Generation of large-scale forest height and disturbance maps through the fusion of NISAR and GEDI along with TanDEM-X/L

Yang Lei, Universities Space Research Associations; Jet Propulsion Laboratory, yang.lei@jpl.nasa.gov (Presenter)
Robert Neil Treuhaft, JPL, California Institute of Technology, robert.treuhaft@jpl.nasa.gov
Paul Robert Siqueira, University of Massachusetts, siqueira@umass.edu
Nathan Torbick, Applied Geosolutions, ntorbick@appliedgeosolutions.com
Richard Lucas, University of New South Wales, richard.lucas@unsw.edu.au
Michael Keller, USDA Forest Service @ JPL, michael.m.keller@jpl.nasa.gov
Michael Schmidt, Remote Sensing Centre, Brisbane, Australia, michael.schmidt@dsiti.qld.gov.au
Mark Ducey, University of New Hampshire, mark.ducey@unh.edu
William A. Salas, Applied GeoSolutions, wsalas@agsemail.com

Large-scale products of forest height and disturbance are essential for understanding the global carbon distribution as well as its changes in response to natural events and human activities. Regarding this scientific need, both NASA’s GEDI and NASA-ISRO’s NISAR are going to be launched in the 2018-2021 timeframe in parallel with DLR’s current TanDEM-X and/or the proposed TanDEM-L, which provides a lot of potential for global ecosystem mapping. A new simple and efficient method of forest height mapping has been developed for combining spaceborne repeat-pass InSAR and lidar missions (e.g. NISAR and GEDI) which estimates temporal decorrelation parameters of repeat-pass InSAR and uses the lidar data as training samples. An open-access Python-based software has been developed for automated processing. As a result, a mosaic of forest height was generated for US states of Maine and New Hampshire (11.6 million ha) using JAXA’s ALOS-1 and ALOS-2 HV-pol InSAR data and a small piece of lidar training samples (44,000 ha) with the height estimates validated against airborne lidar and field inventory data over both flat and mountainous areas. In addition, through estimating and correcting for the temporal decorrelation effects in the spaceborne repeat-pass InSAR coherence data and also utilizing the spaceborne single-pass InSAR phase data, forest disturbance such as selective logging is not only detected but also quantified in subtropical forests of Australia using ALOS-1 HH-pol InSAR data (validated against NASA’s Landsat), as well as in tropics of Brazil using TanDEM-X and ALOS-2 HH-pol InSAR data (validated against field inventory data). The operational simplicity and efficiency make these methods a potential observing/processing prototype for the fusion of NISAR, GEDI and TanDEM-X/L.

Poster Location ID: 101

Session Assigned: Vegetation Dynamics and Distribution