Monitoring Tropical forests with L-band radar: lessons from Indonesian Peat Swamps Matt Waldram, Sue Page, Kevin Tansey www.le.ac.uk Geography Department.

Slides:



Advertisements
Similar presentations
Lab for Remote Sensing Hydrology and Spatial Modeling Dept. of Bioenvironmental Systems Engineering, NTU Satellite Remote Sensing for Land-Use/Land-Cover.
Advertisements

Microwave remote sensing applications and it’s use in Vietnam
REDD+ Methodologies for Regional and Local Land- cover Thelma Krug Co-Chair of the IPCC Task Force on National Greenhouse Gas Inventories Head of INPE´s.
Beyond Spectral and Spatial data: Exploring other domains of information GEOG3010 Remote Sensing and Image Processing Lewis RSU.
Selected results of FoodSat research … Food: what’s where and how much is there? 2 Topics: Exploring a New Approach to Prepare Small-Scale Land Use Maps.
Predicting and mapping biomass using remote sensing and GIS techniques; a case of sugarcane in Mumias Kenya Odhiambo J.O, Wayumba G, Inima A, Omuto C.T,
Time Series Fusion of Optical and Radar Imagery for Improved Monitoring of Activity Data, and Uncertainty Analysis of Emission Factors for Estimation of.
Toward Near Real Time Forest Fire Monitoring in Thailand Honda Kiyoshi and Veerachai Tanpipat Space Technology Applications and Research, School of Advanced.
Estimating Anthropogenic Influence in Tropical Forests Using Charcoal Introduction Jessica Del Greco Advisors: Crystal H. McMichael, Earth System Research.
Forest Monitoring of the Congo Basin using Synthetic Aperture Radar (SAR) James Wheeler PhD Student Supervisors: Dr. Kevin Tansey,
Operational multi-sensor design for forest carbon monitoring to support REDD+ in Kalimantan, Indonesia Stephen Hagen (Applied GeoSolutions) NASA Carbon.
Estimating forest structure in wetlands using multitemporal SAR by Philip A. Townsend Neal Simpson ES 5053 Final Project.
Radar, Lidar and Vegetation Structure. Greg Asner TED Talk.
ReCover for REDD and sustainable forest management EU ReCover project: Remote sensing services to support REDD and sustainable forest management in Fiji.
Multiple Criteria for Evaluating Land Cover Classification Algorithms Summary of a paper by R.S. DeFries and Jonathan Cheung-Wai Chan April, 2000 Remote.
Module 2.5 Estimation of carbon emissions from deforestation and forest degradation REDD+ training materials by GOFC-GOLD, Wageningen University, World.
Overview of Biomass Mapping The Woods Hole Research Center Alessandro Baccini, Wayne Walker and Ned Horning November 8 – 12, Samarinda, Indonesia.
Akira Kato 1, Manabu Watanabe 2, Tatsuaki, Kobayashi 1, Yoshio Yamaguchi 3,and Joji Iisaka 4 1 Graduate School of Horticulture, Chiba University, Japan.
Compton Tucker, GSFC Sassan Satchi, JPL Jeff Masek, GSFC Rama Nemani, ARC Diane Wickland, HQ Terrestrial Biomass Pilot Product: Estimating Biomass and.
Module 2.1 Monitoring activity data for forests using remote sensing REDD+ training materials by GOFC-GOLD, Wageningen University, World Bank FCPF 1 Module.
Principles of Remote Sensing 10: RADAR 3 Applications of imaging RADAR Dr. Mathias (Mat) Disney UCL Geography Office: 113, Pearson Building Tel:
Mapping Fire Scars in Global Boreal Forests Using Imaging Radar Data Written By: L.L. Bourgeau-Chavez, E.S. Kasischke, S. Brunzell, J.P. Mudd, and M. Tukman.
Dual Polarised Entropy/alpha Decomposition and Coherence Optimisation for Improved Forest Height Mapping Z-S Zhou, P. Caccetta, E. Lehmann, A. Held – CSIRO,
Moving on From Experimental Approaches to Advancing National Systems for Measuring and Monitoring Forest Degradation Across Asia Moving on From Experimental.
Microwave Remote Sensing Group 1 P. Pampaloni Microwave Remote Sensing Group (MRSG) Institute of Applied Physics -CNR, Florence, Italy Microwave remote.
Proposed NCEO Land Core posts 1.CEH: Hydrological aspects and coupling with atmosphere 2.KCL: Fire emissions and their use in atmospheric monitoring, modelling.
Peat and GHG Science- KFCP REDD+ Demonstration Grahame Applegate Indonesia- Australia Forest Carbon Partnership International Indonesia Peatland Conversation.
Dr. Florian Siegert, RSS - Remote Sensing Solutions GmbH, München The role of Remote sensing in monitoring land cover and impacts on peatlands.
Comparison of FBD and ScanSAR deforestation detections Martin Whittle (a), Shaun Quegan (a),Kokok Yulianto (b) and Yumiko Uryu (b) (a) CTCD, Department.
