Regional Rates of US Forest Regeneration Measured from Annual Landsat Disturbance History and Ikonos Stereo Imagery

Abstract:

Forests of the Continental United States have been found to be a large contributor to the global atmospheric carbon (C) sink. Here we combine annual Landsat disturbance history with single date IKONOS stereo imagery to estimate the change in forest canopy height with the intention of understanding rates of forest regeneration and C accumulation in the CONUS. We produced canopy height models (CHMs) by differencing digital surface models derived from stereo pairs with national elevation data. CHM data were also derived from Goddard’s airborne LiDAR Hyper-spectral and Thermal Imager (G-LiHT) and linearly correlated to forest inventory and analysis (FIA) data to develop above ground live dry wood biomass coefficients. We used these coefficients to approximate C sequestration rates and through this approach we captured forest regeneration rates. A slow rate of regeneration was found in Florida of 0.30 m yr^‒1 RMSE 7 m, where agro-forestry is less common. We found fast rates of regeneration in the southeastern US of 0.53 m yr^‒1 RMSE < 4.3 m where large tracks of loblolly pine are routinely managed and harvested. Aboveground live dry wood biomass accumulated from 162 gCm^‒2yr^‒1 in Florida to 506 gCm^‒2yr^‒1 in South Carolina. We evaluated the accuracy of our estimates by comparing to field measurements of growth, airborne LiDAR data, and independent model estimates of C accumulation. Landsat disturbance history and existing stereo archives could provide an independent empirical data source to help constrain C-cycle models in forested ecosystems and reduce our uncertainties about the North American C sink.