Mapping Marginal Land Based on Decadal CDL and MODIS Land Cover Products in the Contiguous United States

Renewable energy production will compete limited land resources with food production and natural environment. Utilizing marginal lands which are poorly suited to food crops and easy to convert to bioenergy crops is a key solution. Many studies estimated the spatial distribution of marginal lands, but they were limited in (1) subjective biophysical and environmental thresholds or certain land use types, and (2) coarse resolution or small scale map products. This study developed a new framework to identify marginal lands based on satellite-observed land use change (LUC), produced the first 30m resolution marginal lands dataset across the whole Contiguous United States (CONUS), and evaluated the framework and the dataset from multiple prospects. We hypothesis that LUC is a result of compounding factors. We presented an algorithm to identify croplands in transition from decadal Cropland Data Layer (CDL) data, constrained by MODIS Land Cover C6 product and USDA statistical data. We derived maps of marginal lands, and the total amount across the CONUS is 21 - 31 million acre with confidence and 104 - 128 million acre with uncertainty. By comparing with other datasets, we found (1) satellite-derived high resolution LUC is reliable, (2) marginal lands detected from LUC are biophysically, environmentally and socioeconomically marginal, and (3) LUC-based marginal lands can serve as a lower boundary while other methods upper boundary. Future study will explore the potential of this satellite-derived marginal lands dataset on U.S. land use decision-making.