CBI in the field
Justin Brice, Dustin Pearce, Esti Shay, Kaveh Karimi, Bill Klinkow, Gladwin Joseph, and CBI volunteer, Heather Bishop, recently traveled to Mississippi to gather additional ground-truth data for random forest machine learning models built by Dustin Pearce. The purpose of the models are to accurately predict forest metrics (e.g., basal area, biomass, forest type) for Bottomland Hardwood Forests in Mississippi, which are privately owned lands enrolled in the Conservation Reserve Program (CRP). The CRP pays landowners to maintain Bottomland Hardwood Forests, which provide important ecological benefits such as removing nitrogen and phosphorus from water, storing flood waters and reducing downstream flooding, trapping sediment, and promoting carbon sequestration.
This work is part of a pilot project with the United States Department of Agriculture (USDA) and the Farm Service Agency and depicts ecological and economic features across Bottomland Hardwood Forest Conservation Reserve Program lands in the state of Mississippi. By building an online platform that provides metrics on these Conservation Reserve Program (CRP) lands, the USDA Farm Service Agency will be better able to monitor and evaluate existing acres of Bottomland Hardwood Forests that are part of the CRP.
To validate how well the random forest models predict actual conditions, the CBI team surveyed reforested farmlands enrolled in the CRP, which were planted with trees at least 8 years of age or older. The team set up two 10 or 20 meter plots per CRP parcel and measured species composition, basal area, tree height, above-ground biomass, and stand density.
The random forest model model outputs aligned well with on-the-ground conditions when surveyed by the CBI team. The models use Sentinel 2 satellite imagery and Sentinel 1 Synthetic Aperture Radar data which are updated weekly. This allows for high resolution monitoring over time without having to send an army of people to survey on the ground. As a result, the random forest models are both a time and money-saver.
This pilot project determined the utility of the online platform and remote sensing methods, which can be expanded to all regions where the CRP restores and enhances Bottomland Hardwood Forests.
Photo credits: Justin Brice and Kaveh Karimi