Towards global estimates of burnt area from satellite imagery for use by the atmospheric science...
-
Upload
bridget-norton -
Category
Documents
-
view
217 -
download
5
Transcript of Towards global estimates of burnt area from satellite imagery for use by the atmospheric science...
Towards global estimates of burnt area from satellite imagery for use by the atmospheric science communityGrégoire J-M.1*, K. Tansey1*, I. Marengo1, L. Boschetti1, P.A. Brivio2, D. Ershov3, S. Flasse4, R. Fraser5, D. Graetz6, M. Maggi1, P. Peduzzi7, J.M.N. Pereira8, J. Silva9, A. Sousa10, D. Stroppiana1
1* corresponding authors: Global Vegetation Unit, Joint Research Centre, Italy [[email protected] ; [email protected]]2 IREA-CNR, Milan, Italy; 3 IFI, Moscow , Russia; 4 FC, Maidstone, UK; 5 CCRS, Ottawa, Canada; 6 CSIRO-EOC, Canberra, Australia; 7 UNEP, Geneva, Switzerland; 8 TRI, Lisbon, Portugal; 9 UTL, Lisbon, Portugal; 10 UE, Evora, Portugal
The GBA2000 initiative
References• Boschetti, L., S. Flasse, A. Jacques de Dixmude and S. Trigg, 2001, A multitemporal change detection algorithm for the monitoring of burnt areas with SPOT-Vegetation data. In L. Bruzzone and P. Smith, (Eds.), Analysis of Multi-temporal Remote Sensing Images, Singapore: World Scientific (in press).• Brivio, P.A., M. Maggi, E. Binaghi, I. Gallo, 2002, Mapping burned surfaces in Sub-Saharan Africa based on multi-temporal neural classification, IJRS, submitted.• Ershov, D.V., and Novik, V.P., 2001, Features of burnt area mapping in forest of Siberia using SPOT S1-VGT data. GOFC Fire Satellite Product Validation Workshop, 9 July 2001 (Lisbon).• Fraser, R.H., R. Fernandes, and R. Latifovic, 2002, Multi-temporal mapping of burned forest over Canada using satellite-based change metrics, Geocarto International, submitted.• Grégoire, J-M., K. Tansey, and J.M.N. Silva, 2002, The GBA2000 initiative: Developing a global burned area database from SPOT-VEGETATION imagery, IJRS, in press.• Hansen, M., DeFries, R., Townshend, J. R. G. and Sohlberg, R., 2000, Global land cover classification at 1km resolution using a decision tree classifier, IJRS, Vol. 21, pp. 1331-1365.• Silva, J.M.N., J. M.C. Pereira, A. I. Cabral, A. C. L. Sá, M. J.P. Vasconcelos, B. Mota, and J-M. Grégoire, 2002, The area burned in southern Africa during the 2000 dry season, JGR, submitted.•Stroppiana, D., K. Tansey, J-M. Grégoire, and J.M.C. Pereira, 2002, An algorithm for mapping burnt areas in Australia using SPOT-VEGETATION data. IEEE Trans. on Geoscience & R S, submitted.•Tansey, K., 2002, Implementation of regional burnt area algorithms for GBA2000, Publication of the European Commission, in preparation, April 2002.
to provide the science community with quantitative informationon the areas burnt globally for the major biomes during the year 2000
GOFC/GOLD Fire & IGBP-IGAC/BIBEX workshopImproving global estimates of atmospheric emissions from biomass burning
University of Maryland – July 17-19 2002
Perspectives
Combine GBA2000 product with other global datasetsGLOBSCAR (ESA) and MODIS (NASA) products
Apply GBA processor to other time series of SPOT-VGT data2001, 2002
Develop research in the field of fire intensity and burning efficiency
Accuracy assessment
Performance of the regional algorithm(s)linear regression analysis to compare the burned area estimates derived fromLandsat ETM+ data and the map derived from the SPOT-VGT-S1 for limited test-sites
Accuracy assessment of the global burnt area map
Check project status Select a geographical window
on June 1st 2002, burnt area maps completed forCanada, USA down to 30o South, Europe, Africa,Asia down to 30o South and Australia
the globe has been divided into 6 regions
Download the products for the region and period of interestIllustrative results for window D (sub-saharan Africa) are shown below
Map products: GIS binary files (BSQ format in geographic projection) with file headers for direct importing into ArcInfo (GRID) and/or ENVI, ERDAS...
Statistics on a country basis Quantitative figures are provided to support
country reporting assessment of seasonality land cover – land use studies
Statistics on a grid cell basisA global grid at 1o resolution has beendesigned.
Areas burnt per month and per typeof vegetation cover are providedfor each cell
Advanced products
Latitudinal stratification at continental level Density map at continental level
Product definition A binary map (burnt-unburnt) Monthly non accumulative global burn scar map, at 1 km resolution Statistics of burnt areas Area burnt in 2000 per month and per main types of vegetation cover
Map products & statistics available on the web
Partner institutions in GBA2000 and test-sitesPartner institution and host country Test-siteCanadian Centre for Remote Sensing (CCRS), Canada CanadaNatural Resource Institute (NRI), UK Mexico, Namibia, BotswanaConsiglio Nazionale delle Ricerche (CNR), Italy Africa (northern hemisphere)Universidade Técnica de Lisboa (UTL), Portugal Mozambique, Brazil, Ib. PeninsulaInternational Forest Institute (IFI), Russia Russia (European and East Siberia)CSIRO Earth Observation Centre (EOC), Australia AustraliaJoint Research Centre (JRC), Italy Australia, South and Southeast AsiaUNEP-DEWA (Switzerland), UN development Internet Map Server
General approach Networking: co-investigators from Australia, Canada, EC, Italy, Portugal, Russia, UK, UN Development of regional algorithms: processing SPOT-VGT imagery to burnt area maps Integration of the regional algorithms into a global processing chain at JRC Overall coordination: JRC
Six regional algorithms developed to process SPOT-VGT-S1Algorithm Input data Classification approachBoschetti et al. Single date images Temporal change in the NIR & SWIRBrivio et al. Single date images Multi-Layer Perceptron (MLP) classifier contextual analysisErshov & Novik Single date images Temporal change in the NIR & SWIRFraser et al. 10-day composites Multiple logistic regression modelSilva et al. Monthly composites Linear Discriminant Analysis of temporal change of six variablesStroppiana et al. 10-day composites Classification trees & temporal change in the NIR and SWIR
The global processing chain
Input
6.6 Gb/day(VGT-S1)
Pre-processingMasking for . extreme view angle . SWIR saturation . non-vegetated surf. . cloud, shadow, snow
ProcessingTemporal compositingForest-non forest maskBurnt area algorithms
Post-processingAlgorithm dependentMosaicking to createglobal product
JRC web site http://www.gvm.sai.jrc.it/fire/gba2000_website/index.htm
UNEP-GRID http://www.grid.unep.ch/activities/earlywarning/preview/ims/gba/
Accessing the GBA2000 products
% of 1x1 deg. cell burnt in the year 2000
Development of burnt areas in Central AfricaSPOT-VGT-S1 imagery. November 2nd, 7th, 12th and 17th 1999 (left to right)