Development of indicators of fire severity based on time series of SPOT VGT data Stefaan Lhermitte, Jan van Aardt, Pol Coppin Department Biosystems Modeling,
UN-FCCC Bonn meeting June 2009 Peatlands, carbon and climate change
DOCUMENT OVERVIEW Title: Fully Polarimetric Airborne SAR and ERS SAR Observations of Snow: Implications For Selection of ENVISAT ASAR Modes Journal: International.
20 years of Tropical Forest Research at the University of Leicester Sue Page, Kevin Tansey, Heiko Balzter, Agata Hoscilo, Matthew Waldram,
Long Time Span Interferograms and Effects of Snow Cover on Interferometric Phase at L-Band Khalid A. Soofi (ConocoPhillips), David Sandwell (UCSD, SCRIPPS)
Translation to the New TCO Panel Beverly Law Prof. Global Change Forest Science Science Chair, AmeriFlux Network Oregon State University.
Development and evaluation of Passive Microwave SWE retrieval equations for mountainous area Naoki Mizukami.
Michigan Tech Research Institute (MTRI)  Michigan Technological University 3600 Green Court, Suite 100  Ann Arbor, MI (734) – Phone 
PREFER 1 st Annual Review Meeting, 5-6 Dec 2013, Milano-Italy PREFER WP 3.2, Task 7.
Remote Sensing Realities | June 2008 Remote Sensing Realities.
Soil moisture estimates over Niger from satellite sensors (T. Pellarin, M. Zribi)
THINKING beyond the canopy Removing technical barriers to include tropical peatlands in the REDD+ mechanism Daniel Murdiyarso, Kristell Hergoualc’h and.
Calibration/Validation Efforts at Calibration/Validation Efforts at UPRM Hamed Parsiani, Electrical & Computer Engineering Department University of Puerto.
Niels Wielaard Marcela Quinones Comparison SarVision mid-2007 LULC map and selected reference maps CKPP - Ex Mega Rice Area Prepared for.
Improving carbon cycle models with radar retrievals of forest biomass data Mathew Williams, Tim Hill and Casey Ryan School of GeoSciences, University of.
DN Ordinate Length DN Difference Estimating forest structure in tropical forested sites.
FSU Jena – Department of Earth Observation CREATION OF LARGE AREA FOREST BIOMASS MAPS FOR NE CHINA USING ERS-1/2 TANDEM COHERENCE Oliver Cartus (1), Christiane.
SDCG-4, Caltech, CA, USA 4 th -6 th September 2013 Author/Presenter Indonesia MRV & Reporting Status & Related Space.
Christian N. Koyama University of Cologne IGARSS 2011 Vancouver, July 26 Soil Moisture Retrieval Under Vegetation Using Dual Polarized PALSAR Data Christian.
Comparison of L and P band radar time series for the monitoring of Sahelian area P.-L. Frison, G. Mercier, E. Mougin, P. Hiernaux.
Scaling Up Above Ground Live Biomass From Plot Data to Amazon Landscape Sassan S. Saatchi NASA/Jet Propulsion Laboratory California Institute of Technology.
biomass TO OBSERVE FOREST BIOMASS
0 Riparian Zone Health Project Agriculture and Agri-Food Canada Grant S. Wiseman, BS.c, MSc. World Congress of Agroforestry Nairobi, Kenya August 23-28,
Beyond Spectral and Spatial data: Exploring other domains of information: 3 GEOG3010 Remote Sensing and Image Processing Lewis RSU.
Claire Plagge Research & Discover Internship Summer 2008
Measurement of a Temporal Sequence Of DInSAR Phase Changes Due to Soil Moisture Variations Keith Morrison 1, John Bennett 2, Matt Nolan 3, and Raghav Menon.
The Pacific GIS/RS User Conference Suva, Fiji Island, November 2012 Sharon R. Boe, SPC/GIZ-SOPAC ) SPC/GIZ Regional REDD+ Project:
Time Dependent Mining- Induced Subsidence Measured by DInSAR Jessica M. Wempen 7/31/2014 Michael K. McCarter 1.
Using SAR Intensity and Coherence to Detect A Moorland Wildfire Scar.
Passive Microwave Remote Sensing
Module 2.8 Overview and status of evolving technologies REDD+ training materials by GOFC-GOLD, Wageningen University, World Bank FCPF 1 Module 2.8 Overview.
IFSAR and terrestrial LIDAR for vegetation study in Sonora, Texas
Active Microwave Remote Sensing
PADMA ALEKHYA V V L, SURAJ REDDY R, RAJASHEKAR G & JHA C S
Identifying Forest Change with SAR
Are emission reductions from peatlands MRV-able
UN-FCCC Bonn meeting June 2009 Peatlands, carbon and climate change
Objectives Using a time series of data from radar sensors to detect and measure forest changes Combining different types of data, including: Multi polarisations.
Radar backscattering measurements of paddy rice field using L, C and X-band polarimetric scatterometer ISRS 2007.
(L, C and X) and Full-polarization
Assessing woody carbon stocks in Miombo woodlands of Mozambique (see map for location). We used multiscale sampling of vegetation cover (leaf area index)
Presentation transcript:

