Savanna-Bio: Biomass Estimation with New Spaceborne Missions for MRV in Dry Forests and Savannas
John Armston, University of Maryland, armston@umd.edu (Presenter)
Laura Duncanson, University of Maryland, lduncans@umd.edu
Konrad Johan Wessels, George Mason University, kwessel4@gmu.edu (Presenter)
Paul Robert Siqueira, University of Massachusetts, siqueira@umass.edu
Mikhail Urbazaev, University of Maryland, urbazaev@umd.edu (Presenter)
Xiaoxuan Li, George Mason University, xli50@gmu.edu
Narayanarao Bhogapurapu, University of Massachusetts Amherst, nbhogapurapu@umass.edu
The aim of our CMS research is to develop prototype aboveground biomass and change products for selected tropical savanna regions (India, South Africa and Australia), using NASA’s GEDI and ICESat-2 spaceborne lidar data to reduce the uncertainty of SAR-based models (Sentinel-1A/1B, ALOS PALSAR-1/2, NASA/ISRO NISAR)
Here we present the current status of our project. We have evaluated the efficacy of GEDI and ICESat-2 for the estimation of vegetation height and canopy cover across global savannas. Our analysis is conducted across pilot sites in South Africa and Australia that cover dry, temperate and tropical savannas. We examine measurement error in GEDI Level-2A/2B as well as ICESat-2 ATL03/ATL08 data products by colocation and comparison with equivalent reference estimates simulated from Airborne Laser Scanning data.
Further, we have made substantive progress towards estimating canopy height by inversion of a coupled volumetric and temporal decorrelation model, using repeat-pass Sentinel-1 InSAR image pairs as input, that assumes taller trees will decorrelate more than shorter ones. This temporal decorrelation model is inverted with the Gauss-Newton algorithm, and has been updated to use GEDI observations to localize the fitting of temporal change parameters, instead of assuming scene-wide constant temporal change parameters. To date we have fit these parameters using GEDI canopy height and Sentinel-1 SAR data, to investigate if the temporal decorrelation signature of C-band data can be used for achieving results similar to those that have been demonstrated for L-band.
Our next steps include: 1) development of refined satellite lidar training datasets, 2) application of vegetation height inversion algorithm on ALOS-2 Single Look Complex data, and 3) completion of new airborne/field campaigns and derived products.
Poster: Poster_Armston_3-35_70_35.pdf
Associated Project(s):
Poster Location ID: 3-35
Presentation Type: Poster
Session: Poster Session 3
Session Date: Thu (May 11) 3:00-5:00 PM
CCE Program: TE