A semi-mechanistic model for predicting daily variations in species-level live fuel moisture content

Abstract

• Live Fuel Moisture Content (LFMC) is an essential constraint of wildfire activity. • LFMC estimated from weather/vegetation indices lacks species-specific predictions. • We use a water balance model to predict species-specific LFMC from water potential. • Our model shows an improvement in predictive performance over existing approaches. • Our approach can be implemented within large-scale fire danger forecast systems. Live Fuel Moisture Content (LFMC) is one of the main factors affecting forest ignitability as it determines the availability of existing live fuel to burn. Currently, LFMC is monitored through spectral vegetation indices or inferred from meteorological drought indices. While useful, neither approach provides mechanistic insights into species-specific LFMC variation and they are limited in the ability to forecast LFMC under altered future climates. Here, we developed a semi-mechanistic model to predict daily variation in LFMC across woody species from different functional types by adjusting a soil water balance model which estimates predawn leaf water potential (Ψ pd ). Our overarching goal was to balance the trade-off between biological realism, which enhances model applicability, and parameterization complexity, which may limit its value within operational settings. After calibration, model predictions were validated against a dataset comprising 1659 LFMC observations across peninsular Spain, belonging to different functional types and from contrasting climates. The overall goodness of fit for our model ( R 2 = 0.5) was better than that obtained by an existing models based on drought indices ( R 2 = 0.3) or spectral vegetation indices ( R 2 = 0.1). We observed the best predictive performance for seeding shrubs ( R 2 = 0.6) followed by trees ( R 2 = 0.5) and resprouting shrubs ( R 2 = 0.4). Through its relatively simple parameterization, the approach developed here may pave the way for a new generation of process-based models that can be used for operational purposes within fire risk mitigation scenarios.