Monitoring changes in above-ground biomass in loblolly pine forests using multichannel synthetic aperture radar data.

A study was conducted to evaluate using synthetic aperture radar (SAR) for estimating aboveground biomass in loblolly pine (Pinus taeda L.) forests. The data set for this experiment was a multiple-frequency (C-, L- and P-band), polarimetric SAR data set collected by the NASA/JPL AIRSAR System over the Duke University Research Forest located near Durham, North Carolina. In addition to the SAR data set, a set of ground measurements were collected to describe the tree geometry and biomass characteristics from 59 different stands consisting principally of loblolly pine within the Duke Forest. The aboveground, dry weight woody biomass in these test stands ranges from $$50 kg-m$\sp2$. The first analysis performed on this data set was to produce algorithms to estimate both dry and wet weight biomasses for each of the test stands, and to distribute this biomass amongst various tree components (e.g., boles, branches, and needles/leaves) as well as the different layers within the tree canopy (e.g., canopy, subcanopy and understory) in order to better relate biomass to the radar backscattering measurements. This was accomplished by development of allometric equations to estimate biomass for individual trees, from which stand estimates on an aerial basis were derived. The biomass estimates were then statistically correlated with radar backscatter ($\sigma\sp\circ$) measurements derived from the SAR data set. It was found that $\sigma\sp\circ$ at a variety of radar frequencies (P, L, and C-bands) and linear-polarization combinations (HH, HV, and VV) were significantly correlated (at a level of significance of p = 0.001) to either individual biomass components (e.g., bole biomass, branch biomass, needle/leaf biomass, etc.) or multiple combinations of these components. While the correlations were significant at all linear polarizations at L- and P-bands, they were only significant in the cross-polarized channel at C-band. Finally, a two-step method was developed to estimate aboveground biomass from multichannel SAR data. The error estimate (coefficient of variation) for the various biomass components ranged between 16% and 27%. It is concluded that the pine stand biomasses (dry weight) for which these algorithms are valid range between 15 and 40 kg-m$\sp{2}$.Ph.D.Biological SciencesEcologyForestryNuclear physics and radiationPure SciencesUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/128999/2/9308352.pd