Multitemporal analysis of NDVI and land surface temperature for modeling the probability of forest fire occurrence in central Mexico

Abstract

Forest fire induce drastic, and sometimes extensive changes in the landscape. Data from the Advanced Very High Resolution Radiometer (AVHRR) operated by NOAA have been used in several studies of forest fire. Their features allow detection of active fires as hot-spots, the monitoring of vegetation condition, and estimation of land surface temperature (LST). Multitemporal analysis of Normalized Difference Vegetation Index (NDVI) and LST calculated from NOAA-AVHRR14 in the course of the four months (November to February) previous to the fire seasons (March to May) for the period 1996-2000 allowed to identify a set of the dynamic predictive variables by constructing a logistic model to assess the risk of forest fire over the central region of Mexico. Actual forest fires were detected as hot-spots on nighttime NOAA-AVHRR 14 images from the four fire seasons (March to May) from 1997 to 2000. In addition, elevation, aspect, slope, vegetation type, and precipitation were selected as the static predictive variables of the model. The data base included 846 fires, and 869 random non-fire points from 1997-1999. During 2000 There were 143 forest fires, which were used to assess the accuracy of the model.