Radiative transfer modeling to measure fire impact and forest engineering resilience at short-term

Abstract

Forest managers demand reliable and cost-efficient methodologies to implement forest resilience concepts in post-fire decision-making at different spatio-temporal scales. In this paper, we developed a generalizable remote sensing-based tool to measure disturbance impact and engineering resilience at short-term in forest ecosystems affected by wildland fires. The case study was a mixed-severity wildfire that burned several shrubland (dominated by gorse, broom and heath) and tree forest (dominated by oak and pine) ecosystems. Specifically, we retrieved fractional vegetation cover (FVC) over a time-series of pre and post-fire Deimos-2 imagery (spatial resolution of 4 m) from a radiative transfer model (RTM) hybrid inversion approach (Gaussian processes regression algorithm learned from a simulation dataset generated using the PROSAIL-D model). Pre and post-fire FVC retrieval was validated with field data stratified by dominant ecosystem. High accuracy (>90%) and low error (<7%) were achieved in the retrieval over the time-series, despite the influence of background signal of soil and burned legacies. A random point sampling stratified by ecosystem and burn severity was used to extract validated FVC values for the time-series. A two-way repeated measures ANOVA was performed to evaluate the effect of burn severity along the time-series on FVC for each ecosystem. One-way repeated measures ANOVA and Tukey’s pairwise comparison test were applied to determine the earliest point in the time-series for which the FVC does not differ significantly from the pre-fire FVC. In tree forest ecosystems, the fire impact on FVC was stronger at high burn severity, being similar the impact on shrub ecosystems at medium and high burn severity. Engineering resilience was conditioned both by burn severity and species regenerative strategies. In ecosystems dominated by facultative or obligate seeders, pre-fire FVC was reached later across the time-series, compared to resprouter-dominated ecosystems. The RTM hybrid inversion tool has proved its reliability for assessing disturbance impact and ecosystem engineering resilience at short-term in heterogeneous fire-prone landscapes affected by mixed severity wildfires.