The Amazon Dashboard tracks individual fires in the Amazon region using a new approach to cluster and classify VIIRS active fire detections by fire type (see Methods). Based on the fire location, intensity, duration, and spread rate, each individual event is classified as a deforestation fire, understory forest fire, small clearing & agricultural fire, or savanna fire. The data and summary figures are updated daily using the combined active fire detections from the VIIRS instruments on the Suomi-NPP and NOAA-20 satellites. Please use the map interface to explore specific regions or fires of interest, including fires burning in protected areas or indigenous territories, or download the data in shapefile or GeoJSON format for a complete look at 2023 fire activity across the Amazon biome or other regions in the full study domain (10N – 25S).”
The Amazon Dashboard is based on a new near-real time methodology (Andela et al., 2022) to cluster satellite active fire detections into individual fire events, and classify each event as one of four different Amazon fire types. We differentiate deforestation fires, understory forest fires, small clearing or agricultural fires, and savanna and grassland fires. The algorithm tracks five metrics of fire behavior for each individual fire. These metrics of fire behavior are subsequently combined with information on land cover and historic deforestation within the fire perimeter to identify the specific fire type and confidence class (low, moderate, high) for each event. The classification methods are briefly explained below. Please see the Supplementary Materials section from Andela et al. (2022) for a complete description of the methodology and validation results.
Data on individual fires are updated daily for the South American study region (10°N-25°S) and available in shapefile and GeoJSON format for use in any GIS software. The shapefile attribute table contains information about each fire event, including fire type, data used by the algorithm to estimate fire type, as well as other practical information for land management purposes (Table 1).
For questions, please contact Niels Andela or Douglas Morton.
Shapefile Attribute Table Reference
|Attribute Class||Attribute||Explanation / Units|
|Fire type classification||cluster_ID||Unique identifier number of fire event, changes with each daily updated|
|fire_type||(1) savanna and grassland, (2) small clearing and agriculture, (3) understory, and (4) deforestation fires|
|confidence||(1) low, (2) moderate, (3) high|
|Metrics used for fire type characterization||tree cover||Average tree cover fraction within perimeter (%) from Hansen et al. (2013)|
|biomass||Average biomass within fire perimeter (ton ha-1) from Avitabile (2016)|
|deforestation||Fraction (0-1) of 550m grid cells with historic deforestation (2015-2019) within fire perimeter from INPE (2020)|
|size||Fire size in km2|
|detections||Total number of fire detections within fire perimeter|
|frp||Average fire radiative power (FRP) for all fire detections within fire perimeter in megawatts (MW)|
|persistence||Average fire persistence across 550m grid cells within fire perimeter (days)|
|progression||Average fire progression fraction across 550m grid cells within perimeter (0-1)|
|daytime||Fraction of 1:30pm detections (0-1) for all fire detections within fire perimeter|
|Status||start_DOY||Day of new fire start as of day of year (1-366)|
|last_DOY||Most recent active fire detection within fire perimeter as of day of year (1-366)|
|is_new||New fire start within past 24 hours (1) or existing fire (0)|
|is_active||Fire was active within the past 10 days (1) or not (0)|
|biome||Fire is within (1) or outside (0) the Amazon biome according to RAISG|
|protected||Fraction (0-1) of fire occuring within protected area or indigenous land according to RAISG|
Tracking wildfires in near-real time
Recent work by Oliva and Schroeder (2015) has shown that active fire detections from the Visible Infrared Imaging Radiometer Suite (Schroeder et al., 2014) can be used to map burned area of slow moving forest fires in near real-time. Here we combine the 375 m day and nighttime active fire detections from the VIIRS sensors onboard Suomi-NPP and NOAA20 satellites (see FIRMS) to provide daily burned area estimates at 0.005° resolution (~550 m).
