Evaluation of fire danger indices in Sardinia (Italy)

Fire danger models are commonly used in countries where wildfire is an important natural threat to forests and wooded areas. In the last decades, several research studies were conducted to develop fire danger model (FDM). However, great attention is needed when FDMs are used outside of the geographical areas where they have been calibrated. In addition, although fire danger forecast are available at broad spatial scale (i.e., national), information is not always useful at local scale.

In this paper, a new integrated fire danger index (IFI) developed for an operational use in Sardinia (Italy) was compared with some of the most diffuse fire danger indices (i.e., Canadian Fire Weather Index, Australian MK, Portuguese Index, etc). Daily weather and wildfire data from Sardinia during the period 2000-2007 were used. Daily IFI values were calculated for seven homogeneous climatic areas of Sardinia. For evaluating the performances of each index, a set of statistical tools (linear regression, logistic regression and percentile analysis) was used to compare danger outputs with wildfire events.

The analysis showed a good performance of IFI compared to the other indices. When regression statistics were calculated, good values of the coefficient of determination were obtained for most of the wildfire data (i.e. burned area, number of events, etc.). The percentile analysis showed that IFI was able to capture different fire danger levels better than the other indices, followed by Australian MK4 and FWI. Similar results were obtained from the logistic regression analysis considering the fire danger index as independent variable and the fire activity as dependent variable.

Results confirms that FDMs require site-specific calibrations and give accurate danger predictions. IFI performances are promising and suggest it can be used to assist fire managers in controlling and mitigating the harmful effects of wildfire in Sardinia. IFI has been recently implemented in a user-friendly software application (IFI ver. 1.3) and, from 2005, the Fire Forest Service of Sardinia has been using IFI as a predictive tool for forest fire management.