Forest Biomass Estimation, Australia

Transcription

1 Forest Biomass Estimation, Australia 1 Richard Lucas, 2 John Armston, 1 Peter Bunting, 1 Daniel Clewley, 2 Peter Scarth, 2 Michael Schmidt, 2 Arnon Accad, 3 Paul Siqueira and 3 Yang Lei 1 Institute of Geography and Earth Sciences, Aberystwyth University, Aberystwyth, Ceredigion,Wales, SY23 3DB, Wales, UK Tel: , Fax: , rml@aber.ac.uk 2 Joint Remote Sensing Research Program, School of Geography, Planning and Environmental Management, The University of Queensland, Remote Sensing Centre, Queensland Department of Environment and Resource Management, Brisbane, Queensland, Australia 3 Univesity of Massachusetts, Amherst, US. JAXA K&C Meeting 9 12 th April, 2013

2 The Injune Landscape Collaborative Project, Queensland, Australia Supporting statewide to national characterisation, mapping and monitoring of woody vegetation At multiple scales using multi sensor data Algorithm and method development (e.g., retrieval of biomass, structure, species composition and change) Understanding ecosystem response to human induced and natural change Tree to landscape AIRSAR C, L and P band

3 Tree height (LiDAR) Tree to Stand Level Products Leaf Species (Hyperspectral) Branch Location and density of stems (LiDAR) Trunk Stand-level biomass Total Integration of LiDAR and CASI

4 Tree mortality and loss of understorey through fire: Comparison of 2000 and 2009 LiDAR

5 Ground truth data for estimating AGB Brigalow Forest Regrowth Biomass library sites (2781 plots) Tall Closed Rainforest Open Callitris Forest Eucalypt Woodland Low Acacia Woodland

6 Generation of image mosaics Annual mosaics from 2007 to 2009 Used strip dates captured under driest conditions Mosaic for 2010 compromised by wet conditions

7 Retrieval of AGB ALOS PALSAR and beyond 140 E 145 E 150 E 15 S 15 S 25 S 25 S 20 S 20 S 10 S 10 S Queensland 500 km 140 E 145 E Above Ground Biomass, 50 m, 2009 (dry mosaic) Landsat derived FPC (25 m) 150 E

8 Persistent Green Fraction, Australia Landsat-derived max. min Not per mask Landsat spectral visible, near infrared and shortwave infrared data

9 Potential for Ecosystem Classification

10 National Segmentation Northern Qld HH HV PG

11 National Segmentation Southern Vic HH HV PG

12 Forest Structure ICESat GLAS/ Segmentation of L-band and Landsat FPC Forest Stand Height

13 14 12 y = 1.01x R 2 = 0.60 ICSat Mean Heig Lidar Mean Height Comparison with ICESAT y = 1.08x R 2 = ICSat Heigh ICESat derived height map Lidar Height Comparison with airborne LiDAR

14 Structural Classification based on Landsat FPC, ALOS PALSAR L band HH and HV and ICESat. Segmentation based on FPC and L band HH/HV (40 classes) Similar vertical vegetation profiles for each class (e.g., 17)

15 Forest Growth Stage Mapping Differentiation of early regrowth and remnant forest Landsat FPC and ALOS PALSAR L-band HH and HV (RGB)

16 Composites of Landsat FPC, L band HH and HV Early Regrowth Recovery from fire Mature pine

17 ICESAT Profiles, Injune Early Regrowth Recovery from fire Mature pine

18 Characteristics of Remnant (undisturbed forest): Different Regional Ecosystems Brigalow Belt Bioregion a b FPC (%) HV ( 0 ; db) HH ( 0 ; db) Regional Ecosystem

19 Integration of ALOS PALSAR and Landsat derived FPC Forest Growth Stage: Brigalow Belt Bioregion Regional Ecosystem Mapping Growth stage map

20 Approach to classifying mangroves Define extent of mangroves Existing data layers Landsat FPC Direct classification based on ALOS Separate low from high mangroves SRTM, Tandem X Definition of height locally variable (e.g.., 10 m) Separate high mangroves with/without prop root systems Assign all remaining objects to low mangroves Assign biomass classes (e.g., using relationships with L band HV; upward & downward trends)

21 Approach to Biomass Estimation ALOS PALSAR 25 m mosaics Integration with Landsat derived Persistent Green Cover Based on high density time series of Landsat sensor data. Facilitates detection of dead forest stands Segmentation using ALOS HH and HV and Landsat derived Cover Extrapolation of vertical vegetation profiles (ICESAT) Combinations of all datasets to estimate biomass Approach in development Consideration to different forest types (mangroves, regrowth) Calibration/validation achieved through reference to ground based estimates and LiDAR Spatial maps of uncertainty and variability. Temporal ALOS PALSAR mosaics potentially used to refine biomass estimates as well as detecting change. Links with MODIS time series Links to biodiversity/structure