TY - JOUR A1 - Vogel, Kristin A1 - Weise, Laura A1 - Schröter, Kai A1 - Thieken, Annegret T1 - Identifying Driving Factors in Flood-Damaging Processes Using Graphical Models JF - Water resources research N2 - Flood damage estimation is a core task in flood risk assessments and requires reliable flood loss models. Identifying the driving factors of flood loss at residential buildings and gaining insight into their relations is important to improve our understanding of flood damage processes. For that purpose, we learn probabilistic graphical models, which capture and illustrate (in-)dependencies between the considered variables. The models are learned based on postevent surveys with flood-affected residents after six flood events, which occurred in Germany between 2002 and 2013. Besides the sustained building damage, the survey data contain information about flooding parameters, early warning and emergency measures, property-level mitigation measures and preparedness, socioeconomic characteristics of the household, and building characteristics. The analysis considers the entire data set with a total of 4,468 cases as well as subsets of the data set partitioned into single flood events and flood types: river floods, levee breaches, surface water flooding, and groundwater floods, to reveal differences in the damaging processes. The learned networks suggest that the flood loss ratio of residential buildings is directly influenced by hydrological and hydraulic aspects as well as by building characteristics and property-level mitigation measures. The study demonstrates also that for different flood events and process types the building damage is influenced by varying factors. This suggests that flood damage models need to be capable of reproducing these differences for spatial and temporal model transfers. KW - flood loss KW - Bayesian Network KW - Markov Blanket KW - vulnerability KW - Germany Y1 - 2018 U6 - https://doi.org/10.1029/2018WR022858 SN - 0043-1397 SN - 1944-7973 VL - 54 IS - 11 SP - 8864 EP - 8889 PB - American Geophysical Union CY - Washington ER - TY - JOUR A1 - Vogel, Kristin A1 - Riggelsen, Carsten A1 - Korup, Oliver A1 - Scherbaum, Frank T1 - Bayesian network learning for natural hazard analyses JF - Natural hazards and earth system sciences N2 - Modern natural hazards research requires dealing with several uncertainties that arise from limited process knowledge, measurement errors, censored and incomplete observations, and the intrinsic randomness of the governing processes. Nevertheless, deterministic analyses are still widely used in quantitative hazard assessments despite the pitfall of misestimating the hazard and any ensuing risks. In this paper we show that Bayesian networks offer a flexible framework for capturing and expressing a broad range of uncertainties encountered in natural hazard assessments. Although Bayesian networks are well studied in theory, their application to real-world data is far from straightforward, and requires specific tailoring and adaptation of existing algorithms. We offer suggestions as how to tackle frequently arising problems in this context and mainly concentrate on the handling of continuous variables, incomplete data sets, and the interaction of both. By way of three case studies from earthquake, flood, and landslide research, we demonstrate the method of data-driven Bayesian network learning, and showcase the flexibility, applicability, and benefits of this approach. Our results offer fresh and partly counterintuitive insights into well-studied multivariate problems of earthquake-induced ground motion prediction, accurate flood damage quantification, and spatially explicit landslide prediction at the regional scale. In particular, we highlight how Bayesian networks help to express information flow and independence assumptions between candidate predictors. Such knowledge is pivotal in providing scientists and decision makers with well-informed strategies for selecting adequate predictor variables for quantitative natural hazard assessments. Y1 - 2014 U6 - https://doi.org/10.5194/nhess-14-2605-2014 SN - 1561-8633 VL - 14 IS - 9 SP - 2605 EP - 2626 PB - Copernicus CY - Göttingen ER - TY - JOUR A1 - Vogel, Kristin A1 - Ozturk, Ugur A1 - Riemer, Adrian A1 - Laudan, Jonas A1 - Sieg, Tobias A1 - Wendi, Dadiyorto A1 - Agarwal, Ankit A1 - Roezer, Viktor A1 - Korup, Oliver A1 - Thieken, Annegret T1 - Die Sturzflut von Braunsbach am 29. Mai 2016 – Entstehung, Ablauf und Schäden eines „Jahrhundertereignisses“ T1 - The Braunsbach Flashflood of Mai 29th, 2016-Origin, Pathways and Impacts of an Extreme Hydro-Meteorological Event BT - Teil 2: Geomorphologische Prozesse und Schadensanalyse BT - Part 2: Geomorphological Processes and Damage Analysis JF - Hydrologie und Wasserbewirtschaftung N2 - Am Abend des 29. Mai 2016 wurde der Ort Braunsbach im Landkreis Schwäbisch-Hall (Baden-Württemberg) von einer Sturzflut getroffen, bei der mehrere Häuser stark beschädigt oder zerstört wurden. Die Sturzflut war eine der Unwetterfolgen, die im Frühsommer 2016 vom Tiefdruckgebiet Elvira ausgelöst wurden. Der vorliegende Bericht ist der zweite Teil einer Doppelveröffentlichung, welche die Ergebnisse zur Untersuchung des Sturzflutereignisses im Rahmen des DFG-Graduiertenkollegs “Naturgefahren und Risiken in einer sich verändernden Welt” (NatRiskChange, GRK 2043/1) der Universität Potsdam präsentiert. Während Teil 1 die meteorologischen und hydrologischen Ereignisse analysiert, fokussiert Teil 2 auf die geomorphologischen Prozesse und die verursachten Gebäudeschäden. Dazu wurden Ursprung und Ausmaß des während des Sturzflutereignisses mobilisierten und in den Ort getragenen Materials untersucht. Des Weiteren wurden zu 96 betroffenen Gebäuden Daten zum Schadensgrad sowie Prozess- und Gebäudecharakteristika aufgenommen und ausgewertet. Die Untersuchungen zeigen, dass bei der Betrachtung von Hochwassergefährdung die Berücksichtigung von Sturzfluten und ihrer speziellen Charakteristika, wie hoher Feststofftransport und sprunghaftes Verhalten insbesondere in bebautem Gelände, wesentlich ist, um effektive Schutzmaßnahmen ergreifen zu können. N2 - A severe flash flood event hit the town of Braunsbach (Baden-Wurttemberg, Germany) on the evening of May 29, 2016, heavily damaging and destroying several dozens of buildings. It was only one of several disastrous events in Central Europe caused by the low-pressure system "Elvira". The DFG Graduate School "Natural hazards and risks in a changing world" (NatRiskChange, GRK 2043/1) at the University of Potsdam investigated the Braunsbach flash flood as a recent showcase for catastrophic events triggered by severe weather. This contribution is part two of a back-to-back publication on the results of this storm event. While part 1 analyses the meteorological and hydrological situation, part 2 concentrates on the geomorphological aspects and damage to buildings. The study outlines the origin and amount of material that was mobilized and transported into the town by the flood, and analyses damage data collected for 96 affected buildings, describing the degree of impact, underlying processes, and building characteristics. Due to the potentially high sediment load of flash floods and their non-steady and non-uniform flow especially in built-up areas, the damaging processes differ from those of clear water floods. The results underline the need to consider flash floods and their specific behaviour in flood hazard assessments. KW - flash flood KW - flood risk KW - damaging processes KW - debris flow KW - erosion KW - landslides KW - Braunsbach KW - Sturzflut KW - Hochwassergefährdung KW - Schadensprozesse KW - Erosion KW - Hangrutschungen Y1 - 2017 U6 - https://doi.org/10.5675/HyWa_2017,3_2 SN - 1439-1783 VL - 61 IS - 3 SP - 163 EP - 175 PB - Bundesanst. für Gewässerkunde CY - Koblenz ER - TY - THES A1 - Vogel, Kristin T1 - Applications of Bayesian networks in natural hazard assessments T1 - Anwendungen von Bayes'schen Netzen bei der Einschätzung von Naturgefahren N2 - Even though quite different in occurrence and consequences, from a modeling perspective many natural hazards share similar properties and challenges. Their complex nature as well as lacking knowledge about their driving forces and potential effects make their analysis demanding: uncertainty about the modeling framework, inaccurate or incomplete event observations and the intrinsic randomness of the natural phenomenon add up to different interacting layers of uncertainty, which require a careful handling. Nevertheless deterministic approaches are still widely used in natural hazard assessments, holding the risk of underestimating the hazard with disastrous effects. The all-round probabilistic framework of Bayesian networks constitutes an attractive alternative. In contrast to deterministic proceedings, it treats response variables as well as explanatory variables as random variables making no difference between input and output variables. Using a graphical representation Bayesian