We then use the Global Fire Atlas algorithm (Andela et al., 2019) to track the perimeter of individual wildfires at any given day across the region. Initial comparisons with burned area maps (that are typically available with a 2-3-month delay) indicate good agreement for forested ecosystems (e.g. deforestation and understory fires), but fire spread rates in savannas and grasslands are typically too fast to capture “wall to wall” burned area based on active fire detections alone. This results in artificial fragmentation of some savanna and grassland fire polygons in our dataset. Our main objective, however, is to provide an improved monitoring tool and early identification of forest fire types.
Identifying Fire Type
Large multiday wildfires contribute to most of global burned area and active fire detections (Andela et al., 2019). Aggregating active fire detections into individual fire events therefore provides new insights about the nature of these larger fire events.
For each fire event we combine information about land cover (tree cover, biomass, historic deforestation rates) within the fire perimeter and fire characteristics (e.g., fire radiative power, persistence, progression, day-time detection fraction, and size) based on all satellite fire detections within the larger perimeter. The combination of these data provides a unique signature of each fire type that allows for classification using three confidence intervals (low, moderate, and high).
Deforestation fires typically have high initial fire radiative power, as piled woody debris leads to greater energy release, and long fire persistence, since these piles may smolder for days. By contrast, understory forest fires spread slowly for weeks or months, with lower FRP that leads to a higher fraction of nighttime fire detections.
For example, during 2019, the average understory fire event triggered 54 active fire detections and burned 5.1 km2. Small fires in forested systems (tree cover >50% and equal or less than 5 fire detections) were classified as small clearing or agricultural fires, a broad class encompassing a range of small fire types with short durations.
To separate savanna and grassland fire from other fire types we combine information on tree cover fraction and historic deforestation rates within the fire perimeter.
The current methodology was developed based on the 2019 fire season and thresholds were selected based on historic deforestation data (2014 – 2018) and a number (77) of manually selected understory fires. The methods for the Amazon Dashboard data in 2023 are fully described in the manuscript and Supplementary Materials by Andela et al. (2022). Our research team remains committed to improving the quality and utility of individual fire event data, in partnership with regional stakeholders, and we would welcome the opportunity to collaborate to improve regional training and validation of the fire classification product. Please contact Douglas Morton with any suggested improvements on the algorithm.
- Andela, N., Morton, D. C., Giglio, L., Paugam, R., Chen, Y., Hantson, S., van der Werf, G. R. and Randerson, J. T.: The Global Fire Atlas of individual fire size, duration, speed, and direction, Earth Syst. Sci. Data, 11, 529–552, doi:10.5194/essd-11-529-2019, 2019.
- Avitabile, V., Herold, M., Heuvelink, G. B. M., Simon, L., Phillips, O. L., Asner, G. P., Armston, J., Peter, S., Banin, L., Bayol, N. and Berry, N. J.: An integrated pan-tropical biomass map using multiple reference datasets, Glob. Chang. Biol., 22, 1406–1420, doi:10.1111/gcb.13139, 2016.
- Hansen, M. C., Potapov, P. V., Moore, R., Hancher, M., Turubanova, S. A., Tyukavina, A., Thau, D., Stehman, S. V., Goetz, S. J., Loveland, T. R., Kommareddy, A., Egorov, A., Chini, L., Justice, C. O. and Townshend, J. R. G.: High-resolution global maps of 21st-century forest cover change, Science, 342, 850–853, doi:10.1126/science.1244693, 2013.
- National Institute of Space Research (INPE), PRODES deforestation. http://terrabrasilis.dpi.inpe.br/en/home-page/, accessed March 20, 2020.
- Oliva, P. and Schroeder, W.: Assessment of VIIRS 375m active fire detection product for direct burned area mapping, Remote Sens. Environ., 160, 144–155, doi:10.1016/j.rse.2015.01.010, 2015.
- Schroeder, W., Oliva, P., Giglio, L. and Csiszar, I. A.: The New VIIRS 375m active fire detection data product: Algorithm description and initial assessment, Remote Sens. Environ., 143, 85–96, doi:10.1016/j.rse.2013.12.008, 2014.