1. Detection and Monitoring of Wildfires by MODIS on the Terra and Aqua Satellites Charles Ichoku (SSAI, NASA/GSFC) Yoram Kaufman (NASA/GSFC), Chris Justice (UMd), Louis Giglio (SSAI, NASA/GSFC), Jacques Descloitres (SSAI, NASA/GSFC), Rong-rong Li (SSAI, NASA/GSFC), Wei-Min Hao (USDA Forest Service Fire lab.), David Herring (SSAI, NASA/GSFC) http://modis-fire.gsfc.nasa.gov/ Presented at the 2001 DOE/Contractor Fire Protection Workshop April 23-27, 2001, Augusta, GA

2. All objects on earth emit their own energy and reflect solar radiation

3. In order to observe the Earth and measure changes on it, we “remotely sense” the radiant energy that is reflected and emitted from Earth at various “wavelengths” of the electromagnetic spectrum. Everything— from plants, to water, to clouds, to burning objects, and even particles suspended in the atmosphere (aerosols)—reflects, emits, and absorbs radiant energy in unique and distinct ways. But our eyes are only sensitive to the “visible light” portion of the EM spectrum. Sensors can be sensitive to various parts. The Electromagnetic Spectrum

4. Atmospheric “Windows”

6. EOS Terra–Launch: Dec 18, 1999 Instruments on board MODIS ASTER CERES MISR MOPITT

7. EOS Terra–Polar Orbit

9. EOS/MODIS Product Terminology Granule: Levels 0, 1, 2, Orbit segment containing ~5 minutes of data Tile: Levels 3, 4, ~1200 km x 1200 km grid, ~350 tiles cover earth’s surface

10. Rationale for MODIS Fire Remote Sensing Fire can be a natural hazard with large societal costs and impacts Fire is an important source of trace gas and particulate emissions Fire is a proximate cause / indicator of land cover change Fire is an important biogeochemical process with a major role in the carbon and nitrogen cycles Fire is an important ecological disturbance regime Fire frequency can be expected to change with climate change and variability Fire is a major land management practice in tropical systems Fire frequency will change with population dynamics

11. MODIS Active Fire Detection 1 km fire (Thermal) bands characteristics –Channel 21: 3.96 µm, ~ 450 K saturation –Channel 31: 11.0 µm, ~ 400 K saturation –Channel 22: 3.96 µm, ~ 330 K saturation - lower noise, lower quantization error, 1 km multi-purpose band MODIS Active Fire Algorithm –Absolute and Difference Thresholds for Bands 2, 21,22, and 31 –Also uses: Difference with background brightness temperatures (21/22/31) Standard deviation of surrounding pixel brightness temperatures Mod 35 - Cloud Mask Geolocation Bands 1 and 2 for glint rejection and QA flags

12. MODIS detects Fires! Day 236 Acquired Aug. 23 Produced Aug. 30 True-color image of smoke & burn scars over Montana fire pixels added using 3.75 µm channel By Rong Rong Li

13. MODIS Combined Fire and Reflectance Product: Global Browse (Aug 22 2000) MODIS Fires (red) superimposed on Surface Reflectance J. Descloitres /C.Justice

14. MODIS Fires S. Lake Malawi, August 23, 2000

15. MODIS Fires, NW Australia, Oct 2, 2000

16. TOMS Aerosol Index (Oct 2)

17. MODIS vs. AVHRR fires in Idaho By Rong Rong Li and Yoram Kaufman Ch 20 saturates at 330K, Ch 21 saturates at 450K thus shows the location of the fire fronts (red)

18. MODIS burned-area product (experimental algorithm ) Algorithm operates on multi-temporal data at the pixel level Removes the need for reflectance thresholds, which are sensitive to the spatial and temporal variations of burned areas Takes advantage of the bi-directional reflectance (BRDF) properties of most natural surfaces observed by wide field of view sensors from optical to thermal infrared wavelength Algorithm labels burn scar pixels as those where change is detected in a consistent manner for a specified number of days Residual cloud, sub-pixel cloud, cloud and relief shadow, and any other bad quality data are removed through use of this temporal consistency constraint. Algorithm being run for Southern Africa using MODIS 500m 0.86mm, 1.24mm or 1.64mm land surface reflectance time-series

19. Leftburned area results -> burning over days 249-290 (500m), 2000 Right temporal composite of MODIS day and night active fires detected during the same period (1km) Key purple - beginning of the time series (day 249) red - end of time series (day 290) white - insufficient data in the time series to make a burning decision

20. Fire Product Validation Primary active fire validation measurements include: –Field surveys, –Airborne imagery – MAS –Fine and high-resolution satellite imagery (ASTER, L7) Initial Fire and Burn Scar validation efforts are being conducted through the SAFARI 2000 initiative using a network of regional collaborators: –SAFARI MAS active fires, –SAFARI Landsat-7 Burned Area w. Miombo Science Network –ASTER – contemporaneous –MODIS/AVHRR/TRMM/ATSR inter-comparisons Collaboration with W. M. Hao / Darold Ward - NASA EOS Validation investigation on Montana/Utah Fires, through USFS/Canadian Forest Service Fire Record International fire product validation coordination through the CEOS Cal/Val subgroup on Land Product Validation – GOFC initiative.

22. MISR composite Sept 07 2000 MAS composite 20, 7, 1 Sept 07 2000 MODIS Composite 6, 5, 2 Sept 05 2000 MODIS Composite 6, 5, 2 Sept 05 2000 Skukuza SAFARI Validation Site Landsat 7 Composite 5, 4, 3 Aug 31 2000

23. Five Paths to Operational MODIS Level 1b Data < 1 hour ~ 2 hours ~ 12-14 hours ~ 12-24 hrs ~ 4 hrs Raw data to EDOS L0 data in rec. station < 2 hours MODIS Direct Broadcast data W/in seconds ~ 8-10 hours Data at L0 in GDAAC L1b data in rec. station Data to NOAA via “bent pipe” ~ 2 hours ~ 12-14 hours Data at L1b in GDAAC Data at L1b in MODAPS Fast response processes to L1b L1b ready at NOAA

24. The data were received and processed in real time from the NASA MODIS sensor aboard the TERRA Satellite, at 11AM on January 3, 2001. The TeraScan SX-EOS Direct Broadcast system is located at SeaSpace's Corporate facility in San Diego

25. Towards Operational Monitoring of Fires AVHRR, DMSP and GOES are all used to generate fire data sets - but currently no ‘operational’ commitment for generation of fire products. EU-JRC World Fire Web collecting fire data from AVHRR regional ground stations NOAA-16 AVHRR. Band 3a (1.6µm band) will be activated in the daylight hemisphere, 3b (3.7µm band) will be on at night – capability for monitoring night-time fires only. NOAA-M (NOAA 17) to be launched in June 2001 will be in a midmorning orbit with an equator-crossing time at 10 a.m. local solar time providing two AVHRR’s making daylight observations – one could provide 3.7µm data for daytime fire monitoring. MODIS (Research Systematic Measurements) –Terra Fire products and Rapid Response Fire system for Spring 2001 –MODIS PM (AQUA, Late 2001) will complement TERRA providing four fire observations per day – c. 10.30 and 2.30 am and pm. NPP (NPOESS) VIIRS (2005) will include a mid-IR channel with fire monitoring characteristics. The Global Observation of Forest Cover Project (GOFC) is providing an important international coordination mechanism for operational fire monitoring - http://www.gofc.org/gofc/index.html