A Multisensor Multiresolution Method for Mapping Vegetation Status, Surficial Deposits, and Historical Fires Over Very Large Areas in Northern Boreal Forests of Quebec, Canada

Abstract

Boreal forests have a significant impact on the Earth’s climate and on global warming. In this context, a mapping method was developed and was capable of dealing with very large areas with a lack of support datasets, which could characterize the current forest, surficial deposits, and forest disturbance history. The method based on remote sensing data, images processing techniques, and images interpretation was applied over a very large area ($680\,000\;\text{km}^2$) in Québec, QC, Canada that was dominated by black spruce (Picea mariana [Miller] BSP). It involved five steps: 1) mapping the vegetation based on unsupervised classification, imputation, and segmentation methods; 2) mapping the history of fires that occurred over the mapping area based on archives Landsat images; 3) determining the dominant species characterizing forest stands; 4) mapping surficial deposits; and 5) accuracy assessment of map attributes based on video dataset. Kappa values ranged from 72.5% to 96.3%, indicating substantial agreement when compared with validation dataset. The results demonstrated that our method is a convenient and inexpensive way of mapping forest ecosystems over large areas of northern boreal forest.

Keywords

• Boreal
• fire history
• forest attributes
• Landsat
• RapidEye
• surficial deposits