Introduction:

Estimating the risk of occurrence of a fire contributes to reducing human, infrastructure and natural resource losses; promoting activities to maintain and restore fire regimes; and optimizing resources for suppression.

Objective:

To develop an index of occurrence of forest fires on large areas, called Area at risk of fire (SeR).

Materials and methods:

The index corresponds to the area associated with a probability level measured at the right tail of the density distribution of the area affected annually by forest fires. The density distribution was estimated from the history of the area affected (1970-2018) in Mexico by state. The fit was performed by minimizing the Kolmogorov-Smirnov statistic with four models: exponential, gamma, lognormal and Weibull. Two related indicators are proposed: proportion of forest area affected by wildfires (PSeR) and incremental area at risk (ISeR).

Results and discussion:

all models showed a statistically significant fit (P < 0.05); the lognormal model performed the best. The SeR discriminates territorial units with the largest area affected by fires; additionally, it efficiently predicts the area to be affected by fires. The PSeR facilitates the comparison of the risk of fire occurrence between territorial units of different sizes, while the ISeR estimates the change in the maximum area affected by fires over a period.

Conclusion:

SeR is an extreme event risk index that provides useful information and has a statistically acceptable predictive power.

        · abstract in Spanish     · text in English | Spanish     · English ( pdf ) | Spanish ( pdf )