Predicting forest inventory parameters with ICESat-2

Abstract:

Since its launch in September 2018, the Ice, Cloud and land Elevation Satellite-2 (ICESat-2) has been capturing the three-dimensional features of forests, providing data for potential estimation of a key inventory parameters and indicators of ecosystem health and function. This study investigates the predictive capability of ICESat-2 for estimating critical forest attributes and inventory parameters. Specific objectives are to: (1) Utilize ICESat-2 vegetation product data, ATL08, to estimate biomass, basal area, timber volume, and canopy cover, (2) extract vegetation parameters from corresponding ICESat-2’s geolocated photon data (ATL03) using custom processing algorithms, to estimate forest attributes, and (3) compare predictive performance of models using ATL08 parameters, with those developed from ICESat-2 data processed with custom noise filtering and photon classification algorithms. This work focuses on ICESat-2 tracks over two study sites, one within Sam Houston National Forest (SHNF) in south-east Texas, and another in the Solon Dixon Forestry Education Center (SDFEC) in southern Alabama. Both sites are predominated by pine plantations and consist of vegetation representative of the southeastern US. Methods consist of processing ICESat-2’s ATL03 data for tracks (strong beams) over SDFEC and SHNF, computing canopy metrics from processed data at the ATL08 scale and deriving corresponding ATL08 canopy metrics. Airborne lidar data collected in 2018 and 2019 over SDFEC and SHNF, as part of the U.S. Geological Survey’s 3D Elevation Program (3DEP) are used to directly estimate canopy cover and to develop relationships with field-estimated biomass, volume and basal area, to serve as a reference. Airborne lidar-estimated (canopy height, AGB, volume, basal area) and -measured parameters (canopy cover) are then utilized for building relationships with ATL08 vegetation parameters and separately, with custom-processed ICESat-2 data using linear regression models. Findings from this study serve to demonstrate the capability of ICESat-2 to characterize ecosystems and its utility for estimating critical forest attributes in support of sustainable forest management.