SCANFI v2: the Spatialized CAnadian National Forest Inventory data product
Published by Natural Resources Canada | Ressources naturelles Canada
About this dataset
This data publication contains a set of 30m resolution raster files representing Canadian wall-to-wall maps, at 5-year intervals from 1985 to 2025, of broad landcover type, forest canopy height, degree of crown closure and live aboveground dry tree biomass, along with species composition of several major tree species. The Spatialized CAnadian National Forest Inventory data product (SCANFI) was developed using the newly updated National Forest Inventory photo-plot dataset, which consists of a regular sample grid of photo-interpreted high-resolution imagery covering all of Canada’s non-arctic landmass. SCANFI was produced using temporally harmonized summer and winter Landsat spectral imagery along with hundreds of tile-level regional models based on a novel k-nearest neighbours (k-NN) and random forest imputation method. SCANFI is not meant to replace nor ignore provincial inventories which could include better and more regularly updated inputs, training data and local knowledge. Instead, SCANFI was developed to provide a current, spatially-explicit estimate of forest attributes, using a consistent data source and methodology across all provincial boundaries and territories. SCANFI is the first coherent 30m Canadian wall-to-wall map of tree structure and species composition and opens novel opportunities for a wide range of studies in a number of areas, such as forest economics, fire science and ecology. SCANFI v2 is described here. The main methodological and performance differences between SCANFI v1 (https://doi.org/10.23687/18e6a919-53fd-41ce-b4e2-44a9707c52dc) and SCANFI v2 are briefly outlined below. For a more detailed description of these changes, please consult the accompanying report. For the scientific publication about SCANFI v1, see Guindon et al. (2024). SCANFI version 2 main updates 1. Transition to Landsat Collection 2 Updated all Landsat time series inputs from Collection 1 to Collection 2, providing improved radiometric calibration and geometric consistency across sensors and years. 2. SCANFI time series with LandTrendr-smoothed winter imagery Introduced LandTrendr-smoothed winter imagery, enabling a full 5-year interval time series from 1985 to 2025 with dual-season inputs. 3. Updated climate normals Replaced previous climate layers with smoother 1991–2020 normals (MacDonald et al. 2024), eliminating tiling artifacts. 4. Increased sample selection density of NFI training data Shifted from a photoplot-level 250m systematic grid to a denser sampling strategy: 4 pixels/hectare per polygon (max 100 points, excluding border pixels). This sampling strategy greatly increased the number of training points available, improving representation of inter- and intra-polygon variability and strengthening model performance. 5. More accurate water prediction and delineation of urban areas and croplands Water and cropland prediction is now done using a soon-to-be-published landcover time series layer in order to: (i) prevent the misclassification of recently disturbed areas as water in SCANFI v1; and (ii) convert urban areas and croplands to forested land before 2020 when the Agriculture and Agri-Food Canada (AAFC) land use map used in the original study is not available (Agriculture and Agri-Food Canada, 2022). 6. Forest age map SCANFI v2 introduces new national forest age maps by combining recent disturbance history data with spatialized NFI age estimates for undisturbed stands, producing consistent stand age attribution across Canada from 1985 to 2025, at 5-year intervals. The resulting approach reduces noise relative to single-year estimates and yields spatially complete, temporally stable forest age maps. 7. Modified response and target variable sets Water is no longer used as a target variable, whereas land position (alpine, wetland or upland) was added to the target variable list and is now an output layer. The selection of Landsat bands and indices used in modeling was also refined, improving sensitivity to forest str