Application of multi-scale remote sensing tools for spatial characterization of indicators and drivers of fire-induced ecological impact

Résumé

In recent decades, anthropogenic activity has caused remarkable changes in the fire regime attributes in the western Mediterranean Basin, mainly due to the loss of traditional land use derived from rural abandonment, climate change and the absence of adequate forest management strategies, leading to a dense and continuous accumulation of fire-prone biomass. The new fire regime, characterized by an increase in the frequency of extensive and severe wildfires, affects important ecosystem functions and services, with unprecedented impacts at socioeconomic level. This fact is particularly relevant in wildland urban interface (WUI) areas, where extreme wildfires represent a serious threat to human life and assets. In this context, spatial characterization of fire-induced impact, commonly referred to as burn severity, is crucial to provide scientific basis to design appropriated forest management strategies that enhance adaptive responses to current fire regimes. Field methods are considered highly trustworthy for assessing the impacts on vegetation and soils in burned landscapes, though they often lack spatial exhaustiveness to evaluate large wildfires. Therefore, remote sensing methods have emerged as reliable tools for monitoring and quantifying burn severity because of their cost-effectiveness and synoptic nature. In this context, the main objective of this PhD Thesis is the development of new multiscale remote sensing techniques aimed to identify spatial indicators of fire-induced ecological impacts and evaluate the drivers of extreme wildfire behavior under different fire regimes along an Iberian climatic gradient, with particular focus in WUIs due to their high socioeconomic vulnerability. First, we aimed to improve the estimation of burn severity in forest soils, which are critical ecosystem compartments driving ecosystem functions and processes, by linking ecological indicators of burn severity with the spectral signal of very high spatial resolution remote sensing products obtained with unmanned aerial vehicles (UAV) (Articles I & II). Soil burn severity was assessed in the field 1-month after a wildfire through a Composite Burn Soil Index (CBSI) and, a set of individual indicators (ash depth, ash cover, fine debris cover, coarse debris cover and unstructured soil depth). Furthermore, indicative soil properties of fire-induced changes were analyzed: mean weight diameter (MWD), soil moisture content (SMC), and soil organic carbon (SOC). Simultaneously, post-fire multispectral images from the Sentinel-2A MSI satellite sensor, and RGB and multispectral images from a UAV survey were collected. We found that UAV multispectral products had a better performance than RGB products for estimating fire impacts on soils, being more related to integrative indices (ie., CBSI) than to individual indicators (Article I). Depth and ash cover were the most representative indicators of fire effects on soils. The inclusion of spatially and spectrally enhanced remote sensing data through novel remote sensing techniques, such as the fusion of Sentinel-2 and UAV images, significantly improved the prediction of fire-sensitive soil properties highly related to burn severity, mainly SOC (Article II). This approach provides a powerful tool for estimating fire impacts in complex and heterogeneous landscapes affected by mixed severity wildfires, and consequently to identify priority areas where post-fire restoration actions need to be implemented. Once the potential ecological impact of high severity wildfires has been adequately characterized using new remote sensing techniques, we studied fire regime shifts conducive to extreme fire behavior along an Atlantic-Transition-Mediterranean climatic gradient in the Iberian Peninsula, characterized by the occurrence of extreme wildfire events in the last few years. For this purpose, we analyzed (i) the variation patterns of temporal (recurrence and time since last fire) and magnitude (burn severity) fire regime attributes over 35-years using historical wildfire scars derived from Landsat satellite imagery collection, and (ii) the link between fire regime and pre-fire vegetation characteristics controlling extreme fire behavior. We selected eight extreme wildfires occurring during the period 2017-2022, in which we characterized both (i) the pre-fire fuel type and structure by means of image classification techniques and radiative transfer models (RTMs), and (ii) the ecological impact through the differenced Normalized Burn Ratio (dNBR) derived from bi-temporal Sentinel-2 MSI images. Fire recurrence showed the same downward trend along the climatic gradient, burn severity trends significantly differed among Atlantic and Mediterranean areas. The observed shifts in fire regime attributes had a remarkable influence in shaping fuel types and build-up patterns in landscapes prone to extreme fire behavior along the climate gradient but following distinct pathways as a function of the environmental context. In Atlantic areas, recurrent wildfires of low to moderate severity may foster forest transitions to shrubland stable states prone to high burn severity feedback in subsequent wildfires. A similar pattern was observed in Mediterranean and Transition shrublands after the recurrence of high burn severity wildfires. Under all climatic conditions, long times since the last high-severity wildfires may enhance fuel build-up in conifer forests and shrublands highly prone to extreme fire behavior. Finally, we broadened the generated knowledge about the biophysical contexts shaping extreme fire behavior in wildland urban interface areas to identify the scenarios prone to high burn severity in WUI areas due the growing concern about the socio-economic and environmental implications (Article IV). For this purpose, we chose fourteen large wildfires occurred between 2016 and 2021 across Spain that encompassed different WUI typologies. Density and distance between buildings criteria was used to differentiate isolated, scattered, dense and very dense WUIs, while several pre-fire fuel characteristics inside WUI areas were estimated through multispectral satellite imagery, following the methodology used in the Article III. Then, the combined effect of pre-fire fuel and building density patterns was used to recognize the WUI scenarios most prone to extreme fire behavior. Isolated, scattered and sparsely clustered buildings enclosed in a dense shrub matrix were the WUI typologies with the highest fire hazard. Additionally, WUIs dominated by sparse trees with a dense and continuous shrubby understory constituted another critical typology prone to severe fire impacts. We highlighted the role of pre-fire fuel management to minimize the risk to human lives and assets, particularly under increasing human pressure in WUI areas. The results obtained in this PhD Thesis allowed to predict priority scenarios for effective land use planning, wildfire prevention and management strategies, community education, and collaborative efforts in WUI areas, which are essential to address the challenges posed by new-generation wildfires to population in rural areas. We emphasize that the reduction of homogeneous fuel types, particularly shrub fuels around isolated and dispersed WUIs must be a priority intervention line. These actions should focus on breaking the fuel horizontal continuity and encouraging the development of diverse landscape mosaics to foster resistance and resilience to fire. This target can be achieved by supporting sustainable and traditional activities such as extensive livestock grazing or silvicultural actions by work crews, which is essential for population fixation in sociologically relevant areas such as WUIs.