Predictive Riparian Areas and Wetland Mapping of the Williston Drainage Basin
The overall objectives of this project are as follows:
- Complete pre-typing and attribution of all riparian and wetland ecosystems, for FWCP’s Peace area of interest at a 1:20,000 mapping scale;
- Identify and define the wetland and riparian units to be mapped;
- Review existing available ecosystem mapping data to determine which, if any, of the existing data sets are suitable for inclusion in the riparian/wetland mapping;
- Review existing air photo satellite imagery for quality and extent of coverage;
- Conduct quality assurance on all aspects of the TEM projects internally;
- Perform all work to RISC TEM standards and guidelines;
- Provide all results in the required format for storage in the corporate data management environment, as specified by the RISC TEM and TEI data submission standards.
This is a multi-year project. See Phase 1.
Final Report: Executive Summary
The Ministry of Environment signed a Letter of Agreement with the Fish and Wildlife Compensation Program (FWCP) to map wetland and riparian areas within the FWCP Peace region as available ecosystem mapping products lacked coverage and were cost prohibitive to map the remainder of the area. This mapping project builds on the data and methodology developed in a previous project (BAPID 6492), with the goal to improve the understanding of abundance, distribution and connectivity of wetlands in the FWCP Peace. A Random Forest modelling approach was used to map riparian areas and wetlands across the 7.2 million hectare FWCP Peace region at a 25m pixel resolution. Improved 3-category (water, wetland, upland), 10-category (bog, fen, swamp, marsh, low-bench floodplain, mid-bench floodplain, high-bench floodplain, shallow water wetland, upland, and water) and 46-category predictive wetland raster maps were produced. Model results were validated using field observations collected during the summer of 2017, 2018, and 2019. The 3-category map is 86.7% accurate at predicting wetland, upland and water across the FWCP Peace Region, and had an internal model accuracy of 93% making it a reliable tool for use in wetland and riparian ecosystem planning and management. The 10-category map has an accuracy of 53% and should be used with caution. The 46-category raster prediction is not reliable and should not be used.When compared to models produced for BAPID 6492, model accuracy improved by 13% for the 3-category, and 7% for the 10-category predictions across the FWCP Peace region. At the sub-basin level, the 3-category prediction was improved by 3% in the Finlay sub-basin and 14% in the Parsnip sub-basin where new training points were added. Importantly, methods and products were specifically designed for openness and transparency, thereby increasing value through further use and extension opportunities. Further considerations for extending the product to a wetland management context, as well as data and model improvements as part of continuous improvement cycle are presented herein. Key results from this project include the 3-category raster prediction (20190926-103025_map_recl.tif), 10-category raster prediction (20190627_114031_map_recl.tif), and an Esri ArcGIS interactive mapping application.