Global-scale statistical modelling of the radiative power released by vegetation fires using a doubly truncated lognormal body distribution with generalized Pareto tails

We present a statistical model with 8 parameters that combines three components, namely a truncated lognormal distribution central body with a lower and an upper tail, both consisting of Generalized Pareto (GP) distributions. We fit the model to a dataset of more than 85 million records of the logarithm of fire radiative power (FRP) at the global scale, as derived from satellite observations covering a 19-year period (2002–2021). We use the model to characterize fire activity associated to three fire macro-regimes, designated Wild, Tamed and Domesticated, and to identify a global decreasing trend in fire intensity with moderate release of FRP (10–100 MW) during 2002–2021 that is consistent with agricultural expansion and intensification. The model can be applied to the calibration of meteorological fire danger indices that are used to assist fire prevention and suppression activities.