Land-use and land-management scenarios to support regional-global ecosystem assessments

Land use and land cover change are known to be significant drivers of regional and global changes to the biogeochemical and biophysical properties of the Earth surface, with resulting implications for the climate system, carbon stocks, biodiversity, and economic resilience. To incorporate these effects into Earth System Models (ESMs) and Dynamic Global Vegetation Models (DGVMs), new gridded datasets of land-use states and transitions have been developed and adopted as a required forcing for CMIP6 experiments. Of particular importance in these datasets are the forest-related land-use changes, and to reflect this the second-generation of these Land-Use Harmonization datasets (LUH2) employed Landsat-based forest extent and change data (Hansen et al. 2013) as both a constraint and a diagnostic test of the spatial patterns of wood harvesting. After reconciling forest loss definitions and developing new wood harvest patterns, the LUH2 estimated global forest loss for the period 2000-2012 was reduced from over 8.3 million km² to 1.78 million km² (compared with the remote-sensing-based forest loss of 2.03 million km²).

LUH2 was designed to facilitate global modeling efforts, and at a global scale the new LUH2 forest loss estimates compare very well with Landsat observations. However, there is now also a growing need for land-use change and management datasets to inform regional studies. The Southeast USA is a highly productive and largely forested region with an important timber industry, and is a region in which forest loss discrepancies still remain between LUH2 and Landsat forest loss data. In this region the LUH2 dataset estimates more forest loss than is observed via Landsat. Building upon our prior work in LUH2, we are now developing a Landsat-constrained and regionally-specific LUH dataset for the Southeast region of the USA, that explicitly models regional forest management practices (such as plantation forestry). This work is part of a larger effort towards generating land-use datasets that are both globally consistent and locally accurate.