Figure - uploaded by Emanuele Bevacqua
Content may be subject to copyright.
Source publication
Compound events are multivariate extreme events in which the individual contributing variables may not be extreme themselves, but their joint - dependent - occurrence causes an extreme impact. The conventional univariate statistical analysis cannot give accurate information regarding the multivariate nature of these events. We develop a conceptual...
Similar publications
Compound events (CEs) are multivariate extreme events in which the individual contributing variables may not be extreme themselves, but their joint – dependent – occurrence causes an extreme impact. Conventional univariate statistical analysis cannot give accurate information regarding the multivariate nature of these events. We develop a conceptua...
Citations
... The combination of multiple climate drivers contributing to amplify societal or environmental risk is referred to as a "compound event" (Seneviratne et al. 2012;Reisinger et al. 2014;Zscheischler et al. 2018). The individual climatic event may not have to reach extreme values, but their co-occurrence may often result in severe consequences (Bevacqua et al. 2017). ...
Under global warming, extreme events have been increasing in the last decade and are projected to increase in the future with every increment of global warming. The potential increase in compound drought and hot events may induce a complex web of impacts on societies, ecosystems, and economies, including crop failure, wildfires, and water scarcity. This is particularly concerning for Brazil, where it has been demonstrated to be vulnerable to recent extreme climate events. Using an ensemble of CORDEX-CORE simulations over Tropical Brazil, we investigate changes in compound events in response to changes in radiative forcing and their impact on climate extreme events, including drought and extreme heat. The simulations are conducted at a 25 km horizontal grid spacing using lateral and lower boundary forcing from three Coupled Model Intercomparison Project Phase 5 (CMIP5) climate models. Each model covers the period from 1980 to 2100 under two Radiative Concentration Pathways (RCP2.6 and RCP8.5) in the 21st-century projection period. We used observed data from the Brazilian Daily Weather Gridded Data (BR-DWGD) to evaluate the simulations and perform a quantitative assessment of areas affected by these compound events during the present day. The study finds a generally good agreement between RCM simulations and observed data, with moderate to high correlation coefficients for precipitation, though the strength of these correlations varies across different regions and seasons. The analysis emphasizes the prevalence of compound climate events during the Austral summer season and projects a significant increase in both extreme heat and drought events in the coming decades. These findings underscore Brazil’s vulnerability to compound climate events, highlighting the need for adaptive strategies and policy interventions to mitigate the socio-economic and environmental impacts across various sectors.
... Mediterranean cyclones, resulting from interactions between Atlantic low-24 pressure systems and the Mediterranean Sea, draw energy from the warm25 basin waters (Lionello et al., 2021). These cyclones intensify Acqua Alta through26 the convergence of winds and associated storm surge towards the northern27 Adriatic Sea coast, as well as through higher water levels due to the lowered28 sea level pressure (barometric pressure effect)(Bevacqua et al., 2017). The29 synergy between these cyclones and the Adriatic's geographic features30 magnifies their impact on Venice (Zanchettin et al.Acqua Alta events, Faranda et al. (2023) employed analogues of 33 atmospheric patterns (Faranda et al., 2022). ...
As extreme event attribution (EEA) matures, explaining the impacts of extreme events has risen to be a key focus for attribution scientists. Studies of this type usually assess the contribution of anthropogenic climate change to observed impacts. Other scientific communities have developed tools to assess how human activities influence impacts of extreme weather events on ecosystems and societies. For example, the disaster risk reduction (DRR) community analyses how the structure of human societies affects exposure, vulnerability, and ultimately the impacts of extreme weather events, with less attention to the role of anthropogenic climate change. In this perspective, we argue that adapting current practice in EEA to also consider other causal factors in attribution of extreme weather impacts would provide richer and more comprehensive insight into the causes of disasters. To this end, we propose a framework for EEA that would generate a more complete picture of human influences on impacts and bridge the gap between the EEA and DRR communities. We provide illustrations for five case studies: the 2021–2022 Kenyan drought; the 2013–2015 marine heatwave in the northeast Pacific; the 2017 forest fires in Portugal; Acqua Alta (flooding) events in Venice and evaluation of the efficiency of the Experimental Electromechanical Module, an ensemble of mobile barriers that can be activated to mitigate the influx of seawater in the city; and California droughts and the Forecast Informed Reservoir Operations system as an adaptation strategy.
... between floods and storm surges (Xu et al. 2023;Bevacqua et al. 2017;Gill and Malamud 2014;Ward and Day 2008). The use of joint probability analysis method to study the combined probability of various factors encountered at typical stations in the Feiyun River estuary is an effective approach (Olbert et al. 2023;Zheng et al. 2015;Hawkes 2008). ...
