Estimation of Forest Biomass and Faraday Rotation Using Polarimetric L-/P-band SAR

Licentiate thesis, 2009

At present the greatest source of uncertainty in the global carbon cycle is in the terrestrial ecosystems. In order to reduce these uncertainties it is necessary to provide consistent and accurate global estimates of the biomass contained in the forests of the world. One of the most promising methods for obtaining such estimates is through polarimetric SAR backscatter measurements at L-and P-band frequencies. This thesis aims to investigate the possibility of retrieving forest biomass from L- or P-band SAR backscatter data. It also concerns SAR image calibration. Well calibrated systems are necessary to ensure that errors in estimates of physical properties, e.g. forest biomass, are not enlarged by noise or image distortions. Both system distortions, such as leakage between polarization channels, and distortions caused by the ionosphere are studied. A calibration analysis was made on data from the L-band PALSAR system onboard the Japanese ALOS satellite. The data was shown to be of good quality, especially in terms of radiometric stability and leakage between polarization channels. Following this an analysis of Faraday rotation, i.e. image distortions caused by the ionosphere, was made for the PALSAR images. It was shown that Faraday rotation can be measured and corrected in PALSAR image, at least if the Faraday rotation is small. Forest biomass retrieval algorithms for P- and L-band SAR backscatter were developed and evaluated using data from the BioSAR-I campaign conducted in southern Sweden in 2007. It was found that HV- and HH-polarized backscatter was strongly related to biomass for both Land P-band. However, the L-band backscatter was found to have decreased sensitivity to biomass for forest stands with biomass larger than about 200 tons/ha, resulting in an increased error in the biomass estimates. For P-band this problem was not seen, and it was found that the biomass on stand level could be estimated using HV- and /or HH-polarized Pband backscatter with an error of 20-25% of the mean biomass for stands with biomass ranging from 10 to 290 tons/ha.