RESUMEN

Wildfires pose significant threats worldwide, requiring accurate prediction for mitigation. This study uses machine learning techniques to forecast wildfire severity in the Upper Colorado River basin. Datasets from 1984 to 2019 and key indicators like weather conditions and land use were employed. Random Forest outperformed Artificial Neural Network, achieving 72 % accuracy. Influential predictors include air temperature, vapor pressure deficit, NDVI, and fuel moisture. Solar radiation, SPEI, precipitation, and evapotranspiration also contribute significantly. Validation against actual severities from 2016 to 2019 showed mean prediction errors of 11.2 %, affirming the model's reliability. These results highlight the efficacy of machine learning in understanding wildfire severity, especially in vulnerable regions.

RESUMEN

Extreme fire events have increased across south-eastern Australia owing to warmer and drier conditions driven by anthropogenic climate change. Fuel reduction burning is widely applied to reduce the occurrence and severity of wildfires; however, targeted assessment of the effectiveness of this practice is limited, especially under extreme climatic conditions. Our study utilises fire severity atlases for fuel reduction burns and wildfires to examine: (i) patterns in the extent of fuel treatment within planned burns (i.e., burn coverage) across different fire management zones, and; (ii) the effect of fuel reduction burning on the severity of wildfires under extreme climatic conditions. We assessed the effect of fuel reduction burning on wildfire severity across temporal and spatial scales (i.e., point and local landscape), while accounting for burn coverage and fire weather. Fuel reduction burn coverage was substantially lower (∼20-30%) than desired targets in fuel management zones focused on asset protection, but within the desired range in zones that focus on ecological objectives. At the point scale, wildfire severity was moderated in treated areas for at least 2-3 years after fuel treatment in shrubland and 3-5 years in forests, relative to areas that did not receive fuel reduction treatments (i.e., unburnt patches). Fuel availability strongly limited fire occurrence and severity within the first 18 months of fuel reduction burning, irrespective of fire weather. Fire weather was the dominant driver of high severity canopy defoliating fire by ∼3-5 years after fuel treatment. At the local landscape scale (i.e., 250 ha), the extent of high canopy scorch decreased marginally as the extent of recently (<5 years) treated fuels increased, though there was a high level of uncertainty around the effect of recent fuel treatment. Our findings demonstrate that during extreme fire events, very recent (i.e., <3 years) fuel reduction burning can aid wildfire suppression locally (i.e., near assets) but will have a highly variable effect on the extent and severity of wildfires at larger scales. The patchy coverage of fuel reduction burns in the wildland-urban interface indicates that considerable residual fuel hazard will often be present within the bounds of fuel reduction burns.

Asunto(s)

RESUMEN

In temperate forests, elevated frequency of drought related disturbances will likely increase the incidence of interactions between disturbances such as bark beetle epidemics and wildfires. Our understanding of the influence of recent drought and insect-induced tree mortality on wildfire severity has largely lacked information from forests adapted to frequent fire. A recent unprecedented tree mortality event in California's Sierra Nevada provides an opportunity to examine this disturbance interaction in historically frequent-fire forests. Using field data collected within areas of recent tree mortality that subsequently burned in wildfire, we examined whether and under what conditions wildfire severity relates to severity of prefire tree mortality in Sierra Nevada mixed-conifer forests. We collected data on 180 plots within the 2015 Rough Fire and 2016 Cedar Fire footprints (California, USA). Our analyses identified prefire tree mortality as influential on all measures of wildfire severity (basal area killed by fire, RdNBR, and canopy torch) on the Cedar Fire, although it was less influential than fire weather (relative humidity). Prefire tree mortality was influential on two of three fire-severity measures on the Rough Fire, and was the most important predictor of basal area killed by fire; topographic position was influential on two metrics. On the Cedar Fire, the influence of prefire mortality on basal area killed by fire was greater under milder weather conditions. All measures of fire severity increased as prefire mortality increased up to prefire mortality levels of approximately 30-40%; further increases did not result in greater fire severity. The interacting disturbances shifted a pine-dominated system (Rough Fire) to a cedar-pine-fir system, while the pre-disturbance fir-cedar system (Cedar Fire) saw its dominant species unchanged. Managers of historically frequent-fire forests will benefit from utilizing this information when prioritizing fuels reduction treatments in areas of recent tree mortality, as it is the first empirical study to document a relationship between prefire mortality and subsequent wildfire severity in these systems. This study contributes to a growing body of evidence that the influence of prefire tree mortality on wildfire severity in temperate coniferous forests may depend on other conditions capable of driving extreme wildfire behavior, such as weather.

Asunto(s)

RESUMEN

In recent years, the Mediterranean area has witnessed an increase of both the frequency and severity of large fires, which appears to be intimately associated with climate and land use changes. To measure the impact of wildfires on living organisms, diverse indicators have been proposed. These indicators of fire severity traditionally rely on quantifying the damage caused to the vegetal component of ecosystems. However, the use of bacterial communities as severity indicators has received less attention. Here, we studied the differences between bacterial communities of three different Mediterranean ecosystems, two shrubby and one arboreal, two months after a large wildfire. Two levels of severity were compared to a control unburnt soil. The results showed that greater fire severity triggers a reduction in the diversity of soil bacterial communities. In high-severity fires, this reduction reached 40.6 and 58.6% of the control values for richness and Shannon's diversity, respectively. We also found that the greatest differences between communities could be attributed first to the severity of the fire, and second to the ecosystem from which they originated. Importantly, species of just five families of bacteria: Oxalobacteraceae, Micrococcaceae, Paenibacillaceae, Bacillaceae and Planococcaceae, became dominant in all three ecosystems. The average frequency increase for particular species was 100 times. However, due to random uncontrolled factors, the species that became dominant in each community were not always the same.

Asunto(s)

RESUMEN

Disturbance events play an important role in ecosystem services management and species biodiversity. In this sense, species biodiversity may constitute a large proportion of the total ecosystem value, mainly in natural protected areas. The present research proposes a methodology for the economic valuation of flagship species; the value of charismatic species was estimated using two complementary approaches based on recovery programs and contingent valuation method (CVM). While recovery programs approach is related to government expenditure, CVM is associated with survey results according to the society's willingness to pay. There are significant differences between both approaches as flagship species are highly valued by the society. In this sense, a difference of 43.75% on the species value can be found depending on the scenario of CVM (all respondents or only affirmative respondents). Our research was done on the integration of economic tools and wildfire severity of two burned areas in order to evaluate the effects caused in their habitat and, as a consequence, in the food chain. The results obtained from both the studied areas emphasized the importance of wildfire impacts on flagship species (209,619.08-445,495.88 € from Doñana wildfire and 634.68-5792.98 € from Segura wildfire) which are often omitted in valuation reports. The use of Geographic Information Systems helps to identify flagship species impacts per unit area (74.89-159.17 €/ha from Doñana wildfire and 0.76-6.98 €/ha from Segura wildfire) and to prioritize restoration activities on the most susceptible areas. This methodology could be extrapolated to any territory and spatial resolution based on the revision of the questionnaires regarding flagship species. The availability of cartography of flagship species´ susceptibility could play a critical role in budget optimization and the decision-making process on restoration planning.

Asunto(s)