Pléiades potentialities :

Transcription

1 GT2 Risque et Aide humanitaire Pléiades potentialities : Assessment of clearing levels for operational management of forest fires in the Maures massif Marechal D., Thierion V., Kabar B., Ayral P.-A., Salze D., Sauvagargues- Lesage S.

2 Expectations and needs Dedicated tool for automatic update of DFCI tracks clearing levels Application on the whole region of intervention (Maures Massif): quickness, automation and efficiency Need for consensus between operational actors (firemen and foresters)

3 Assessment of clearing levels in the DFCI tracks areas PHASE 1 : Are Pleiades products well adapted to operational management of clearing operations? PHASE 2 : How Pleiades products combinated with OTB enable operational management of clearing operations?

4 Assessment of clearing levels in the DFCI tracks areas PHASE 1 : Are Pleiades products well adapted to operational management of clearing operations? PHASE 2 : How Pleiades products combinated with OTB enable operational management of clearing operations?

5 Scientific outcomes of the first phase ( ) Good clearing Intermed. clearing Bad clearing

6 Scientific outcomes of the first phase ( ) First approach Tree / Ground QuickBird Panchromatic PELICAN fusion Second approach Multi-layer «Pixel» method QuickBird fusion «Object» method QuickBird fusion + Spatial analysis

7 Scientific outcomes of the first phase ( ) First approach Tree / Ground QuickBird Panchromatic PELICAN fusion Second approach " Multi-layer " «Pixel» method QuickBird fusion «Object» method QuickBird fusion 81% Bare soil Herbaceous layer Shrub layer Tree layer 96% PIXEL OBJECT

8 Scientific outcomes of the first phase ( ) Spatial Analysis : From ecological approach to operational use Tree neighbouring + Shrub layer density Spatial analysis Bad Intermed. Good

9 PHASE 1 : Are Pleiades products well adapted to operational management of clearing operations? A priori, spatial and spectral resolution allows an precise assessment of ecological types and a fortiori of the clearing levels. Situation in 2008 Quickbird DFCI / non-dfci 4 ecological types 3 clearing levels

10 2009 PHASE 1 : Are Pleiades products well adapted to operational management of clearing operations? A priori, spatial and spectral resolution allows an precise assessment of ecological types and a fortiori of the clearing levels. PHASE 2 : How Pleiades products combinated with OTB enable operational management of clearing operations? 1. Methodological validation 2. Multi-temporal clearing levels evaluation between 2006 and Specifications and prototype of dedicated tool

11 PHASE 2 : How Pleiades products combinated with OTB enable operational management of clearing operations? 1. Methodological validation EMA Students (Desbois, Dutault, Recordet, 2009) 2. Multi-temporal clearing levels evaluation between 2006 and 2008 EMA Students (Abrial, Champion et Morello, 2008 et 2009) 3. Specifications and prototype of dedicated tool Master degree internship financed by the CNES (Kabar, 2009)

12 Methodological validations Remote sensing methods validation 3 study sites (Le Laïre, Barral et Rayol) Object approach

13 2006 and 2008 images Multi-temporal analysis Rayol

14 2006 and 2008 images Multi-temporal analysis Le Laïre

15 PHASE 2 : How Pleiades products combinated with OTB enable operational management of clearing operations? 1. Methodological validation EMA Students (Desbois, Dutault, Recordet, 2009) 2. Multi-temporal clearing levels evaluation between 2006 and 2008 EMA Students (Abrial, Champion et Morello, 2008 et 2009) 3. Specifications and prototype of dedicated tool Master degree internship financed by the CNES (Kabar, 2009)

16 Pleiades simulated imagery (Quickbird ) Orfeo Toolbox ORFEO Motivate future imagery exploitation Team «Risques industriels et naturels» (LGEI) Issues: Methods Tool specifications SDIS 83 and SIVOM Management of operational clearing Do Pléiades products associated with OTB allow an operational management of clearing? 1. Analysis of existing methods (ENVI, Definiens, ArcGIS) 2. OTB implementation 3. Enhancement

17 Issue OTB Results Perspectives Why OTB?: Free and Flexible Numerous applications for dedicated tools

18 Issue OTB Results Perspectives OTB Workflow Technological Consensus OTB Object approach

19 Issue OTB Results Perspectives DFCI extraction (a) Quickbird RGB, Rayol zone. (b) Intensity-ARVI with «Mean Shift» segmentation; δs = 3, δr = 35, min size region = 8000, scale= 1. (c) Labeled image «forest» and «DFCI». (a) (b) (c). Good discrimination. Few errors of classifications. Not directly usable for multi-layer classification Buffer zone around DFCI tracks

20 Issue OTB Results Perspectives Multi-layer classification Mean Shift clustering (a) Pansharpened image (b) Clustered image B-V-R-PIR (pansharpened)-arvi with: δs = 2, δr = 30, min region size = 1, scale = 1. (c) Focus on non-clustered image. (d) Focus on clustered image

21 Issue OTB Results Perspectives Multi-layer classification Combination: R-PIR-PAN-ARVI-IC Extent : 4245,6m x 3220,2m SVM classification Kappa: 97% Bare soil Tree layer Shrub layer Herbaceous layer Non-DFCI

22 Issue OTB Results Perspectives Multi-layer classification Local control Reference: Google Earth (2006) Bare soil Tree layer Shrub layer Herbaceous layer Non-DFCI Some errors of confusion: optimisation of training sample delineation, features combination

23 Issue OTB Results Perspectives Results:. Similar quality of classification compared to previous methods. Still some imprecisions. No spatial analysis Futur works:. Real object oriented classification (use of texture, geometry and neighbouring). Correlation between spectral signatures and ecological layer. Operational validation (with operational actors). Performing spatial analysis in OTB? Dedicated tool specifications

24 Issue OTB Results Perspectives Apply OTB algorithms to the whole region of interventions First tool building (OTB) Quantitative analysis of the classification quality Spatial analysis integration Spatial analysis validation 2010 Internship Scale Multi-temporal analysis (3 images) Septembre 2009 classification Automatisation with OTB (Test) Link with clearing technics efficiency

25 Issue OTB Results Perspectives To an operational use Price, delay and programmation (1 /year) Concertation with local and regional administrations for imagery requests Zones of interventions monitoring during 3 or 4 years: ¼ of the territory of interest every year (?) Assessment of financial interests (field work vs imagery)

26 Thanks Pierre-Alain Ayral : pierre-alain.ayral@mines-ales.fr Boris Kabar : boris.kabar@mines-ales.fr Denis Maréchal : denis.marechal@mines-ales.fr David Salze : david.salze@mines-ales.fr Vincent Thierion : vincent.thierion@mines-ales.fr

27 Operational scale of validation? 1 (few m 2 ) 2 (several m 2 ) Scale of validation Remote sensing efficiency 3 (ha) Technological needs