1. QUEST, Exeter 27-28 October 2005 Stephen Plummer (IGBP@ESA), Olivier Arino (ESA), Franck Ranera and Muriel Simon (SERCO) Kevin Tansey (Univ. Leicester), Luigi Boschetti (UMD), H. Eva (JRC) and Freddy Fierens (VITO Consortium) THE GLOBCARBON INITIATIVE: Populating the Earth with Burned Area

2. QUEST, Exeter 27-28 October 2005 General User Needs For Atmospheric Chemistry and Dynamic Global Vegetation Models:  There is a particular need for information on vegetation amount (ideally biomass or leaf area index), area burned, and vegetation temporal variability.  These should be global, in a consistent format, and all data products should be available from one place.  Consistency is more important than outright accuracy (within limits).  The products should be multi-annual with 5 years being the minimum but incorporating both average and extreme conditions e.g. El Ninõ.  Products should come with spatial heterogeneity information ideally at the highest available resolution.  The spatial resolution requirements are 0.5°, 0.25° and 10km.  The temporal resolution initially on a time step of 1 month but better higher, possibly bi-weekly.

3. QUEST, Exeter 27-28 October 2005 develop a service quasi-independent of the original Earth Observation source. focus on a system to estimate:  Burned area  f APAR and LAI  Vegetation growth cycle cover six complete years: 1998 to 2003 (now up to 2007) cover VEGETATION, ATSR-2, ENVISAT (AATSR, MERIS) be applicable to existing archives and future satellite systems be available at resolutions of ¼, ½ degree and 10km with statistics build on the existing research experience Objectives

4. QUEST, Exeter 27-28 October 2005 GLOBCARBON uses the experience of these and some of the algorithms to produce a single burned area product – multi-annually. uses revised versions of algorithms it has associated with it confidence information (detection confidence from individual algorithms plus collocation with available active fire products). Burned Area - History GBA-2000 Globscar Year 2000 – two independent demonstrators of global burned area: GLOBSCAR and GBA-2000

5. QUEST, Exeter 27-28 October 2005 Original = 6 regional algorithms GLOBCARBON = 1 global and 2 regional algorithms it has associated with it confidence information (detection confidence from individual algorithms) Burned Area - Approach GBA-2000Globscar Results are merged to one product Collocation with available active fire products improves confidence Original = 2 global algorithms and burn when both agree. GLOBCARBON = each algorithm and sub-parts given a probability. The resulting probability determines occurrence of a pixel as burned (confidence information)

6. QUEST, Exeter 27-28 October 2005 Results – 1 km (Madagascar) July 1998 63-74% 75-87%88-100% Confidence Rating Index (CRI)

7. QUEST, Exeter 27-28 October 2005 Results – 1 km (Madagascar) July 1998 GLOBSCAR onlyGBA onlyBoth algorithms Algorithm Detection (GLOBSCAR, GBA, Both)

8. QUEST, Exeter 27-28 October 2005 Results – 1 km (Italy) July 1998 63-74% 75-87%88-100% Confidence Rating Index (CRI)

9. QUEST, Exeter 27-28 October 2005 Results – 1 km (Italy) July 1998 GLOBSCAR onlyGBA onlyBoth algorithms Algorithm Detection (GLOBSCAR, GBA, Both)

10. QUEST, Exeter 27-28 October 2005 Results – 1 km (Angola) July 1998 63-74% 75-87%88-100% Confidence Rating Index (CRI)

11. QUEST, Exeter 27-28 October 2005 Results – 1 km (Angola) July 1998 GLOBSCAR onlyGBA only Both algorithms Algorithm Detection (GLOBSCAR, GBA, Both)

