Catastrophe (cat) models are tools, typically used in the (re)insurance industry, that evaluate the risks to a given portfolio by modelling the impact of thousands of years of synthetic hazard events. Of particular interest to users is an evaluation of the low probability (tail) risks. This includes asking questions such as, “what is the worst loss event that will be exceeded, on average, every 200 years?” 

An assessment of tail risks is inherently uncertain. This is compounded by a large number of uncertain or free parameters throughout the modelling chain which may be set via expert (subjective) judgement or via a process of calibration. The calibration process would take a given portfolio with known historical losses and adjust some of the free parameters to match the historical losses. This process may be reframed as creating a structured ensemble of catastrophe models with a range of each of the free or uncertain parameters. The process would then compare the modelled losses from each of the ensemble members to the known historical record and select the model that best represents the historical losses. 

A major limitation of the ensemble approach to catastrophe model calibration is the short historical record from which to select the most representative model. This work uses a flood catastrophe model ensemble to explore the calibration process by creating a short synthetic loss record from a single ensemble member and examining the downstream effects of using this loss record for model selection. 

How to cite: Lamb, C., Haylock, M., Wing, O., and Sloan, O.: Using an ensemble of flood catastrophe models to explore the interplay of loss variability and the catastrophe model calibration process, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-6304, https://doi.org/10.5194/egusphere-egu25-6304, 2025.

The HuT (The Human-Tech Nexus) project aims at finding effective strategies to manage the risks associated with extreme climate events by means of specific demonstrators over the European territory in which different Disaster Risk Reduction strategies are prototyped and tested. In this context, we show here two distinct innovative insurance prototypes to cope with risks associated with wildfires and landslides over two peculiar areas in Sardinia and Campania regions (Italy). While the hazard posed by the two perils show distinct characteristics and origins, in both cases an insurance product can play a crucial role in the aftermath of the events for communities and private stakeholders. Since the risk assessment is crucial both in terms of financial structure and pricing strategies of a natural hazard insurance product, prototypes are developed through a Nat Cat modeling-based hazard assessment, while the vulnerability and finance considerations are related to the specific characteristics of the area of interest. Eventually, two prototypes are fully developed: “Landslide First Rescue”, a semi-parametric product designed to cope with the immediate economic needs after a landslide events; and “Fire Safe Community”, proposed as a community-based efficient tools to restore the economic losses related to wildfires. The prototypes present specific discounts if the policy holders are willing to implement risk reduction solutions to cope with the specific natural hazard. Results prove that the final premium associated with the products would be affordable and several consultations with interested stakeholders have shown how these kinds of products could also play a role in the development of nature-based solutions over broader regions.

How to cite: Lo Conti, F., Gallotti, G., Tirri, A., Santoro, A., Rianna, G., Bacciu, V., and Calvello, M.: Disaster Risk Reduction through innovative insurance solutions , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7030, https://doi.org/10.5194/egusphere-egu25-7030, 2025.

To mitigate the growing intensity, duration, and frequency of wildfires in recent years, leveraging the latest forecasting tools and maximizing their capabilities is essential. The FireHUB platform, provided by Beyond Operational Unit of the National Observatory of Athens, has been a reliable decision-support system utilized by numerous decision-makers and public bodies. It is also a continuously evolving platform. The most recent enhancement, implemented under the framework of the MedEWSa project, involves the deployment of a brand-new fire-spread model, offering several comparative advantages that are presented in this study.

A variety of atmospheric and soil parameters (e.g., wind, air/soil temperature and humidity, fuel density) are necessary to accurately predict fire spread information. Many of these factors are influenced by local topographical features, making high-resolution forecasts crucial. Additionally, the ability of a fire-spread model to ingest and process spatiotemporally variable fields is critical. Deploying the ForeFIRE code in combination with finer grid scales from our atmospheric operational forecasts (2-km resolution) demonstrated significant strengths over the existing system. In a series of simulated fire episodes, predictions from the old model and the new model are compared against satellite-derived burnt scar maps to evaluate their performance. The new system is expected to operate in a pseudo-operational mode alongside the existing service during the 2025 fire season and to fully replace the operational fire-spread model by 2026.

How to cite: Bartsotas, N. S., Herekakis, T., Girtsou, S., and Kontoes, C.: First results from the implementation of a new fire-spread model in FireHUB platform, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9693, https://doi.org/10.5194/egusphere-egu25-9693, 2025.

Freeze hazard represents the costliest peril associated with winter weather in the United States (US). This study focuses on the development and validation of a Freeze Index (FI) to model the impact of freeze effectively. The FI integrates both the intensity and duration of freeze events, offering a more accurate modelling of freeze hazards. The updated FI is used to select US-wide events targeting mainly the spatial scale of cold air outbreaks (CAOs). Validation of the hazard footprints is performed against historical data, including the December 2022 CAO and the Texas freeze of 2021. The findings underscore the importance of considering both temperature and duration in freeze hazards to model the damages accurately.

