Land cover classification in an era of big and open data: Optimizing localized implementation and training data selection to improve mapping outcomes (English, HTML)

View metadata and access the original English HTML file: Land cover classification in an era of big and open data: Optimizing localized implementation and training data selection to improve mapping outcomes.

Format

HTML

Access

file-only

Publisher-reported update

2026-03-16

Wall-to-wall map of water bodies across Canada's forested ecosystems for the year 2022, derived from the "water" class of the annual Virtual Land Cover of Engine (VLCE) product. It is developed within the framework of Canada’s National Terrestrial Ecosystem Monitoring System (NTEMS). The VLCE maps are based on Landsat image time-series composites and represent annual land cover classifications from 1984 to 2022 at a spatial resolution of 30 m. The classification process integrates forest change information and ancillary topographic and hydrologic variables, applying a regional modeling framework based on a 150x150 km tiling system ( Hermosilla et al., 2022). Training data are drawn from multiple land cover sources and selected proportionally to land cover distributions using a distance-weighted approach. Classifications are refined over time using a Hidden Markov Model to ensure consistency and reduce classification noise between years. Hermosilla, T., Wulder, M.A., White, J.C., Coops, N.C. 2022. Land cover classification in an era of big and open data: Optimizing localized implementation and training data selection to improve mapping outcomes. Remote Sensing of Environment. 268,…

Access original file