The tidal section of the Feiyun River is affected by the dual effects of the upstream flood discharge and downstream tidal uplift, resulting in complex water level changes. To determine the extreme water levels of the tidal section, the methods of the statistical analysis, harmonic calculation and joint probability analysis are utilized to study the correlation between the water level changes and the downstream estuarine tides and upstream floods based on the observation data and explore the extreme water level characteristics caused by flood-tide encounters. The study results indicate that the interannual water level correlation between Rui'an Hydrological Station and the up-downstream is not significant, but the correlation of their monthly water levels is enhanced. During the rainy season, there is a positive correlation between the daily average water levels of the upstream and Rui'an Hydrological Station. The water levels at Feiyun River (Rui'an section) are more affected by the combined influence of the astronomical tides and storm surges than by the upstream runoffs and water levels. The main contributing factor to the extreme skew surges at Rui'an Hydrological Station is the offshore storm surges. There is a significant time difference between the upstream flood peaks and the skew surge peaks at Rui'an Hydrological Station. The skew surge peaks at Rui'an Hydrological Station are at the forefront of the flood peaks superimposed by the extreme skew surges of the storm surges. The combined maximum total water levels for 1 year, 5 years, 10 years and 50 years calculated by the joint probability analysis method are 4.16 m, 4.44 m, 4.59 m and 4.90 m, respectively, which are basically consistent with the analysis results of measured data in the past 50 years. According to the joint probability distribution of the storm surges and astronomical tides, the reliability of short period calculation results is relatively high.
... Additionally, in engineering, threedimensional copulas are used to study the connections between three components in complex systems, thereby facilitating reliability analysis and the improvement of system design. Some of the best-known of three-dimensional copulas have been recently practically involved in [10], [11], [12], [13], and [14], among others. New theoretically oriented methodologies on this topic can be found in [15], [16], [17], [18], [19], [20], [21], and [22]. ...
... Any inference on this level would be highly uncertain since 231 it is based on a single event (e.g., Zscheischler & Fischer, 2020). For shorter time series, 232 the maximum empirical T also decreases such that extreme event estimation suffers from 233 high uncertainties (Bevacqua et al., 2017). Instead of event counting, we here fitted cop-234 ulas, i.e., multivariate probability distributions, to the bivariate distributions (Zscheischler 235 & Fischer, 2020). ...
Compound dry and hot extremes (CDHE, such as recent summers 2015, 2018 and 2022 in Europe) have wide ranging impacts: Heat exacerbates moisture shortages during dry periods whereas water demand rises. Climate change will likely increase the intensity, frequency, and duration of CDHE events in Europe. However, current studies focus on drivers and impacts in coarse-resolution global climate models and likely miss spatial details of CDHE characteristics. To overcome this issue, we exploit a regional 50-member single-model initial condition large ensemble (SMILE) at 12 km spatial resolution. Hence 1000 model years per 20 year-periods provide an extensive database of CDHE and robust estimations of their occurrence changes across Europe in high geographical detail. CDHE occurrences are investigated in a current climate and at two global warming levels (+2 °C, +3 °C). We identify Northern France, Southern Germany, Switzerland, Southern Ireland, and the western coasts of the Black Sea with currently low CDHE frequencies as emerging hotspots. These regions experience a tenfold occurrence increase under global warming conditions. Apart from Western Europe, temperature is the dominant contributor to frequency increases. Furthermore, tail dependencies strengthen in regions with high CDHE frequency increases. In European agricultural areas, soil moisture shows very strong negative correlations with CDHE extremeness. Last, our results suggest a halving of CDHE in a +2 °C world compared to a +3 °C world, highlighting the necessity of climate mitigation with respect to this hazard type.
... It has been widely used to construct multivariate joint distributions to analyze the combinations of events. Bevacqua et al. (2017) constructed the joint distribution of sea and river levels to analyze compound flood risk in Ravenna. Zhang et al. (2021) constructed the joint distribution of flood peak, volume, and duration to evaluate the flood risk of Guadalupe and Mission River basins. ...
