Probabilistic Risk Model Research Articles

Adaptive risk prediction system with incremental and transfer learning

Currently, popular methods for prenatal risk assessment of fetal aneuploidies are based on multivariate probabilistic modelling, that are built on decades of scientific research and large-scale multi-center clinical studies. These static models that are deployed to screening labs are rarely updated or adapted to local population characteristics. In this article, we propose an adaptive risk prediction system or ARPS, which considers these changing characteristics and automatically deploys updated risk models.8 years of real-life Down syndrome screening data was used to firstly develop a distribution shift detection method that captures significant changes in the patient population and secondly a probabilistic risk modelling system that adapts to new data when these changes are detected. Various candidate systems that utilize transfer -and incremental learning that implement different levels of plasticity were tested.Distribution shift detection using a windowed approach provides a computationally less expensive alternative to fitting models at every data block step while not sacrificing performance. This was possible when utilizing transfer learning. Deploying an ARPS to a lab requires careful consideration of the parameters regarding the distribution shift detection and model updating, as they are affected by lab throughput and the incidence of the screened rare disorder. When this is done, ARPS could be also utilized for other population screening problems.We demonstrate with a large real-life dataset that our best performing novel Incremental-Learning-Population-to-Population-Transfer-Learning design can achieve on par prediction performance without human intervention, when compared to a deployed risk screening algorithm that has been manually updated over several years.

Scenario-based economic and societal risk assessment of storm flooding in Shanghai

PurposeQuantitative and spatial-explicit flood risk information is of great importance for strengthening climate change adaptation and flood resilience. Shanghai is a coastal megacity at large estuary delta with rising flood risks. This study aims to quantify the overall economic-societal risks of storm flooding and their spatial patterns in Shanghai.Design/methodology/approachBased on multiple storm flood scenarios at different return periods, as well as fine-scale data sets including gridded GDP, gridded population and vector land-use, a probabilistic risk model incorporating geographic information system is used to assess the economic-societal risks of flooding and their spatial distributions.FindingsOur results show that, from 1/200 to 1/5,000-year floods, the exposed assets will increase from USD 85.4bn to USD 657.6bn, and the direct economic losses will increase from USD 3.06bn to USD 52bn. The expected annual damage (EAD) of assets is around USD 84.36m. Hotpots of EAD are mainly distributed in the city center, the depressions along the upper Huangpu River in the southwest, the north coast of Hangzhou Bay, and the confluence of the Huangpu River and Yangtze River in the northeast. From 1/200 to 1/5,000-year floods, the exposed population will rise from 280 thousand to 2,420 thousand, and the estimated casualties will rise from 299 to 1,045. The expected annual casualties (EAC) are around 2.28. Hotspots of casualties are generally consistent with those of EAD.Originality/valueIn contrast to previous studies that focus on a single flood scenario or a particular type of flood exposure/risk in Shanghai, the findings contribute to an understanding of overall flood risks and their spatial patterns, which have significant implications for cost-benefit analysis of flood resilience strategies.

Featured Cover

The cover image is based on the Original Article Incorporation of technical, human and organizational risks in a dynamic probabilistic fire risk model for high-rise residential buildings by Samson Tan et al., https://doi.org/10.1002/fam.2872.

Improving astronaut cancer risk assessment from space radiation with an ensemble model framework

A new approach to NASA space radiation risk modeling has successfully extended the current NASA probabilistic cancer risk model to an ensemble framework able to consider sub-model parameter uncertainty as well as model-form uncertainty associated with differing theoretical or empirical formalisms. Ensemble methodologies are already widely used in weather prediction, modeling of infectious disease outbreaks, and certain terrestrial radiation protection applications to better understand how uncertainty may influence risk decision-making. Applying ensemble methodologies to space radiation risk projections offers the potential to efficiently incorporate emerging research results, allow for the incorporation of future models, improve uncertainty quantification, and reduce the impact of subjective bias. Moreover, risk forecasting across an ensemble of multiple predictive models can provide stakeholders additional information on risk acceptance if current health/medical standards cannot be met for future space exploration missions, such as human missions to Mars. In this work, ensemble risk projections implementing multiple sub-models of radiation quality, dose and dose-rate effectiveness factors, excess risk, and latency are presented. Initial consensus methods for ensemble model weights and correlations to account for individual model bias are discussed. In these analyses, the ensemble forecast compares well to results from NASA's current operational cancer risk projection model used to assess permissible mission durations for astronauts. However, a large range of projected risk values are obtained at the upper 95th confidence level where models must extrapolate beyond available biological data sets. Closer agreement is seen at the median ± one sigma due to the inherent similarities in available models. Identification of potential new models, epidemiological data, and methods for statistical correlation between predictive ensemble members are discussed. Alternate ways of communicating risk and acceptable uncertainty with respect to NASA's current permissible exposure limits are explored.

