Survival functions of holdover time of lightning-ignited wildfires

Abstract

Lightning-ignited wildfires (LIWs) can go unreported for hours, days or even weeks before being reported. This is due to the fact that some LIWs present an intermediate phase between ignition and detection characterized by a smoldering combustion. Holdover time is generally defined as the time between lightning-induced ignition and fire detection. This study aims at obtaining survival functions to estimate the probability of a LIW reaching a certain holdover time. To this end, we fitted nine different probability distributions (exponential, chi-squared, log-normal, log-logistic, F, gamma, Weibull, Pareto, and Gompertz) to data from a database gathering 42 frequency distributions of holdover time obtained from more than 152,375 LIWs in 13 countries from 1921 to 2020. Gamma distributions provide the best fits to the observed holdover times. Accordingly, we estimated several survival functions derived from gamma distributions fitted to holdover time data. The survival functions are monotonically decreasing functions characterized by high probabilities for short holdover times and low probabilities for long holdover times. These survival functions can be used for holdover times of LIWs occurring globally as well as in boreal, Mediterranean and temperate coniferous forest biomes. Survival functions are intended to provide a more reliable way to assess holdover time-based probabilities of lightning causing wildfires.