Synoptic circulation control on wild fire occurrence

Abstract

Wild fires have gained increasing interest during the last decade worldwide. Millions of hectares are burned out every year in various countries in all continents. Most of them are burned by the villagers to increase the areas which are available to cultivation but many fires also start by natural causes. Especially in the Mediterranean regions the demise of the traditional landscape management practices along the last decades is another relevant cause of wild fires, since large amounts of fuels are accumulated in the forests, increasing the risk of forest fires. The continuous drought, the rarity of precipitation and the intensity of the winds play a significant role in the onset and evolution of a wild fire. In this work we are trying to link synoptic weather types with events of forest fires that took place in Greece during the 20-year period 1985–2004. To find possible associations with the weather systems prevailing in an area we used the synoptic classification proposed by the ESF’s COST 733 action-release 1.2 and the classification software v.17-01. Since the onset of a wild fire is a rather complex phenomenon and is largely determined by the combination of high air temperatures and absence (or rareness) of precipitation during a certain period before the event, we used the sequential option of the classifications. Specifically we applied 3, 7 and 15 days sequences, for 9 and 18 classes, and for KMEANS, DKMEANS, KHC, LUND, HCl and SANDRAS synoptic catalogues in Domain 10 of the COST 733 scheme, in which Greece is located. The analysis shows that if we use KMEANS and HCl for both the 9 and 18 classes and for all sequences the number of synoptic types that are associated with wild fires is reduced to 2–3. These 2–3 categories are associated with almost 70–90% of the total wild fires events in Greece. On the other hand the KHC and LUND catalogues are not very successful since wild fire events are almost evenly dispersed among the categories.