Eliciting Model Structures for Multivariate Probabilistic Risk Analysis

Notionally objective probabilistic risk models, built around ideas of cause and effect, are used to predict impacts and evaluate trade-offs. In this paper, we focus on the use of expert judgement to fill gaps left by insufficient data and understanding. Psychological and contextual phenomena such as anchoring, availability bias, confirmation bias and overconfidence are pervasive and have powerful effects on individual judgements. Research across a range of fields has found that groups have access to more diverse information and ways of thinking about problems, and routinely outperform credentialled individuals on judgement and prediction tasks. In structured group elicitation, individuals make initial independent judgements, opinions are respected, participants consider the judgements made by others, and they may have the opportunity to reconsider and revise their initial estimates. Estimates may be aggregated using behavioural, mathematical or combined approaches. In contrast, mathematical modelers have been slower to accept that the host of psychological frailties and contextual biases that afflict judgements about parameters and events may also influence model assumptions and structures. Few, if any, quantitative risk analyses embrace sources of uncertainty comprehensively. However, several recent innovations aim to anticipate behavioural and social biases in model construction and to mitigate their effects. In this paper, we outline approaches to eliciting and combining alternative ideas of cause and effect. We discuss the translation of ideas into equations and assumptions, assessing the potential for psychological and social factors to affect the construction of models. We outline the strengths and weaknesses of recent advances in structured, group-based model construction that may accommodate a variety of understandings about cause and effect.

Assessing regional risk of COVID-19 infection from Wuhan via high-speed rail.

This paper demonstrates that transportation networks may be used to assess and predict the regional risk of COVID-19 infection from the outbreak. We use China's high-speed rail (HSR) network at the scale of prefecture level to assess, based on a probabilistic risk model, the risk of COVID-19 infection from Wuhan to the country's 31 province-level regions at the early stage of domestic spread. We find that the high-risk regions are mainly distributed along the southern half of Beijing-Hong Kong HSR line, where a large number of infection cases have been confirmed at the early stage. Furthermore, the two components of the infection risk, namely, the probability (proxied by the region's correlation with Wuhan through HSR) and the impact (proxied by the region's population with mobility), can play different roles in the risk ranking for different regions. For public health administrators, these findings may be used for better decision making, including the preparation of emergency plans and supplies, and the allocation of limited resources, before the extensive spread of the epidemic. Moreover, the administrators should adopt different intervention measures for different regions, so as to better mitigate the epidemic spread according to their own risk scenarios with respect to the probability of occurring and, once occurred, the impact.

Developing an integrated framework for source apportionment and source-specific health risk assessment of PAHs in soils: Application to a typical cold region in China

Here, a new integrated methodology framework has been proposed for source apportionment and source-oriented risk evaluation, and applied to identify the characteristics, sources and health risks of PAHs in the soils of a typical cold region in Northeastern China. To this end, a large-scale data set containing 1780 soil samples and 16 priority PAHs has been collected from the study area. Two advanced receptor models, positive matrix factorization (PMF) and multivariate curve resolution-weighted alternating least-squares (MCR-WALS), have been comparatively employed to apportion the pollution sources of soil PAHs, with the help of a set of modified literature PAH source fingerprints. Further, the apportionment results have been incorporated into a probabilistic incremental lifetime cancer risk model for assessing the source-specific health risk of soil PAHs. Notably, the PMF and MCR-WALS models have apportioned essentially same results. The coal combustion and gasoline engine are identified as the main contributors of soil PAHs, with contributions of 57.9–58.1% and 25.2–22.2%, respectively. The health risks posed by PAHs in the soils are negligible for both adult and children; relatively, source-oriented risk assessment shows coal combustions make the largest contribution to the total risk of PAHs (56.1%), followed by gasoline engine (22.5%) and coke oven (21.4%).

Probabilistic health risk assessment of exposure to carcinogens of Chinese family cooking and influence analysis of cooking factors.