The complex topology of river networks and the numerous factors influencing streamflow make it challenging to forecast streamflow in large river basins. Improving the accuracy of streamflow forecasting in these basins is of great importance for water resource planning and management. In this paper, considering the spatial variability of impacts of human activities on streamflow, a streamflow forecasting method with a hybrid physical process-mathematical statistic is proposed to obtain better results by realizing the complementary advantages of conventional methods and the streamflow into Hongze Lake in China is forecasted by the proposed method. Firstly, the physical process-based Soil and Water Assessment Tool (SWAT) is used to forecast the streamflow of tributaries with few water storage projects such as reservoirs in the catchment area. Secondly, considering the correlation between mainstream streamflow and tributaries streamflow as well as the correlation between mainstream streamflow and its previous streamflow, streamflow forecasting models based on the vine copula function are constructed to forecast the mainstream streamflow and its confidence intervals, where there are many water storage projects in the catchment area. Then, the total streamflow into Hongze Lake and its confidence intervals are obtained by coupling the outputs of the above two parts. Finally, statistical indicators are chosen to evaluate the forecasting effects in terms of the credibility of deterministic forecasting as well as the reliability and acuity of probabilistic forecasting. The results demonstrate that the proposed method outperforms existing SWAT and long and short-term memory (LSTM) neural network methods in terms of forecasting performance. Consequently, it presents an effective alternative for addressing complex hydrological forecasting tasks in large river basins.
... These driving forces co-occurred or in close succession, limiting the univariate frequency distribution approach's reliability or associated return periods. Incorporating a parametric-based multivariate probability distribution framework could permit the mutual concurrencies between the contributing variables of compound events (i.e., Serinaldi 2015;Aghakouchak et al. 2014;Bevacqua et al. 2017). Compared to the empirical estimation procedure, the statistical approach minimizes the uncertainties of statistical properties that we want to examine from the given observational datasets. ...
Compound flooding is a multidimensional consequence of the joint impact of multiple intercorrelated drivers, such as oceanographic, hydrologic, and meteorological. These individual drivers exhibit interdependence due to common forcing mechanisms. If they occur simultaneously or successively, the probability of their joint occurrence will be higher than expected if considered separately. The copula-based multivariate joint analysis can effectively measure hydrologic risk associated with compound events. Because of the involvement of multiple drivers, it is necessary to switch from bivariate (2D) to trivariate (3D) analyses. This study presents an original trivariate probabilistic framework by incorporating multivariate hierarchal models called asymmetric or fully nested Archimedean (or FNA) copula in the joint analysis of compound flood risk. The efficacy of the derived FNA copulas model, together with symmetric Archimedean and Elliptical class copulas, are tested by compounding the joint impact of rainfall, storm surge, and river discharge observations through a case study at the west coast of Canada. The obtained copula-based joint analysis is employed in multivariate analysis of flood risks in trivariate and bivariate primary joint and conditional joint return periods. The estimated joint return periods are further employed in estimating failure probability statistics for assessing the trivariate (and bivariate) hydrologic risk associated with compound events. The statistical tests found the fully nested Frank copula outperforms symmetric 3D copulas. Our work confirms that for practical compound flood risk analysis together with bivariate or univariate return periods, it is essential to account for the trivariate joint return periods to assess the expected compound flood risk and strength of influence of different variables if they occur simultaneously or successively. The bivariate (also univariate) events produce a lower failure probability than trivariate analysis for the OR-joint cases. Thus, ignoring the compounding impacts via trivariate joint analysis can significantly underestimate failure probability and joint return period.
... During this combination, either the river flow becomes blocked or a back wave is formed; in both cases, in the lower reaches, water level rises and increases the risk of a flood (Khanam et al. 2021, Hsiao et al. 2021. Individual components can be non-extreme, but their general interdependency can cause extreme situations (Emanuele Bevacqua et al. 2017). In order to determine the anthropogenic effects on different characteristics on compound floods these flood types require a systematic approach (Zscheischler et al. 2020). ...
The moisture content in the atmosphere column is a major
characteristic that determines the potential of precipitation
formation. In the atmosphere, horizontal air flows
dominates, therefore, a water vapor can be redistributed
inhomogeneously due to its horizontal advection. For
precipitation, physical processes, which lead to lifting air
mass and moisture condensation, are needed. If there is a
large amount of water vapor in a significant layer of the
troposphere, heavy precipitation can occurred. These cases
include the so‐called 'atmospheric rivers'.
Atmospheric rivers are often associated with extreme
winter storms and heavy rainfalls along the west coast of
continents, including the US and Western Europe. In some
cases, atmospheric rivers penetrates inland and causing
intensive precipitation and flash floods (Lavers and Villarini.
2015). The term ‘atmospheric river’ was introduced in 1998
(Zhu and Newell, 1998). Further, atmospheric rivers have
been defined as anomalous moisture content and strong
horizontal moisture transport, which are concentrated into
long narrow corridors, typically 400–600 km wide (but ≤
1000 km) and more than 2000 km long (Gimeno et al., 2014).
Vertically integrated horizontal water vapor transport (IVT)
is used as a physical threshold of the atmospheric river, in
which IVT ≥ 250 kg m−1 s−1.