12. QUEST, Exeter 27-28 October 2005 MODIS Comparison – 1 km July 2000 MODIS GLOBCARBON

13. QUEST, Exeter 27-28 October 2005 MODIS Comparison – 1 km Sept 2000 MODIS GLOBCARBON

14. QUEST, Exeter 27-28 October 2005 Results –10 km (Mongolia) Percentage of pixels with CRI > 80 in a 10*10km box 2000 May 2000 1-10 11-20 51-60 21-30 31-40 41-50 71-80 61-70 81-90 91-100

15. QUEST, Exeter 27-28 October 2005 Results –10 km (Australia) 90 100 1-10 11-20 51-60 21-30 31-40 41-50 71-80 61-70 81-90 91-100

16. QUEST, Exeter 27-28 October 2005 NW Australia – 10km May 99 May overlain with May Vectors

17. QUEST, Exeter 27-28 October 2005 NW Australia 10km – May 99 May overlain with June Vectors

18. QUEST, Exeter 27-28 October 2005 the process of assessing by independent means the quality of the data products derived from the system outputs Inter-comparison – not senso stricto validation because comparison with similar products which may or may not be reliable. However, it remains the only achievable method for burned area. Sampling – Ideal situation is a statistical sample in all land covers but major problems induced by data availability and cost CEOS WGCV Definition Validation - Definition

19. QUEST, Exeter 27-28 October 2005 Validation - Criteria  Representativeness: sample all fire types (stand replacing, ground etc)  Biome type: sample the variability of biomes and conditions encountered over the Earth’s surface.  Number of sites: an adequate sampling of each of the qualifying land cover classes.  Season: cover range of vegetation conditions exhibited over the course of a growing season.

20. QUEST, Exeter 27-28 October 2005 Validation - Approach  Theory:  Algorithm: Classification of the known (in 2000) performance of GLOBSCAR/GBA-2000  Statistical: sampling of all classes (strata) to determine which locations to focus on  Data Needed: Each site characterised by two Landsat ETM images 1 month apart  Problems:  Cost: 2 scenes of Landsat ETM for all samples prohibitive  Data Availability: Often two scenes not even available (1 year or less apart)  Solution: Use what we can afford + seek national/continental databases

21. QUEST, Exeter 27-28 October 2005 Validation - Results South Sudan, Stratum 1, R 2 = 0.81Montana, Stratum 1, R 2 = 0.80

22. QUEST, Exeter 27-28 October 2005 Validation - Results Sudan South, Stratum 1, R 2 = 0.81Montana, Stratum 1, R 2 = 0.80 July 13 2000 Aug 14 2000 Sept 15 2000Mar 30 2001 Dec 24 2000 Apr 28 2000

23. QUEST, Exeter 27-28 October 2005 LAI Results – 10 km water no data > 10 9-10 8-9 7-8 6-7 5-6 4-5 3-4 2-3 1-2 0-0.5 0.6-1 June 1999

24. QUEST, Exeter 27-28 October 2005 Phenology Results – 10km water no value 201-300151-200126-150101-12551-1001-50 Leaf On 1998

25. QUEST, Exeter 27-28 October 2005 To feed in to the Global Carbon Project Earth observation must deliver long time series, consistent estimates of global vegetation behaviour complete with accuracy/quality figures. GLOBCARBON will deliver 10 complete years (1998 to 2007) of global vegetation products to the DGVM and atmospheric chemistry modelling community at resolutions of ¼, ½ degree and 10 km. Timeline: GLOBCARBON has now entered the operational production phase. Validation and inter-comparison are key elements (and being conducted by the industrial contractor). Validation and inter-comparison An ongoing process so validation data still useful – MODIS, Landsat ETM Product Release (1998-2003): January 2006 Conclusions

26. QUEST, Exeter 27-28 October 2005 Acknowledgements Many thanks to all the people who have contributed data for the validation of GLOBCARBON Burned Area especially: –Louis Giglio for TRMM hotspot information –Robert Fraser and colleagues (FireM3 and validation data) –Tom Bobbe for US burn area vectors –Chris Schmullius/Charles George for Siberia validation –Joao Silva for Africa validation data –DOLA for burned area validation in Australia