Tropical forest measurement by interferometric height modeling and P-band radar backscatter

Abstract

A new approach to tropical forest biomass monitoring with airborne interferometric X and P-band synthetic aperture radar (SAR) data is presented. Forest height, basal area, and aboveground biomass are modeled from remote sensing data for a study site in the Brazilian Amazon, Radar data quality has improved: A novel digital model of vegetation height from X- and P-band interferometry is available along with the usual P-band backscatter information. The digital vegetation height model is derived from the interferometric surface models of the ground (from P-band) and the forest canopy (from X-band), The difference between the surface models is called "interferometric height," and is used as a measure of vegetation height, Interferometric height is shown to relate to a subset of the forest trees that changes according to the forest successional stages. The suitability of radar backscatter and interferometric height as a means for forest and biomass monitoring was explored by relating forest parameters as measured in the field to remote sensing data. Basal area and biomass were related to radar backscatter with limited precision of r2 = 0.19 and r2 = 0.34, respectively. Mean forest height is shown to relate to interferometric height with good precision (r 2 = 0.83, RMSE = 4.1 m), A statistical model for forest biomass as a function of both P-band backscatter and interferometric height information not only arrives at high values of precision (with r2 = 0.89 and a RMSE from cross-validation of only 46.1 t/ha), but also overcomes the well-known issue of backscatter saturation. This research shows that tropical forest biomass can be quantified and mapped over large areas for a range of forest structures with reasonably tight and similar errors. Copyright © 2005 by the Society of American Foresters.