Monitoring Tropical forests with L-band radar: lessons from Indonesian Peat Swamps Matt Waldram, Sue Page, Kevin Tansey Geography Department

Why do we need to monitor forests? Conservation of forest carbon (e.g. UN REDD+) becoming a reality? Need for frequent estimates of forest/carbon loss to support MRV activities. Radar remote sensing has many advantages particularly in tropical areas and numbers of radar (and other) satellites looks set to increase.

ALOS PALSAR L-Band Synthetic Aperture Radar in Fine Beam Dual Polarisation mode (no clouds!). Dual channel FBD (3, with interferometric coherence) & 25m pixel size (2*9 multilooks). Field data includes plot based estimates of above ground biomass, dead fuel load and water table measurements (largely collected by KFCP REDD demonstration project). Methods

Study area 1: Ex Mega Rice Project and surrounding areas in Central Kalimantan, Indonesia. Covered by 8 (3 strips) PALSAR scenes & ~12 dates over 4 years. Very seasonal with long dry season. Mode of deforestation is fire. 28,500km % of Borneo by area (roughly).

Study area 2: Kamp peninsula and surrounding areas Riau, Indonesia. Covered by 5 PALSAR scenes (2 strips) & 10 dates over 4 years. Aseasonal with brief dry season. Mode of deforestation is clear felling for plantations.

From Mitchard ET, Saatchi SS, Woodhouse IH, et al. Using satellite radar backscatter to predict above-ground woody biomass: A consistent relationship across four different African landscapes. Geophysical Research Letters. 2009;36(23) Much effort had gone in to producing backscatter : biomass calibration curves. In order to provide direct estimates of biomass (and hence carbon). Most radar/biomass studies done using one time period (mapping) or two separate time periods for (change detection). But I am interested in monitoring, i.e. having multiple time periods (i.e. time as a continuous variable). How does the radar backscatter change with time in response to disturbance?

So when temporal backscatter variation is included the picture becomes much more complicated! Signal over intact forest is quite stable (saturation of radar signal). What is driving signal in change in degraded areas? Is it noise or is there information contained within the temporal behaviour?

Peatland water table measured via a network of dipwells (data courtesy of KFCP project). Non-linear relationships exist between water table depth and radar backscatter and coherence but only in low biomass (degraded) areas. Changes in backscatter caused by increasing dielectric of wet peat ( 0m).

Actual relationship probably between surface soil moisture and radar variables. Surface soil moisture linearly related to water table at depths down to ~-30cm at which point they become decoupled. This correlation allows us to model water table depth and to predict water table depth from PALSAR signals (within limits). R 2 =0.5 for excluded points

What temporal signal does forest disturbance produce? In Kalimantan forest loss occurs via fire. This produces a very different signal response to other modes of deforestation, e.g. clear cut logging in kampar.

HH HV Kalimantan (fire)Kampar (plantation logging) Temporal backscatter signatures of two modes of forest loss, (4 year PALSAR time series)

Long term changes in radar signal in response to disturbance. Generated using landsat derived burn history map and extracting radar signal over burn scars of different age.

We need statistical techniques to map spatial and temporal patterns of radar signal variation. Generally I have found the literature to be lacking, particularly when it is considered. Need to develop new techniques able to be applied to large data sets. One technique I have looked at is empirical orthogonal functions (EOFs).

R(EOF1),G(EOF2),B(EOF3) = 94.4% Red=EOF1, 63.5% Green=EOF2, 22.7% Blue=EOF3, 8.2%

R(EOF1),G(EOF2),B(EOF3) =92.0% Red=EOF1, 60.1% Green=EOF2, 18.4% Blue=EOF3, 13.5%

Looking at a single data PALSAR scene could be very misleading! Including temporal trends can be very confusing but information is contained with in temporal variation. Temporal patterns in the radar signal are effected by: Mode of forest loss (fire, clear cutting plantations) Moisture regime/rainfall seasonality Structure of vegetation type. Bottom up approach in contrast to top down approach (e.g. GEO FCT). Conclusions & thoughts…..