Atmospheric rivers are associated with the processes of
cyclogenesis, where the horizontal flow of water vapor get
ascending movement and turn into precipitation. In the
basins of the Mediterranean and Black Sea, intense
mesoscale cyclones are sometimes observed, with
characteristics and development mechanism similar to
tropical cyclones. These cyclones have been named as
“medicanes“ (Mediterranean Hurricanes) or Tropical‐Like
Cyclones (TLC) (Miglietta, 2019). Medicanes are
characterized by the development of intensive convection
and convective phenomena, such as strong winds, heavy
precipitation and flooding on coastal areas. The main
process of medicanes intensification is the release of latent
heat of condensation associated with convection processes
and interaction of sea surface with the atmosphere. But in
the early stages of development, the main mechanism is a
baroclinic instability of the atmosphere,since these cyclones
are developing under the deep upper‐level cutoff low, in
which contains a large amount of moisture (Emanuel, 2005).
The purpose of this study is to analyze the integrated
water vapor transport in a tropical‐like cyclone, which
formed over the Black Sea in August 2021, and led to strong
rainfalls with flooding in the northern regions of Turkey and
in the Krasnodar Krai of the Russian Federation.
... It is interesting to note that recent research has focused on the analysis of Compound Events (CE), i.e., multivariate occurrences in which the contributing variables may not be extreme themselves, but their joint instances may nevertheless cause severe impacts: traditional univariate statistical analyses cannot give information regarding the multivariate nature of CE's and the hazard/risks associated with them (Schölzel and Friederichs 2008;Bevacqua et al. 2017). Attention to CE's has also increased at an international policy-makers level. ...
The particular structure and configuration of the Venice lagoon represents a paramount case study concerning coastal flooding which affects natural, historical/cultural properties, together with industrial, commercial, economical and port activities. In order to defend Venice (and other sites) within the lagoon from severe floods, the Italian Government promoted the construction of a complex hydraulic/maritime system, including a movable storm surge barrier named Experimental Electromechanical Module (MoSE), to be activated when specific water levels occur. When the MoSE barriers are raised, the only access to the lagoon for commercial and cruise ships is represented by the Malamocco lock gate, provided that suitable safety conditions (involving the significant wave height) are satisfied. In addition, the Italian Government has recently established that, in the near future, large ships will always have to enter/exit the lagoon only through the Malamocco entrance. In turn, the navigation within the Venice lagoon is (will be) controlled by the combined MoSE-Malamocco system, ruled by both univariate and bivariate paradigms/guidelines. As a novelty, in the present work, for the first time, the statistics of significant wave heights and water levels in the Venice lagoon (both univariate and bivariate ones) are investigated: in particular, these variables turn out to be dependent, and their joint occurrence (statistically modeled via Copulas) can determine the stop of ship navigation, yielding significant economic losses. Here, univariate and bivariate Return Periods and Failure Probabilities are used to thoroughly model the statistical behavior of significant wave heights and water levels, in order to provide useful quantitative indications for the management of the tricky hydraulic, maritime and economical system of the Venice lagoon.
... Studying compound events is a complex task, which often requires a multidisciplinary 99 approach involving the understanding of the underlying physical processes beyond the 100 impact, climate, and weather elements, and advanced process-based and statistical mod-101 elling (Bevacqua et al., 2017). Consequently, for users, it is often difficult to identify the 102 key elements (i.e., physical variables, relevant spatial and temporal scales as well as suit-103 able analysis tools and datasets) necessary to address a given compound event-related 104 research question. ...
Compound weather and climate events are combinations of climate drivers and/or hazards that contribute to societal or environmental risk. Studying compound events often requires a multidisciplinary approach combining domain knowledge of the underlying processes with, for example, statistical methods and climate model outputs. Recently, to aid the development of research on compound events, four compound event types were introduced, namely (1) preconditioned, (2) multivariate, (3) temporally compounding, and (4) spatially compounding events. However, guidelines on how to study these types of events are still lacking. Here, we consider four case studies, each associated with a specific event type and a research question, to illustrate how the key elements of compound events (e.g., analytical tools and relevant physical effects) can be identified. These case studies show that (1) impacts on crops from hot and dry summers can be exacerbated by preconditioning effects of dry and bright springs. (2) Assessing compound coastal flooding in Perth (Australia) requires considering the dynamics of a non-stationary multivariate process. For instance, future mean sea-level rise will lead to the emergence of concurrent coastal and fluvial extremes, enhancing compound flooding risk. (3) In Portugal, deep-landslides are often caused by temporal clusters of moderate precipitation events. Finally, (4) crop yield failures in France and Germany are strongly correlated, threatening European food security through spatially compounding effects. These analyses allow for identifying general recommendations for studying compound events. Overall, our insights can serve as a blueprint for compound event analysis across disciplines and sectors.
