The Vegzone layer was derived from overstory and understory species composition and abundance (percent cover) information in existing Gradient Nearest Neighbor (GNN) vegetation maps developed by the Landscape Ecology, Modeling, Mapping, and Analysis (LEMMA) group in the Pacific Northwest Research Station (http://lemma.forestry.oregonstate.edu/).

Species composition in the imputations is determined from forest inventory plot data (1990-2019). A combination of FIA-Periodic, FIA-Annual, and R6/R5 CVS plot designs were used to determine PVT at the vegzone (Series) and subzone (Subseries or Plant Association Group (PAG)) levels. FIA-Periodic and R6/R5 CVS Inventory plots were used where they were co-located with FIA-Annual plots to help interpret previous or subsequent disturbance and succession. These Interpretations are based on multiple observations (2-5 obs per location) of species composition through time. Each plot location was assigned 1 vegzone and 1 subzone across all observations based on a vegzone/subzone ruleset (Key) programmed in R.

Map Data (1986-2017) consists of annual imputations of FIA-Annual Plots (2001-2016) only, using GNN K-1 (single nearest neighbor) methodology (Ohmann 2001). Each raster call was evaluated across all imputation years (1986-2017) for subzone Majority, subzone Min Rank (earliest subzone in the ruleset), and subzone Variety. Subzone Variety was used as an index of disturbance through time. Where subzone Variety >4, the cell was considered disturbed and subzone Min Rank was used in the final PVT map for that raster cell. Where subzone Variety was <=4, the cell was considered undisturbed and subzone Majority was used in the final PVT map.

Presence and percent cover of tree species were used to infer potential forest vegetation zones. Other indicator species (trees not used in vegetation zone delineation and understory species) were grouped by moisture and temperature regimes. The abundance and number of species in a group served as indicators for moisture and temperature subzones within a vegetation zone. This delineation often indicated different disturbance regimes, rates of development after disturbance, and ultimately what type of forest would dominate in the absence of disturbance. GNN maps for the entire 1986-2017 LandTrendr normalized Landsat time sequence were used to account for species composition changes owing to recent disturbances such as fire or timber harvest. Where imputations from different map years were highly variable (>4 different vegzone/subzones) indicating 1 or more disturbances, the higher ranked (generally more mesic) vegetation zone and subzone was assigned.