Colloquium

Monitoring Wetland Restoration Effects with Classification using Remote Sensing

Organised by Laboratory of Geo-information Science and Remote Sensing
Date

Wed 15 January 2025 11:15 to 11:45

Venue Gaia, building number 101
Droevendaalsesteeg 3
101
6708 PB Wageningen
+31 (0) 317 - 48 17 00
Room 2

By Chris Hurink

Abstract
This research investigates the potential of multi-sensor remote sensing for monitoring restoration interventions in wetland ecosystems. Sentinel-1 Synthetic Aperture Radar (SAR) and Sentinel-2 Multispectral Imagery (MSI) were used to classify wetland vegetation types and detect changes across a five-year period (2018–2022) in a wetland ecosystem in Utrecht, the Netherlands. A machine learning model with 49 features was developed to classify wetland type, achieving an overall accuracy of 75.3%. Change detection revealed detectable changes in 33% of the study area, this is in alignment with the 27% change suggested by trusted land cover reference data. The study also evaluated the detectability of the three main intervention types undertaken in this area —Excavation, Deforestation, and Sodcutting— finding that Excavation was the most impactful, with detectable changes in 83% of the targeted areas, compared to 61% for Deforestation and 54% for Sod-cutting. These findings demonstrate the feasibility of using multi-sensor, multitemporal remote sensing for tracking the spatial and temporal effects of wetland restoration interventions. This research offers a starting point for future studies to refine methodologies and enhance the application of satellite imagery data in wetland management and wetland restoration practices.