Synoptic patterns associated with large summer forest fires in Portugal

Time series of the total annual burnt area in Portugal reveal two main features, a large inter-annual variability and a positive trend since the early 80 s. Here we show that inter-annual variability is partly due to the amount of precipitation in the fire season and in the preceding late spring season and partly to the occurrence of atmospheric circulation patterns that induce extremely hot and dry spells over western Iberia. On the other hand, the observed positive trend of burnt area is mainly related to changes in farming and land use. Meteorological conditions play a fundamental role, both in the ignition and during the fire spread. The description of spatial and temporal variability of wildfire characteristics is performed using the comprehensive fire data set (between 1980 and 2000) from the Portuguese forest service. We show that the vast majority of the burnt area in Portugal (80%) is due to fire events that occurred on in a very small number (10%) of summer days. Large-scale climatic and dynamical meteorological fields were retrieved from the NCAR/NCEP Reanalyses data sets for the 1961-2000 period and composites were then obtained for the 10% of summer days associated with the highest values of burnt area. Anomaly fields of climate variables (e.g. 850 hPa temperature and relative humidity) are interpreted based on physical mechanisms associated with dynamical variables such as the surface wind field or the 500 hPa geopotential height. Overall, one may state that synoptic patterns of most analysed meteorological fields present statistically significant anomalies over western Iberia. In particular, composites of geopotential height for mid (500 hPa) and lower (850 hPa) troposphere show that large forest fires in Portugal occur when the atmospheric circulation forms a prominent ridge over the Iberian peninsula with the flow being dominated by a strong meridional component. Near the surface, wind and sea level pressure anomalies show that these days are associated with south-easterly conditions, with a strong anomalous advection from northern Africa that is further heated when crossing the central Iberian plateau. Large asymmetries between minimum and maximum temperatures composites are analysed taking into account the lack of cloud cover and corresponding precipitation. Finally, we present a linear model based on the monthly precipitation and the occurrence of previously identified wildfire prone atmospheric patterns. The developed model gives a correlation coefficient of 0.8 between the observed and modeled extent of burnt area during the summer. (c) 2005 Elsevier B.V. All rights reserved.