Cooking oil fume (COF) have adverse health effects for people. A probabilistic health risk assessment model with risk parameters as random variables considering the differences in exposure concentration and exposure time of different cooking event was proposed to assess the inhalational incremental lifetime cancer risk (ILCR). The exposure of carcinogens such as benzene, formaldehyde, PM2.5-bound polycyclic aromatic hydrocarbons (PPAHs) and PM2.5-bound heavy metals (PHMs) of Chinese family cooking was studied and the exposure concentrations of carcinogens were predicted by computational fluid dynamics (CFD). In addition, the influence of five key cooking factors (cooking method, the weight of ingredients (meat and vegetables), type of meat, ratio of meat to vegetables, and type of oil) that affect the generation of COF was explored. The ILCR of COF is assessed comprehensively in present study by the probabilistic health risk assessment model. The result showed that the sum of the risks of assessed carcinogens (total ILCR of COF) determined by Monte Carlo simulation method with a 95% confidence interval (95%CI) is 2.45×10-4 to 1.61×10-3, which far exceeds the acceptable limit of 1.00×10-6. Generally, the ILCR of assessed carcinogens decreases in the following order: PHMs [ILCR (95%CI): 2.08×10-4 to 1.54×10-3]>formaldehyde [ILCR (95%CI): 9.04×10-6 to 6.87×10-5] and PPAHs [ILCR (95%CI): 5.97×10-6 to 4.51×10-5]>benzene [ILCR (95%CI): 2.99×10-7 to 3.00×10-6]. The results indicated that more attention should be paid to the ILCR of PM2.5. Cooking method significantly affect the ILCR of carcinogens in COF excluding formaldehyde. The ILCRs of COF from water-based cooking methods are greater than those of oil-based cooking ones.

Estimates of lung burden risk associated with long-term exposure to TiO2 nanoparticles as a UV-filter in sprays.

Titanium dioxide (TiO2) nanoparticles (NPs) are employed as an ultraviolet filter in sunscreen products because of their high ultraviolet absorptivity. However, sunscreen sprays may pose health risks due to the toxicity of inhaled TiO2 NPs. Therefore, we estimated the potential human health risk posed by inhaled TiO2 NPs emitted from sunscreen sprays. The physiology-based lung model was employed to predict the lung TiO2 NPs burden caused by long-term exposure. A Hill-based dose-response model described the relationship between lung inflammation and TiO2 NP accumulation. The Weibull threshold model was used to estimate the threshold amount of accumulation inducing 0.5% of the maximum increase in neutrophils. The potential health risk was assessed using a hazard quotient-based probabilistic risk model. All data obtained to date indicate that application of sunscreen sprays poses no significant health risk. However, using data simulations based on the threshold criterion, we discovered that in terms of practical strategies for preventing the risks posed by inhaled TiO2 NPs emitted from spray products, the suggested daily use amount and pressing number are 40 g (95% confidence interval: 11-146 g) and 66 (18-245), respectively. In this study, we successfully translated the potential health risk of long-term exposure to NP-containing sunscreen sprays and recommendations for daily application into mechanistic insights.

Modelling the dynamics of the cancer risk due to potentially toxic elements in agricultural soils, in the upper Crocodile River catchment, North-West province, South Africa

Contaminated agricultural soil with potentially toxic elements (PTEs) poses a threat to food safety and can adversely affect human health. This study evaluates the potential health risk caused by PTEs contamination in cultivated soils and the possible related health effect. Sixty-four (64) soil samples from cultivated soils before crop cultivation and post-crop were harvested and twenty-four (24) soil samples from a natural game reserve park were collected and. The PTEs were analysed using the inductively coupled plasma-mass spectrometry (ICP-MS) for the following elements As, Cd, Cr, Cu, Ni, Pb, Zn, Mn, Al, and Fe. The contamination factor (CF), pollution load index (PLI), hazard quotient (HQ), and hazard index (HI) was calculated to assess the human health risks for the different pathways to the exposed population. A carcinogenic probabilistic risk model based on the mathematical derivation and PTEs transmission pathways is presented. The mean concentrations of the PTEs in the cultivated soil varied in order of Al > Fe > Mn > Cr > Ni > Zn > Cu > Pb > As > Cd. Cr and Ni concentrations in all the cultivated soils were found to be higher than the Canadian guidelines and the Finnish permissible threshold guidelines for agricultural. The CF in the soil indicates extreme contamination suggestion an ecotoxicological effect. The PLI value also suggests that the soils have undergone some form of deterioration, particularly with Cr and Ni. The ingestion seems to be the major pathway followed by dermal to children. The HQ and HI values for the children were > 1 indicating an adverse health effect for the children residing around the cultivated fields. The simulated results indicate that the percentage of the affected humans that may be consuming crop derived from the contaminated soil increases by a factor of 2 as the contamination factor increases. The transmission compartments can be used as an effective measure in which mitigation can be effectively used.

