Beyond Radar Backscatter: Estimating Forest Structure and Biomass with Radar Interferometry and Lidar Remote Sensing

Friday, 19 December 2014
Marco Lavalle and Razi Ahmed, Jet Propulsion Laboratory, Pasadena, CA, United States
Mapping forest structure and aboveground biomass globally is a major challenge that the remote sensing community has been facing for decades. Radar backscatter is sensitive to biomass only up to a certain amount (about 150 tons/ha at L-band and 300 tons/ha at P-band), whereas lidar remote sensing is strongly limited by poor spatial coverage. In recent years radar interferometry, including its extension to polarimetric radar interferometry (PolInSAR), has emerged as a new technique to overcome the limitations of radar backscatter. The idea of PolInSAR is to use jointly interferometric and polarimetric radar techniques to separate different scattering mechanisms and retrieve the vertical structure of forests. The advantage is to map ecosystem structure continuously over large areas and independently of cloud coverage. Experiments have shown that forest height - an important proxy for biomass – can be estimated using PolInSAR with accuracy between 15% and 20% at plot level.

At AGU we will review the state-of-art of repeat-pass PolInSAR for biomass mapping, including its potential and limitations, and discuss how merging lidar data with PolInSAR data can be beneficial not only for product cross-validation but also for achieving better estimation of ecosystem properties over large areas. In particular, lidar data are expected to aid the inversion of PolInSAR models by providing (1) better identification of ground under the canopy, (2) approximate information of canopy structure in limited areas, and (3) maximum tree height useful for mapping PolInSAR temporal decorrelation. We will show our tree height and biomass maps using PolInSAR L-band JPL/UAVSAR data collected in tropical and temperate forests, and P-band ONERA/TROPISAR data acquired in French Guiana. LVIS lidar data will be used, as well as SRTM data, field measurements and inventory data to support our study. The use of two different radar frequencies and repeat-pass JPL UAVSAR data will offer also the opportunity to compare our results with the new airborne P-band ECOSAR and L-band DBSAR instruments developed at the NASA Goddard Space Flight Center.