Forest Fire Alert System: a Geo Web GIS prioritization model considering land susceptibility and hotspots - a case study in the Carajás National Forest, Brazilian Amazon

• Published in: International journal of geographical information science : IJGIS 2010-04, Vol.24 (6), p.873-901
• Main Authors: Roberto Barbosa, Marcelo, Carlos Sícoli Seoane, José, Guimarães Buratto, Mario, Santana de Oliveira Dias, Leonardo, Paulo Carvalho Raivel, João, Lobos Martins, Flavio
• Format: Article
• Language: English
• Subjects: alert system; algorithm; Carajás National Forest; forest fire; geographical information system; hotspots; susceptibility maps
• ISSN: 1365-8816; 1362-3087
• DOI: 10.1080/13658810903194264

To increase the monitoring potential of forest fires, an alert classification methodology using satellite-mapped hotspots has been established to help forest managers in the prioritization of which hotspot to be verified in the field, thus potentially improving the distribution of fire-fighting resources. A computer application was developed based on web-distributed geographical information technology whose main function is to interact automatically generated satellite hotspots and risk areas indicated in fire-susceptibility maps and classify them into five alert levels. The location of the hotspots is available continuously every 4 h, and a susceptibility map is produced daily through map algebra algorithm, which uses static (topography, vegetation and land use) and dynamic (weather) variables. Every process runs through automated geoprocessing routines. The methodology was tested during the dry period of 2007 in the Carajás National Forest, in the Brazilian Amazon, within an area of 400,000 ha. It is a critical area constantly threatened by fires caused by invasions and deforestation owing to intense agribusiness advances and mining activities in its surroundings. This situation results in observations of many hotspots inside the study area for the same day and almost the same time period, in places of extreme opposites, demanding complex rapid analysis and hindering the decision of the displacement of fire-fighting teams. Further, a major mining company operates within the National Forest area, maintaining actions of protection as part of its environmental mining license. Results are presented under three aspects: (i) the credibility of the daily susceptibility map (algorithm), which showed strong correlation between areas of greatest risks and the confirmed forest fires; (ii) the reliability of hotspots (alert levels), confirming 71% of fires; (iii) accuracy in the decision of which hotspot to be checked, which revealed the same number of verifications at different alert levels, 82% confirmed alert 5 hotspots (maximum) and only 50% from alert 1 (minimum), resulting in faster fire-fighting actions, minimizing burned areas and, in some cases, allowing fire control before its spreading. Therefore, the methodology demonstrated that GIS routines are able to determine the relationship between a reality-based, interpreted susceptibility map of the area and satellite-generated hotspots, highlighting the ones of highest hazard level through the alert classifica