Comparing an insurer's perspective on building damages with modelled damages from pan-European winter windstorm event sets: a case study from Zurich, Switzerland

Abstract. With access to claims, insurers have a long tradition of being knowledge leaders on damages caused by windstorms. However, new opportunities have arisen to better assess the risks of winter windstorms in Europe through the availability of historic footprints provided by the Windstorm Information Service (Copernicus WISC). In this study, we compare how modelling of building damages complements claims-based risk assessment. We describe and use two windstorm risk models: an insurer's proprietary model and the open source CLIMADA platform. Both use the historic WISC dataset and a purposefully built, probabilistic hazard event set of winter windstorms across Europe to model building damages in the canton of Zurich, Switzerland. These approaches project a considerably lower estimate for the annual average damage (CHF 1.4 million), compared to claims (CHF 2.3 million), which originates mainly from a different assessment of the return period of the most damaging historic event Lothar–Martin. Additionally, the probabilistic modelling approach allows assessment of rare events, such as a 250-year-return-period windstorm causing CHF 75 million in damages, including an evaluation of the uncertainties. Our study emphasizes the importance of complementing a claims-based perspective with a probabilistic risk modelling approach to better understand windstorm risks. The presented open-source model provides a straightforward entry point for small insurance companies.

Environmental risk assessments for contrasting Arctic shelf ecosystems

The article presents environmental risk assessments for two types of contrasting ecosystems of the Arctic shelf: highly productive and low-productive. Calculations were carried out using a probabilistic risk model with variations in the impact on the ecosystem from stressors, taking into account the risks from the technological processes of extraction and transportation of mineral resources of the shelf. Comparison of the results of risk modeling for highly productive and low-productive ecosystems showed that due to the low productivity of marine ecosystems, reliability requirements for technological processes ensuring the safety of resource development should not be reduced. The nontrivial conclusion obtained confirms the nonlinearity of the ecosystem response to impacts, which justifies the involvement not only of expert assessments, but also of all available methods, including modeling, to minimize the risks of Arctic resource development.

Watershed-scale assessment of surface water-related risks from shale gas development in mountainous areas, China

Water risks are one of the key issues dominating environmental debates on shale gas development. Water withdrawals and wastewater discharges in shale gas fields of mountainous areas are more complicated than in plain areas due to different climatic, topographical and hydrological conditions, which would impact water resources. This research identifies the surface water-related risks from shale gas development in mountainous areas as water shortage and water pollution. Conceptions of accessibility for both water supply and water pollution are proposed to describe the vulnerability of water resources and the exposure to water pollution. Based on a risk probability model, a water risk assessment method for mountainous areas is constructed from the perspectives of dangers, exposures and vulnerabilities. Finally, the assessment method is applied in Chongqing, China. The results show that, from 2010 to 2020, the water consumption of shale gas development has a little impact on regional water resources in total, but more significant impacts are seen in a few areas, including the seasonal water-deficient areas in Western Chongqing, the urban and suburban areas with high pollutant loadings in Midwest Chongqing, and other areas with high pollutant accessibility and vulnerable water environments. The surface water-related risks of the shale gas development in Chongqing are principally composed of low and relatively low levels of risks, which cover 60% of the total area of Chongqing and display a spatial difference of west > northeast > southeast areas. Based on Monte Carlo method, the results of uncertainty analyses show the model is reliable. This research provides a reference for water comprehensive risk assessment of shale gas development in mountainous areas.

Use and Misuse of MCDA to Support Decision Making Informed by Risk.

