Conference Agenda

Wildfires pose an urgent social problem in Europe due to demographic and climatic changes. As the threat increases, community preparedness has a critical role to play in mitigating the risk and easing the burden on civil protection personnel. This presentation focuses on wildfire awareness and preparedness among at-risk communities in Switzerland. Building on an online survey and interviews with residents in the Cantons of Bern, Wallis, Ticino and Graubünden, we examine the cultural, socioeconomic, political and environmental factors that influence risk perceptions, awareness raising and coping strategies. The results provide valuable insights into dominant narratives, local needs, motivations and vulnerabilities among different communities. These insights can assist official and community efforts to build wildfire resilience in Switzerland before the predicted threat becomes acute on the southern and northern sides of the Alps. They also contribute to a multi-year comparative study of different European countries as part of the SNSF-funded FiRES project.

The increasing risk and prevalence of wildfires are strongly associated with human-induced climate change. An example is Australia, where the most destructive wildfires in decades occurred in 2019-2020. However, there is still a challenge in developing effective models to understand wildfire susceptibility and pre-fire vegetation conditions. The recent launch of NASA’s ECOSTRESS presents an opportunity to monitor fire dynamics with a high resolution of 70m by measuring ecosystem stress and drought conditions preceding the wildfires. We incorporated ECOSTRESS data, vegetation indices, rainfall, and topographic data as independent variables and fire events as dependent variables into machine learning algorithms. We predicted over 90% of all wildfire occurrences one week ahead of these wildfire events. Our models identified vegetation conditions with a three-week time lag before wildfire events in the 4th week and predicted the probability of wildfire occurrences in the subsequent week (5th week). ECOSTRESS water use efficiency (WUE) consistently emerged as the leading factor in all models predicting wildfires., Results suggest that the pre-fire vegetation was affected by wildfires in areas with WUE above 2 g C kg ⁻¹ H ₂O at 95% probability level. Additionally, the ECOSTRESS evaporative stress index (ESI) and slope data were identified as significant contributors in predicting wildfire susceptibility. These results indicate a significant potential for ECOSTRESS data to predict and analyze wildfires and emphasize the crucial role of drought conditions in wildfire events, as evident from ECOSTRESS data. Our approaches developed in this study and outcome can help policymakers, fire managers, and city planners assess, manage, prepare, and mitigate wildfires in the future.

Extreme temperatures and drought in the summer 2018 resulted in widespread early leaf discoloration in Switzerland’s European Beech (Fagus sylvatica L.) forests. In subsequent years, early discolored trees exhibited higher rates of crown dieback and tree mortality. With more frequent and severe droughts expected under a changing climate, the future resilience of European Beech on the Swiss plateau remains unclear. Climate-smart forest management which accounts for species resistance to drought under an uncertain future is required to maintain important ecosystem services like biodiversity, timber production, and protection from gravitational hazards, but requires a more robust understanding of European Beech vulnerability to drought. To that end, we use a novel seasonal standardization of Sentinel-2 derived vegetation indices and in-situ field observations in a support vector classifier to model empirical European Beech discoloration with 90% accuracy (as compared to a subset of withheld field observations). This model is applied to predict monthly European Beech discoloration across Switzerland from 2017 to 2023 at a 10 m spatial resolution and is validated using independent manual mapping of discoloration in PLANET data. This unprecedented multi-temporal record reveals spatio-temporal hot spots of European Beech discoloration across multiple years suggesting that, independent of meteorological forcings, site specific factors significantly predispose some stands to discoloration over others (and thereby increase the likelihood of tree mortality). Using this newly developed empirical dataset and a combination of high-resolution soil maps, meteorological data, topographic derivatives, and information on Swiss forest structure, we are training additional ensemble machine learning models to examine which site-specific factors predispose European Beech to early discoloration. Forward applying this environmental model will 1) allow us to identify European Beech stands vulnerable to drought under a changing climate, 2) evaluate the influence of management strategies on European Beech vulnerability, and 3) provide a series of high-resolution risk maps for European Beech under various climate and management scenarios.

Extreme meteorological events, such as heat and drought, can induce significant damage to vegetation and ecosystems. The frequency and intensity of extreme events are subject to change due to anthropogenic global warming. It is therefore crucial to quantify the impact of such events for better preparedness.
Here, we focus on forest damage in Europe, defined as negative anomalies of the normalized difference vegetation index (NDVI, a measure of vegetation greenness). Compound drought and heat wave events are known to trigger low NDVI events in summer. A dry summer combined with moist conditions during the previous autumn can also have a negative impact. Hence, our study aims to find, among all the hydro-meteorological variables available as output from the sub-seasonal to seasonal forecasts in the ECMWF model, the most relevant ones to predict forest damage. For this purpose, we apply a Random Forest procedure to identify the compound hydro-meteorological conditions leading to low NDVI events at the S2S timescale. We train the model using ERA5 and ERA5-Land reanalysis datasets for the explicative variables. These variables include temperature, precipitation, dew point temperature, surface latent heat flux, soil moisture, and soil temperature. We provided an automated procedure with strong predictive performance for identifying low-greenness events during summer based on prior hydro-meteorological conditions. The most essential preceding periods and variables are location and forest-type dependent.