Applied Geomatics Research Group. 2023. Project Report: Validating remote sensing tools to identify effective management measures to protect eelgrass. Applied Geomatics Research Group, NSCC Middleton, NS. https://www.nscc.ca/appliedresearch/docs/dfo-agrg-eelgrass-classification-2023-report.pdf
Executive Summary
The Nova Scotia Community College – Applied Geomatics Research Group (NSCC-AGRG) utilized a support vector machine (SVM) classifier to determine the presence of eelgrass in 14 bays along the Gulf of St. Lawrence. The SVM process was chosen due to its fast processing time and accurate results compared to other classification methods. The classification was conducted on 2022 imagery for seven sites in New Brunswick, six sites in Prince Edward Island, and one in Nova Scotia. The accuracy of the classification was limited by the quality of the imagery and was challenged by sun glint and vertical banding as a result of the sensor collection. Although there was a lack of synchronous and high spatial precision ground truth data, the accuracy assessment reports of the SVM classification demonstrated a generally high detection rate of eelgrass in the bays. Single beam data collected using a BioSonics MX echosounder was processed to analyze plant canopy and bathymetry and provide a secondary form of validation of the SVM classification.
Figure 17 Eelgrass presence/ absence in New London Bay, classification done on WorldView 2 imagery collected on October 1st, 2022, classification completed in ArcGIS Pro using the SVM classifier.
