Speaker
Description
The global dry tropics are currently the largest, most sensitive, and fastest increasing component of the land carbon sink, but proper characterization of their role requires accurate estimates of the forest processes occurring within this region. L-band Synthetic Aperture Radar (SAR) observations are currently our best option to consistently map these processes: i) they are available globally since 2007 (JAXA ALOS-1 and ALOS-2), and ii) are sensitive to forest structure, such as aboveground biomass up to ~ 100 t/ha.
Here we use high-quality ground observations from Africa, Australia and Latin America to understand the impact of forest structure and site characteristics on L-band radar data. Fully-polarimetric L-band SAR data from ALOS-1 PALSAR-1 and ALOS-2 PALSAR-2 sensors over the ground sites were interpreted in terms of scattering mechanisms (surface scattering, volume scattering, double bounce) inferred using the Freeman & Durden, van Zyl and Cloude & Pottier polarimetric decompositions. Within the forest sites, volume scattering is clearly identified as the dominant mechanism over the wooded vegetation types though it is mixed with a weaker return from surface scattering and, in most cases, a much weaker return from double bounce. Double bounce is usually small, probably because of weaker scatter from dry soils and the prevalence of multi-stem tree types. However, across the landscape there is much greater variation, with surface scattering dominating in non-wooded areas and a more even balance of surface and volume scattering in mixed areas. Volume scattering is sensitive to above-ground biomass density (AGBD) up to ~100 t ha-1, after which saturation occurs, but the response to AGBD is strongly affected by stocking density (much less so by average tree biomass).
We will present these results in the context of structural equation modelling and discuss their implications for forest information recovery using only the more widely available ALOS dual-pol data.
