RETRIEVAL OF FOREST BIOMASS FROM SAR DATA

In recent years, there has been an increasing interest in the use of r adar data for observations of forest ecosystems. In particular, it was shown that the intensity of SAR images at L band was proportional to the forest aboveground biomass. More recently, the analysis of NASA/JP L SAR data over the Landes forest (south-west France) revealed strong correlation between P-band backscatter and biomass of pine trees and r elated characteristics (tree age, height, diameter). Also, similar rel ations have been obtained on two different forests (maritime pine at L andes forest, France, and loblolly pine at Duke forest, U.S.A.). This paper presents a step further in the understanding of the observations , using theoretical modelling applied to calibrated SAR data to explai n the radar backscatter from the forest canopy under study. The study is presented at P band, which was found to be an optimal frequency ban d for forest observations. It was found that the HH return is physical ly related to both trunk and crown biomass, whereas VV and particularl y HV returns are linked to crown biomass. Through allometric equations , relations between radar backscatter intensities and most of the fore st characteristics can thus be explained. The validated model has been used in a next step to simulate a diversity of forest and observation conditions. The results have been used for a preliminary assessment o f the possibility of inverting SAR data into forest biomass.