Assessing post-fire recovery in Mediterranean forests using spectral metrics and physiological models

Abstract Background Monitoring forest recovery after wildfires is crucial for informing management decisions aimed at promoting regeneration. The wide range of Earth observation methods available makes selecting the most suitable methodology challenging. Here, we compare two distinct frameworks for monitoring post-fire recovery, considering both forest cover and functional recovery, through a spectral approach and a model-based approach. We evaluated 17 wildfires that occurred in Catalonia (Spain) in 2003, analyzing time series of satellite derived Normalized Burn Ratio (NBR) and modelled Net Primary Productivity (NPP) from 2000 to 2020. For both approaches, we obtained metrics of recovery rate and time to recovery and compared their outcomes in order to evaluate the similarities between them. To assess the influence of topography, weather, pre-fire vegetation conditions and fire magnitude on observed differences, we conducted correlation analyses and statistical tests for significant differences. Results After 17 years, 91% and 98% of the burned area had recovered, according to the spectral (NBR) and model-based (NPP) approaches, respectively. Recovery rates estimated using both approaches were strongly correlated (0.76), and the correlation increased to 0.96 when rates were normalized by the magnitude of change. The analyses suggested water availability, vapor pressure deficit, fire magnitude, elevation and pre-fire NBR as key explanatory factors associated with forest recovery. The spectral approach indicated faster recovery than the model-based approach under conditions where water availability was not limiting. The model-based approach offered a more conservative recovery estimation showing a strong negative association with the magnitude of fire. Conclusions Spectral and model-based approaches produce comparable recovery rates and times in the long-term. However, in the short-term, the spectral approach related to forest cover indicated faster recovery than functional recovery. These findings suggest that integrating spectral and model-based approaches could provide a more complete view of post-fire recovery by capturing both vegetation cover and ecosystem productivity.