Annual Maps of Forest Structure and Fuel Attributes
These maps were created following the two-step modeling methodology outlined in Hudak et al. (2020) in a project funded by the NASA Carbon Monitoring Systems (CMS) Program. First, field observations contributed by project stakeholders across the western United States were related to airborne lidar metrics of canopy structure and topography, and to west-wide climate variables, to predict and map forest structure variables across airborne lidar extents. Second, the lidar-derived maps were used to train a model that annually predicts lidar forest structure estimates from Landsat image time-series metrics derived from LandTrendr, topography, and west-wide climate variables.
Annual forest structure maps were bias-corrected with simple linear regression models and annual FIA field inventory data available from 2011 through 2019 across the 11 western states. A square-root transformation was applied to FIA measurements and Random Forest map values. Bayesian inference provided a posterior predictive distribution, allowing predictions to be returned to their original scale while avoiding conventional back-transformation bias.
A forest/nonforest mask limits estimates and inferences to forested areas. Following the annual tree-canopy-cover methodology of Egorov et al. (2023), hindcast to 1985 by Egorov et al. (2024), pixels were classified as forest when more than 5 of the 35 annual maps from 1985-2019 showed greater than 10% tree canopy cover.
The mapped attributes include aboveground and belowground biomass, basal area, board feet, canopy base height, canopy bulk density, canopy fuel load, downed dead woody debris, forest floor fuel, shrub and herbaceous fuel, stand density index, stand height, snag density, tree density, and tree volume.
Egorov, A., Roy, D.P., and Boschetti, L. 2023. Generation and comprehensive validation of 30 m conterminous United States Landsat percent tree cover and forest cover loss annual products. Science of Remote Sensing, 7, 100084. https://doi.org/10.1016/j.srs.2023.100084.
Egorov, A., Roy, D.P., and Boschetti, L. 2024. Tree Cover Dynamics CONUS Dataset (Version 1.0) [Data set]. University of Idaho. https://doi.org/10.7923/s6d0-3b36.
Emick, E., Babcock, C., White, G.W., Hudak, A.T., Domke, G.M., and Finley, A.O. 2023. An approach to estimating forest biomass while quantifying estimate uncertainty and correcting bias in machine learning maps. Remote Sensing of Environment, 295, 113678. https://doi.org/10.1016/j.rse.2023.113678.
Hudak, A.T., et al. 2020. A carbon monitoring system for mapping regional, annual aboveground biomass across the northwestern USA. Environmental Research Letters, 15(9), 095003. https://doi.org/10.1088/1748-9326/ab93f9.
Kennedy, R.E., Yang, Z., and Cohen, W.B. 2010. Detecting trends in forest disturbance and recovery using yearly Landsat time series: 1. LandTrendr—Temporal segmentation algorithms. Remote Sensing of Environment, 114, 2897-2910.
This work was funded by NASA Carbon Monitoring Systems Program Awards 80HQTR20T0002 and NNH24OB24A.
This work used the Alpine High-Performance Computing resource at the University of Colorado Boulder. Alpine is jointly funded by the University of Colorado Boulder, the University of Colorado Anschutz, Colorado State University, and the National Science Foundation (award 2201538).
