Agricultural Data Analytics for Environmental Monitoring in Canada

Abstract The development of accurate international reports and domestic policies related to environmental sustainability requires high-accuracy, high-resolution, and multitemporal national resource maps. Although a number of land cover and vegetation maps have been produced for Canada over the past 30 years, the classification schemes often differ, spatial resolution and coverage varies between products, and classification accuracy generally remains at or below 82%–86%. This chapter outlines an approach that uses appropriate analytical procedures to integrate a wide variety of raster and vector spatial products to generate a series of land use maps that reflect the best estimate of the class at each of over 6 billion 30-m pixels in Canada. The premise of the approach was that careful evaluation of inputs and rigorous development of rules regarding the accuracy and preponderance of evidence should bring out the best of each input and allow the development of output maps with higher class and overall accuracies than any of the individual input products. The project resulted in maps for 1990, 2000, and 2010, all at the same scale and categorized according to the six classes of the Intergovernmental Panel on Climate Change: Forest, Cropland, Grassland, Wetland, Settlement, and Otherland.