The freeze events obtained above are used to investigate trends in duration and FI, using 2-metre temperature (T2M) from the reanalysis data (1950 – 2024) and compared with the events from the detrended T2M. In the detrended set, no significant trend is observed in the duration of events from 1950 onwards. The average FI obtained from the footprints of each event also did not show a significant trend. The freeze events obtained from the non-detrended T2M also do not show a significant trend in duration and average FI for the events. However, there is a clear decrease in the occurrence of long-duration events with only four events greater than 10 days from 1990 onwards compared to thirteen events in the 1950 – 1985 period.

How to cite: Ali, M., Ait-Chaalal, F., Dobbin, A., and Grieser, J.: Modelling Freeze Hazard for the North American Winters , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9897, https://doi.org/10.5194/egusphere-egu25-9897, 2025.

Historical cyclone data indicate significant variations in cyclone activity during different phases of the El Niño-Southern Oscillation (ENSO). However, the impact of these variations on cyclone risk and damage has not been thoroughly investigated due to limited historical loss record. Understanding these variations could be crucial for effective risk management.

This study examines the variation in cyclone risk associated with ENSO phases, utilizing the cyclone risk assessment model by Impact Forecasting for Australia. The model employs a stochastic event set of cyclones, representing about forty-two thousand years of basin-wide activity, developed using environmental data from reanalysis and machine learning techniques. Our analysis demonstrate that the stochastic event set accurately reflects the seasonal variation in cyclone activity due to ENSO phases, making it a reliable tool for risk assessment.

To evaluate risk by ENSO phases, we segregated the stochastic event set using the Oceanic Nino Index and estimated wind-driven losses for each phase. The model results shows a significant variation in cyclone risk in Australia during El Niño and La Niña. However, the risk during the Neutral phase is found to be comparable with the long-term average. Annual average losses (AAL) during La Niña increases by 40%, while El Niño phases show a 37% reduction compared to the long-term average. Additionally, a one-in-hundred-year event during La Niña can result in 21% higher losses, whereas losses are 28% lower during El Niño compared to the long-term average.

The modeled loss variations across ENSO phases are consistent with observed changes in cyclone activity in Australia and are supported by the historical loss records.

How to cite: Bongirwar, V., Abraham Joseph, L., Ranjan Tripathy, R., Martin Kalbermatter, D., Roy, T., and Yang, P.: Modeling cyclone risk variations in Australia by ENSO phases., EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-10365, https://doi.org/10.5194/egusphere-egu25-10365, 2025.

Climate change has modified climatic hazards features and requires to reconsider agro-climatic risks. Among these, drought is one of the risks with the strongest impact on both crop production and crop weather insurance performance (Brisson et al., 2010). Understanding the effects of climate change on agro-climatic risks at regional to local scale is therefore a major challenge for the agricultural sector, specifically for insurers offering crop weather insurance policies. This work, resulting from a collaboration between an insurer and a research laboratory, focuses on the development of a drought index that well explain the evolution of crop weather insurance loss ratio. As maize is a major crop in the company's portfolio, the study focuses on this crop in particular. The aim of this work is to find the optimal set of parameters that maximizes the correlation between the drought index and the drought-related losses on crop weather insurance.

The Safran-Isba-Modcou reanalysis produced by Météo France provides spatially and temporally continuous climate data over metropolitan France of relevant interest to address this topic (Le Moigne et al., 2020; Soubeyroux et al., 2008). At the regional scale, these data allow us to quantify the evolution of climate hazards related to the water cycle from 1960 to present day. Taking into account the vulnerability of the crop of interest through the use of a simplified two reservoirs water balance model provides an opportunity to assess changes in maize water stress (Jacquart and Choisnel, 1995). The definition of a water stress threshold leads to the development of an annual drought index (Laurent et al., under review). The correlation with the crop weather insurance loss ratio due to drought is tested at various spatial scales (municipality, production basin), for different varieties, different sowing dates and different stress thresholds.

Our results indicate that climate change has affected the frequency and intensity of drought risk on maize crops in France, depending on the French production area studied. The significance of the correlation depends on maize variety, sowing date and hydric stress threshold. It seems that using drought index computed with low stress thresholds and analyzing correlations at large spatial scales gives the best results.

For non-irrigated maize area at production basin scale, our drought index can explain a significant part of drought-related losses in crop weather insurance. The results suggest that such an index may be relevant to improve the actuarial loss model of the insurer. However, further analysis is required in areas where correlations are weaker, particularly in production basins with high irrigation levels.