Recent guidelines for risk-informed decision making (RIDM) provide a gold-standard for how to incorporate probabilistic risk models in conjunction with other considerations in a decision process. Nevertheless, risk quantification using probabilistic and statistical methods is difficult in situations where threat, vulnerability, and consequences are highly uncertain and risk quantification. In such situations a wider variety of methods could be employed, which we call decision making informed by risk (DMIR) combining risk and decision analytics. Risk informed decision making (RIDM) can be considered as a special case of DMIR. Multi criteria decision analysis (MCDA) often serves as a basis for DMIR in order to flexibly accommodate different levels of analytical detail. DMIR often involves artful use of proxy variables that correlate with, and are more measurable than, underlying factors of interest. This article introduces the notion of DMIR and discusses the use of MCDA in its application in the context of risk-based problems. MCDA-based risk analyses identify metrics associated with threats of concern and system vulnerabilities, characterize the way in which alternative actions can affect these threats and vulnerabilities, and ultimately synthesize this information to compare, prioritize, or select alternative mitigation strategies. Simple linear additive MCDA models often integrate these inputs, but the same simplicity can limit such approaches and create pitfalls and more advanced models including multiplicative relationships can be warranted. This essay qualitatively explores the critical practitioner questions of how and when the use of linear multicriteria models creates significant problems, and how to avoid them.

Probabilistic seismic risk assessment of India

This study presents a comprehensive open probabilistic seismic risk model for India. The proposed model comprises a nationwide residential and non-residential building exposure model, a selection of analytical seismic vulnerability functions tailored for Indian building classes, and the open implementation of an existing probabilistic seismic hazard model for India. The vulnerability of the building exposure is combined with the seismic hazard using the stochastic (Monte Carlo) event-based calculator of the OpenQuake engine to estimate probabilistic seismic risk metrics such as average annual economic losses and the exceedance probability curves at the national, state, district, and subdistrict levels. The risk model and the underlying datasets, along with the risk metrics calculated at different scales, are intended to be used as tools to quantitatively assess the earthquake risk across India and also compare with other countries to develop risk-informed building design guidelines, for more careful land-use planning, to optimize earthquake insurance pricing, and to enhance general earthquake risk awareness and preparedness.

Accounting for directivity-induced pulse-like ground motions in building portfolio loss assessment

Earthquake-induced pulse-like ground motions are often observed in near-source conditions due to forward-directivity. Recent worldwide earthquakes have emphasised the severe damage potential of such pulse-like ground motions. This paper introduces a framework to quantify the impact of directivity-induced pulse-like ground motions on the direct economic losses of building portfolios. To this aim, a simulation-based probabilistic risk modelling framework is implemented for various synthetic building portfolios located either in the fault-parallel or fault-normal orientations with respect to a case-study strike–slip fault. Three low-to-mid-rise building typologies representative of distinct vulnerability classes in the Mediterranean region are considered: non-ductile moment-resisting reinforced concrete (RC) frames with masonry infills, mainly designed to only sustain gravity loads (i.e. pre-code frames); moment-resisting RC infilled frames designed considering seismic provisions for high ductility capacity (i.e. special-code frames); special-code steel moment-resisting frames. Monte Carlo-based probabilistic seismic hazard analysis is first performed, considering the relevant modifications to account for the pulse-occurrence probability and the resulting spectral amplification. Hazard curves for sites/buildings located at different distances from the fault are obtained, discussing the spatial distribution of the hazard amplification. A set of pulse-like ground motions and a set of one-to-one spectrally-equivalent ordinary records are used to perform non-linear dynamic analysis and derive fragility relationships for each considered building typology. A vulnerability model is finally built by combining the derived fragility relationships with a (building-level) damage-to-loss model. The results are presented in terms of intensity-based and expected annual loss for synthetic portfolios of different sizes and distribution of building types. It is shown that, for particularly short-period structures (e.g. infilled RC frames), the influence of near-source directivity can be reasonably neglected in the fragility derivation while kept in place in the hazard component. Overall, near-source directivity effects are significant when estimating losses of individual buildings or small portfolios located very close to a fault. Nevertheless, the impact of pulse-like ground motions on losses for larger portfolios can be considered minimal and can be neglected in most of the practical large-scale seismic risk assessment applications.

