Real-time AVHRR forest fire detection in Mexico (1998-2000)

Abstract

Using our NOAA satellite ground receiving station we have implemented a real-time multispectral fire detection procedure. The Mexican territory is divided into five regions. For each region, time-dependent dynamic temperature threshold values are determined. To avoid solar contamination, nocturnal and early morning Advanced Very High Resolution Radiometer (AVHRR) Local Area Coverage (LAC) images are used. The model determines hot points using channel 3 data. Channel 4 data are used for determining the environmental (surface) temperature. Channel 1 data are used to detect smoke and pixels having a high albedo (4-7%) due to clouds and ground features. Channel 2 data are used to identify land irregularities. Detected hot points are then brought into a geographic information system for ecosystem classification on a national vegetation map which identifies natural protected areas, fragmented forest and agricultural areas. Two daily reports were delivered during 1999 and 2000 to the corresponding national and state authorities so they could take immediate actions. The fire season of 1998 was the most severe season. Satellite imagery detected many fires not only in Mexico but also in Guatemala and Honduras during April and May. South-westerly winds transported the smoke haze over the Gulf of Mexico reaching the USA from New Mexico to southern Florida and sometimes as far north as over the Great Plains. Comparison of the results show that affected areas have been reduced by 7.3 ha per event (1998), 4.0 ha per event (1999) and 2.8 ha per event (2000), respectively. The most affected regions of the country were as follows--in the south: States of Chiapas, Oaxaca, Tabasco and Quintana Roo; in the centre: Estado de México, Guerrero, Jalisco, Zacatecas and Tamaulipas; in the north-east: Coahuila, Nuevo León; and in the north-west: Chihuahua and Sonora. The areas classified as Fragmented Forest show the maximum number of events (80%). This may be because these areas tend to have considerable agricultural activities. It is important to take measures to eliminate these activities.