References:

Brisson et al., 2010. Field Crops Res. 119, 201–212. https://doi.org/10.1016/j.fcr.2010.07.012
Jacquart, Choisnel, 1995. La Météorologie 8ème série, 29–44. https://doi.org/10.4267/2042/51939
Laurent et al., under review. J. Agric. For. Meteorol.
Le Moigne et al., 2020. Geosci. Model Dev. 13, 3925–3946. https://doi.org/10.5194/gmd-13-3925-2020
Soubeyroux et al., 2008. La Météorologie 8, 40. https://doi.org/10.4267/2042/21890

How to cite: Laurent, L., Ullmann, A., and Castel, T.: How drought risk evolution impacts crop weather insurance loss ratio in France?, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-10429, https://doi.org/10.5194/egusphere-egu25-10429, 2025.

Clusters of storms are defined as sequences of multiple storms occurring within a short time frame and a limited spatial extent. In this study, storm clusters are identified using a Lagrangian approach combined with an absolute frequency metric within a 96-hour time window, reflecting reinsurance contract specifications for an insurance company. Compound storms are further constrained to affect a common area, determined by the intersection of their footprints. Those footprints can be delineated using various radii of different sizes, depending on the desired granularity for compounding analysis.

The motivation for this definition stems from the potentially severe impacts of such events on the insurance sector. Storms are known to be among the costliest events for Insurance in Europe, with an average annual insured loss of €217 billion [Copernicus, 2023]. The repetition of such intense wind and strong precipitation events is no exception. The successive storms Lothar and Martin in December 1999 remain the costliest events observed in France with an estimated loss of €17 billion [EEA, 2023]. Despite the substantial risks associated with these compound events, few studies have investigated their role in amplifying both the hazard and the vulnerability.

We apply this approach to Generali, an Italian insurance company with approximately 5% market share in France. Using Generali’s historical claims data from 1998 to 2024, we propose a novel methodology linking high-resolution claims to individual storm events. This approach represents a significant advance in understanding loss drivers. Applied to storm clusters, the methodology distinguishes the relative contribution of each storm in a cluster to the total observed loss. By comparing the findings with Generali’s portfolio from 2018 to 2024, we identify key factors contributing to the additional damages caused by storm clusters. These insights are crucial for enhancing risk prevention and adapting current insurance strategies to better address compound storm events.

How to cite: Hasbini, L., Yiou, P., and Boissier, L.: Analysis of the insurance impacts of storm clusters: a case study with Generali France, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-11503, https://doi.org/10.5194/egusphere-egu25-11503, 2025.

Storm surge, a coastal flooding phenomenon driven by high-speed winds pushing water onshore poses a significant natural hazard across the globe. In recent decades, Europe has experienced several destructive extratropical cyclones that have severely impacted coastal communities and economies, such as Eunice (2022), David (2018), or Xaver (2013). Storm Xynthia in 2010 was especially notable, with substantial fatalities and material losses in France, highlighting the need for accurate storm surge risk assessment for societies and the (re)insurance industry involved. Yet, current modelling solutions are limited. Main commercial models only provide partial coverage of the risk in Europe, with a primary focus on the United Kingdom. To address this gap, AXA proposes a scenario-based approach to assess storm surge risk across North-Western Europe. Using the SCHISM 2D hydrodynamic model, we reproduced 10 significant historical events notably affecting France, Germany, and the United Kingdom, then perturbed them along three parameters: wind speeds, storm sizes and tide timings, generating 480 scenarios. The study presents the challenges of scenario selection and variability representation. It further provides findings on the modelling results by parameter and country, and on the estimation of the loss potential using a representative North-Western Europe insured market portfolio. Finally, key limitations are discussed, related to unmodelled defences and Digital Elevation Model accuracy. The approach provides valuable insights for AXA’s risk assessment and is a crucial step towards building a robust understanding of our risk.

How to cite: Diouf, A., Pranantyo, I. R., Joffrain, M., and Bruneau, N.: Designing representative European storm surge scenarios for insurance risk assessment: challenges, results, and limitations, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-17961, https://doi.org/10.5194/egusphere-egu25-17961, 2025.

Due to intense destructive winds and heavy rainfall associated with storm surges, large waves and flooding, tropical cyclones are one of the most damaging natural catastrophes. They are a major threat to human lives and properties across the globe. When travelling over the ocean and approaching shallow water regions, tropical cyclones generate storm surge and waves that can devastate coastal communities and local economies.

In the recent years, Typhoons Hato (2017) and Mangkhut (2018) produced material surge damages to insurers in the Northwest Pacific basin, and therefore raised the need for accurate natural catastrophe models. Cat models consist of very large catalogues of synthetic but realistic events also called “event sets”. These event sets are consistent with experienced historical data but allow extrapolation beyond what was observed. 