Complexity, Power Laws and a Humean Argument in Risk Management: The Fundamental Inadequacy of Probability Theory as a Foundation for Modeling Complex Risk in Banking

Whenever risk managers are confronted with deep uncertainty and organized complexity, probabilistic inference methods are used. These seem able to allow for crisp inputs and precise results. However, as has been noted by several thinkers (e.g., von Hayek in Studies in philosophy, politics and economics, Routledge, London, 1967; Weaver in American Scientist, 36: 536–544, 1948), such methods cannot be used effectively in such situations. This might basically sound like old wine in a new bottle and, in fact, objections to, and limitations of conventional, i.e., probabilistic risk modeling are anything but unheard of in the literature of banking and finance. However, this paper introduces for the first time an argument, inspired by reflections on the old riddle of induction, from which those shortcomings of the limited suitability of probability can be derived. It demonstrates that any choice of a particular probability distribution for a given risk management purpose is necessarily arbitrary, i.e., it is not grounded in the data but in the choices of the statistician or risk manager, and cannot be justified by appealing to something more objective. Thereby, this paper unmasks the illusion that financial data and extreme losses are well-described by non-standard probability functions such as power laws that have been embraced at the expense of bell curves in the aftermath of the global financial crisis of 2008. Moreover, although we do not propose a positive solution, we believe that articulating the real, and as yet unnoticed, source of the problem is a key step towards developing a principle and tractable response.

Incorporation of technical, human and organizational risks in a dynamic probabilistic fire risk model for high‐rise residential buildings

SummaryWhile probabilistic risk assessment (PRA) is an explicit methodology for complying with the performance requirements of the Building Code of Australia (BCA) or similar codes, it traditionally focuses only on technical risks of fire safety systems in a building. There are growing concerns that performance‐based fire engineering designs underestimate safety risk levels in high‐rise residential buildings. Existing fire risk models account for failures of technical systems but ignore human and organizational errors (HOEs) and the complex interactions among these variables. Probabilistic models in other applications, such as offshore platforms and nuclear plants, demonstrate the importance of HOE inclusion in risk models and the resulting impacts on overall risk. This paper proposes a comprehensive technical‐human‐organizational risk (T‐H‐O‐Risk) methodology to enhance the PRA approach by quantifying human and organizational risks in a probabilistic model using Bayesian Network (BN) analysis of HOEs and System Dynamics (SD) modelling for dynamic characterization of risk variations over time. While risk modelling itself is not novel, the current research develops unique and specific enhancements to existing risk approaches by integrating HOE risks with technical risks in a comprehensive dynamic and probabilistic model for high‐rise residential buildings. Three case studies are conducted to demonstrate the application of this comprehensive approach to the designs of various high‐rise residential buildings ranging from 18 to 24 storeys. Societal risks are represented in F‐N curves. Results show that in general, fire safety designs that do not consider HOEs underestimate overall risks generally by~20%—and can reach up to 42% in an extreme case. Furthermore, risks over time due to HOEs vary by as much as 30% over a 10‐year period. A sensitivity analysis indicates that deficient training, poor safety culture and ineffective emergency plans have significant impact on overall risk.

Probabilistic risk modelling for port investments: A practical approach

Economic and financial analyses of port projects are essential to ensure project viability in terms of defined objectives, adequate returns to private and public stakeholders, and to society in general. Moreover, including early-stage risk analysis ensures better decision making and better outcomes. However, project risk management is not very frequently used in port infrastructure projects. To improve this area and to ensure the kind of transparency, with clear figures, that would make ports attractive to private investors, a general methodology is described for including risk analysis in an economic-financial analytical model of a port investment project that considers probabilistic distributions for selected input and output variables to yield more information than could be obtained from a deterministic model. The proposed methodology is illustrated by applying it to a port terminal investment project by the Cartagena Port Authority (Spain).

Event generation for probabilistic flood risk modelling: multi-site peak flow dependence model vs. weather-generator-based approach

Abstract. Flood risk assessment is an important prerequisite for risk management decisions. To estimate the risk, i.e. the probability of damage, flood damage needs to be either systematically recorded over a long period or modelled for a series of synthetically generated flood events. Since damage records are typically rare, time series of plausible, spatially coherent event precipitation or peak discharges need to be generated to drive the chain of process models. In the present study, synthetic flood events are generated by two different approaches to modelling flood risk in a meso-scale alpine study area (Vorarlberg, Austria). The first approach is based on the semi-conditional multi-variate dependence model applied to discharge series. The second approach relies on the continuous hydrological modelling of synthetic meteorological fields generated by a multi-site weather generator and using an hourly disaggregation scheme. The results of the two approaches are compared in terms of simulated spatial patterns of peak discharges and overall flood risk estimates. It could be demonstrated that both methods are valid approaches for risk assessment with specific advantages and disadvantages. Both methods are superior to the traditional assumption of a uniform return period, where risk is computed by assuming a homogeneous return period (e.g. 100-year flood) across the entire study area.