2023-05-04T13:11:58.897473

2023-05-04T13:11:58.897475

Global

Woods Hole Research Center. Unpublished data. Accessed through Global Forest Watch Climate on [date]. climate.globalforestwatch.org

Aboveground live woody biomass density

Woods Hole Research Center

[Creative Commons CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)

This work generated a global-scale, wall-to-wall map of aboveground biomass (AGB) at approximately 30-meter resolution. This data product expands on the methodology presented in Baccini et al. (2012) to generate a global map of aboveground live woody biomass density (megagrams biomass ha-1) at 0.00025-degree (approximately 30-meter) resolution for the year 2000. Aboveground biomass was estimated for more than seven hundred-thousand quality-filtered Geoscience Laser Altimeter System (GLAS) lidar observations using allometric equations that estimate AGB based on lidar-derived canopy metrics. Forty-seven allometric equations were compiled from more than 20 scientific publications, with each equation developed for a different region and forest type. The most appropriate equation was determined for each shot, accounting for land cover, burned status, and Terrestrial Ecoregion of the World (TEOW) ecoregion data (Olson et al. 2001) for each shot. The equations were applied to the GLAS data, generating an AGB estimate for each shot in the global GLAS dataset. A subset of shots was classified as zero-biomass based on GLAS data and tree canopy cover and were assigned an AGB of 0 Mg ha-1. The global set of GLAS AGB estimates was used to train random forest models that predict AGB based on spatially continuous data. The predictor datasets include Landsat 7 Enhanced Thematic Mapper Plus (ETM+) top-of-atmosphere reflectance and tree canopy cover from the Global Forest Change version 1.2 dataset (Hansen et al. 2013), 1 arc-second SRTM V3 elevation (Farr et al. 2007), GTOPO30 elevation from the U. S. Geological Survey (for latitudes greater than 60° N), and WorldClim climate data (Hijmans et al. 2005). The predictor pixel values were extracted and aggregated for each GLAS footprint in order link the GLAS AGB estimates with the predictor data. A random forest model was trained for each of six continental-scale regions: the Nearctic, Neotropic, Palearctic, Afrotropic, Tropical Asia, and Australia regions. The six regions were delineated based on aggregations of TEOW ecoregions. The predictor layers were stacked (the elevation and climate layers were resampled to match the 30-meter resolution of the Landsat inputs), and each random forest model was applied to all pixels within its region. The data are AGB density values (megagrams biomass/hectare); aboveground carbon density values can be estimated as 50 percent of biomass density values. In addition to the AGB density map, there is an error map for an earlier version of the AGB map. This map of the uncertainty in AGB density estimation accounts for the errors from allometric equations, the LiDAR based model, and the random forest model. The error map for the current version of the biomass density map is not available yet.

Shows carbon density values of aboveground live woody biomass

It is recommended that both aboveground biomass density and uncertainty values be used together for biomass assessments and verification. The map will provide accurate estimates of aboveground biomass stock and aboveground biomass density when aggregated to large areas (5,000 to 10,000 ha) for project and regional level assessments. The biomass density value of a single pixel may have large uncertainty when compared with small plots for verification. The uncertainty map currently available is from an earlier version of the biomass density map.

Core data set for GFW Climate

2bd671ee-50b5-465c-8012-21905edf6202