In this study, we focus in winds and surges on the Philippines and Hong Kong regions. Driven by an existing tropical cyclone wind event set, over 10k full-physic simulations of storm surge and waves are computed for each region to estimate the complete distribution of coupled wind and surge losses over an exposure dataset. Due to computationally expensive dynamical simulations of storm surges and waves,  we first rank and select a subset of events (10k) based on an IKE (Integrated Kinetic energy) index. For each of these 10k event, the Semi-implicit Cross-scale Hydroscience Integrated System Model (SCHISM; Zhang & Baptista, 2008, Zhang et al., 2016) is forced by atmospheric winds and pressure fields to derive wave and surge footprints.

Second, we use adjusted Hazus (FEMA) damage functions to convert the water heights and windspeeds from the simulated events into damage factors. These factors are then multiplied to the considered exposure to derive losses. Third, we study the relationship between the wind and the surge modeled losses based on two criteria, (i) the event level correlation between IKE and surge losses, to ensure this index stands as a robust risk representation, and (ii) the event level proportion of surge losses out of the wind losses, which provides a set of reusable inter perils correlation factors.

How to cite: Joffrain, M., Pranantyo, I. R., and Bruneau, N.: Evaluating the relationship between wind and storm surge risk in the Philippines and Hong-Kong, an insurance industry perspective., EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18013, https://doi.org/10.5194/egusphere-egu25-18013, 2025.

How to cite: Azemar, F., Shaylor, M., Bruneau, N., Loridan, T., Swain, D., and Joffrain, M.: Development of a climate-driven stochastic event catalogue for Wildfire in Europe, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18168, https://doi.org/10.5194/egusphere-egu25-18168, 2025.

With global temperatures continuing to rise year on year, drought conditions are becoming increasingly frequent and severe, across all continents. More and more, the negative effects of these worsening drought conditions are being experienced by people across the world both directly, through damage to agricultural systems, water scarcity or damage to homes from subsidence, as well as indirectly, through cascading effects on other perils such as heatwaves and wildfires, which in turn may devastate communities and drive great economic losses. For these reasons, drought is of growing concern to the (re)insurance industry, as an emerging peril. It is therefore essential that reinsurers have access to tools which can aid in their understanding of drought hazard and risk in a changing climate. One such tool we present here – a climate driven, globally connected stochastic drought hazard model, which responds dynamically to the climate of any given year, enabling this understanding of how drought conditions change with the climate.

In this presentation, we describe the novel methodology applied to generate this globally connected and climate-driven stochastic drought model. The model is generated in two stages, the first addressing global variability in drought trends and teleconnections, and the second looking at continental scale patterns. In the first instance, we apply a dimensionality reduction to a selection of historical drought indexes over different time scales, allowing extraction of the key modes of variability of drought at the global scale. We then condition the top key modes of variability to the climate state using reanalysis (ERA5) data, allowing us to drive our stochastic set at the global scale, based on the global climate state.

Once these global patterns have been determined, we use the residual drought signal to condition a regional (continental) model using similar reduction and conditioning techniques. This regional layer is then effectively layered onto the global model, allowing us to recreate globally and regionally consistent drought variability in the stochastic set. A Bayesian framework is used to sample a range of realistic drought conditions, aligned with the climate of any given year. Global and regional drought conditions are then combined in order to generate >100K stochastic years of global drought severity as well as duration of drought for three severity levels (moderate, severe, extreme). This framework can also be applied to any other climate model data (for example, CESM LENS2) to generate a stochastic event set up to the year 2100. Here we present initial results from this stochastic catalogue, showcasing the spatial and temporal variation in drought hazard from 1950 – 2100, return periods, and comparisons to historical records. This work also builds upon a previous, continental only version of the drought model.

How to cite: Shaylor, M., Bruneau, N., Azemar, F., and Loridan, T.: Development of a Globally Connected, Climate-Driven, Stochastic Drought Model for Hazard Assessment using Machine Learning Techniques, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18838, https://doi.org/10.5194/egusphere-egu25-18838, 2025.

Diagnosing the drivers of changing insured losses year on year is an important component of developing a sustainable insurance portfolio. The common assumption is that losses for most perils are increasing year on year. However, there are many factors that could drive the change in losses: economic versus insured losses, impacts of inflation, changes in societal wealth over time, movement toward riskier property locations as well as potential changes in the frequency and severity of European wind and flood events. This presentation will quantify each of the above factors to determine the drivers of changes in insured losses over time.

How to cite: Milner, C. and Mulder, K.: Insured Losses from European Natural Catastrophes: Is there a trend over time?, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7007, https://doi.org/10.5194/egusphere-egu25-7007, 2025.