Flow-line and wet-areas conformance testing of wetland locations using LiDAR and SRTM elevation data

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
University of New Brunswick
This thesis reports on discerning flow networks and wetland borders across forested lands using digital elevation models (DEMs) and Light Detection and Ranging (LiDAR)- generated point cloud data for two contrasting forest zones. This selection refers to the boreal forest zone in northern Alberta’s Ecosystem Management Emulating Natural= Disturbance (EMEND) study area, north of Peace River, and to the temperate forest zone typical of the Acadian Forest in central New Brunswick, as represented by the University of New Brunswick forest in Fredericton (UNB Forest) study area. The DEMs refer to globally available Shuttle Radar Topography Mission (SRTM) elevation rasters with 30 and 90 m spatial resolution, and bare-earth DEMs generated from classified LiDAR point cloud data, interpolated at 1 metre (m) and resampled at 10 m and 30 m spatial resolutions. The methodology involves comparing how DEM-delineated flow-line and wetland border predictions at 1, 10, 30 and 90 m spatial resolution relate with corresponding infield GPS-tracks. It was found that wetland delineations were best when using a combination of DEM-generated wet area model thresholds pertaining to: DEM resolution at 1 m spatial resolution; cartographic depth-to-water index (DTW) < 1 m with flow lines formed at a 4 hectare (ha) minimum upstream contributing area threshold and presence of LiDAR-discerned hydrophytic vegetation patterns, as in raised bogs. The resulting best fitted wetland borders conformed to the GPS-tracked borders within ±20 m nine times out of ten, while false positive and false negative wetland area determinations dropped below 20%. Flow-line locations were best derived from the 1m LiDAR DEMs once hydroconditioned through general depression and road-specific breaching. Flow-line and wetland-border differences between the EMEND and UNB Forest delineations were mainly due to sharper wetland-upland transitions and deeper incision of ephemeral, intermittent and permanent flow channels on rugged (UNB Forest) as opposed to flat terrain (EMEND).