EVALUATION OF APPROACHES TO ESTIMATING ABOVEGROUND BIOMASS IN SOUTHERN PINE FORESTS USING SIR-C DATA

A study was performed to evaluate various techniques for estimating ab oveground woody plant biomass in pine stands found in the southeastern United States, using C- and L- band multiple polarization radar image ry collected by the Shuttle Imaging Radar-C (SIR-C) system. The biomas s levels present in the test stands ranged between 0.0 and, 44.5 kg m( -2). Two SIR-C data sets were used: one collected in April, 1994, when the soil conditions were very wet and the canopy was slightly wet fro m dew and a second collected in October, 1994, when the soils and cano py were dry. During the October mission, pine needles were completely flushed and the foliar biomass was twice as great in the forest stands as in April. Four methods were evaluated to estimate total biomass: o ne including a straight multiple linear correlation between total biom ass and the various SIR-C channels; another including a ratio of the L -band HV/C-band HV channels; and two others requiring multiple steps, where linear regression equations for different stand components (heig ht, basal area, and crown or branch biomass) were used as the basis fo r estimating total biomass. lt was shown that the data collected in Oc tober (dry soil conditions) were better for estimation of biomass than the data collected in April (wet soil conditions). Overall, a multist ep approach resulted in the lowest root mean square (RMS) errors (5.91 kg m(-2)) when biomass levels were (20 kg m(-2)). For all biomass lev els, the simple regression technique resulted in the lowest RMS errors (8.1 kg m(-2)). The multiple-step approaches have the additional adva ntage of being able to provide estimates of different components of st and structure and biomass, such as average tree height, basal area, br anch biomass, canopy biomass, trunk biomass, and foliage biomass. The LHV channel is the critical element in all the biomass equations, as w ould be expected from the body of literature. The addition of other ch annels-generally, CHV or CHH-significantly improves biomass estimates, whether as a ratio or as additional terms in a regression equation. ( C) Elsevier Science Inc., 1997.