holders: Copyright © 2023 The Authors
• Publisher Policy: Reproduced with the permission of the authors

Project Code	Award No	Project Name	Principal Investigator	Funder's Name	Funder Ref	Lead Dept
309062		IAPETUS 2 - Delivering excellence in PhD training across thespectrum of environmental science	Cristina Persano	Natural Environment Research Council (NERC)	N/A	GES - Geography