Characterization of Wildfires in Portugal

An increase in forest fires in Portugal supports climate change models suggesting that the two phenomena are linked. In recent decades the occurrence of wildfires, their severity, and the area burned, have all increased. In an effort to help in formulating a fire management plan, Marques et al. (2011) conducted a study to characterize wildfires in Portugal. The object of the study was to demonstrate trends in fire activity and examine how fuel type, fuel load, elevation, and socioeconomic factors have bearing on fire severity. What the authors found was that fire behaved selectively based on fuel type, slope and elevation, and proximity to roads and populated areas. Furthermore, they established that shrubs displayed the most significant fire activity potential, especially at higher elevations on slopes greater than 5% and further away from socioeconomic influences. –Lindon Pronto

Marques, S.,  Borges, J. G. J., Garcia-Gonzalo, Moreira, F.,  Carreiras, J. M. B., Oliveira, M. M., Cantarinha, A.,  Botequim, B., Pereira, J. M. C., 2011. Characterization of wildfires in Portugal. European Journal of Forest Research 130, 775–784.

Using historical fire information data, a 33-year-long period from 1975 to 2007 was used as a basis for observing trends in fire occurrence, proximity, and severity. Burned area mapping was established through the use of high-resolution remote sensing data by the Remote Sensing Laboratory of Instituto Superior de Agronomia. The study was broken down into three separate 5 year sub-periods (1987–1991, 1990–1994, and 2000–2004) in an attempt to minimize the effects of land cover changes over time. From land cover maps the authors were able to identify fuel types and distribution. In addition to devising 10 classes of cover types for the purpose of the study, Marques et al. identified altitude, slope, proximity to roads, and population density as four additional variables for modeling purposes. Altitude and slope data were obtained from the country’s digital terrain model (DTM); GIS overlays from the Instituto Nacional de Estatística provided data on road proximities and population density. Relationships between Ecological and socioeconomic variability and fire occurrences during the three sub-periods were largely based on statistical models.

Over the 33 year period fire perimeter data show that there were 35,194 wildfires which were greater than 5 ha in size. Area burned per year ranged from 15,500 ha in 1977 to 440,000 ha in 2003, where a single fire was responsible for 58,000 ha alone. The first sub-period (1987–1991) had 7,672 starts; the second sub-period (1990–1994) exhibited significantly calmer fire activity with 5,703 starts. The third and final sub-period (2000–2004), was characterized by a significant increase in fire occurrence and size; while the period had 7,383 starts, the area burned was over 43% greater than the first sub-period and 182% for the second sub-period respectively. Most notably the final period exhibited the occurrence of four very large fires being greater than 20,000 ha each in size. There were no fires greater than 20,000 ha during the first 25 years of data.

Weighted generalized linear models (WGLM) proved that the number one high risk fuel was shrubs, followed by mixed stands, softwoods and hardwoods; individual species added variance based on fuel loading, resin, and foliage essential oil content. Marques et al. also observed that fires occurred more frequently at higher elevations, which was attributed to higher lightning activity levels (LAL) and escaped pastoral burns. An additional factor found more generally at higher elevations was that of greater slope which contributed to faster rate of spread. In populated areas in Portugal, although human activity is the number one cause of wildfires, their proximity to roads and population centers allows for a very quick response time from firefighters who are often able to extinguish the fires when they are still small. When fires occurred away from populated areas where there was limited or no access, data show that these fires tend to become very large, especially in mountainous areas where slope accelerates rate of spread. There was a positive correlation between distance from populated areas and the area burned.

Through this study, Marques et al. were able to characterize wildfire in Portugal with special attention to socioeconomic influences, fuel type, and landscape specific variability. The technique which made this approach possible was the use of weighted generalized linear models which highlighted the relationships of ecological and socioeconomic factors. This study is intended to provide a starting point for policy makers to develop an appropriate and effective fire management plan that is current with wildfire activity trends and congruent with the possible effects of climate change. It provides a context for developing fire prevention practices and policies; furthermore, it suggests continued work in this subject area to translate these results into functioning fire prevention and suppression models for the country of Portugal.