networks encode the dependency relations between the variables in a directed acyclic graph: variables are represented as nodes and (in-)dependencies between variables as (missing) edges between the nodes. The joint distribution of all variables can thus be described by decomposing it, according to the depicted independences, into a product of local conditional probability distributions, which are defined by the parameters of the Bayesian network. In the framework of this thesis the Bayesian network approach is applied to different natural hazard domains (i.e. seismic hazard, flood damage and landslide assessments). Learning the network structure and parameters from data, Bayesian networks reveal relevant dependency relations between the included variables and help to gain knowledge about the underlying processes. The problem of Bayesian network learning is cast in a Bayesian framework, considering the network structure and parameters as random variables itself and searching for the most likely combination of both, which corresponds to the maximum a posteriori (MAP score) of their joint distribution given the observed data. Although well studied in theory the learning of Bayesian networks based on real-world data is usually not straight forward and requires an adoption of existing algorithms. Typically arising problems are the handling of continuous variables, incomplete observations and the interaction of both. Working with continuous distributions requires assumptions about the allowed families of distributions. To "let the data speak" and avoid wrong assumptions, continuous variables are instead discretized here, thus allowing for a completely data-driven and distribution-free learning. An extension of the MAP score, considering the discretization as random variable as well, is developed for an automatic multivariate discretization, that takes interactions between the variables into account. The discretization process is nested into the network learning and requires several iterations. Having to face incomplete observations on top, this may pose a computational burden. Iterative proceedings for missing value estimation become quickly infeasible. A more efficient albeit approximate method is used instead, estimating the missing values based only on the observations of variables directly interacting with the missing variable. Moreover natural hazard assessments often have a primary interest in a certain target variable. The discretization learned for this variable does not always have the required resolution for a good prediction performance. Finer resolutions for (conditional) continuous distributions are achieved with continuous approximations subsequent to the Bayesian network learning, using kernel density estimations or mixtures of truncated exponential functions. All our proceedings are completely data-driven. We thus avoid assumptions that require expert knowledge and instead provide domain independent solutions, that are applicable not only in other natural hazard assessments, but in a variety of domains struggling with uncertainties. N2 - Obwohl Naturgefahren in ihren Ursachen, Erscheinungen und Auswirkungen grundlegend verschieden sind, teilen sie doch viele Gemeinsamkeiten und Herausforderungen, wenn es um ihre Modellierung geht. Fehlendes Wissen über die zugrunde liegenden Kräfte und deren komplexes Zusammenwirken erschweren die Wahl einer geeigneten Modellstruktur. Hinzu kommen ungenaue und unvollständige Beobachtungsdaten sowie dem Naturereignis innewohnende Zufallsprozesse. All diese verschiedenen, miteinander interagierende Aspekte von Unsicherheit erfordern eine sorgfältige Betrachtung, um fehlerhafte und verharmlosende Einschätzungen von Naturgefahren zu vermeiden. Dennoch sind deterministische Vorgehensweisen in Gefährdungsanalysen weit verbreitet. Bayessche Netze betrachten die Probleme aus wahrscheinlichkeitstheoretischer Sicht und bieten somit eine sinnvolle Alternative zu deterministischen Verfahren. Alle vom Zufall beeinflussten Größen werden hierbei als Zufallsvariablen angesehen. Die gemeinsame Wahrscheinlichkeitsverteilung aller Variablen beschreibt das Zusammenwirken der verschiedenen Einflussgrößen und die zugehörige Unsicherheit/Zufälligkeit. Die Abhängigkeitsstrukturen der Variablen können durch eine grafische Darstellung abgebildet werden. Die Variablen werden dabei als Knoten in einem Graphen/Netzwerk dargestellt und die (Un-)Abhängigkeiten zwischen den Variablen als (fehlende) Verbindungen zwischen diesen Knoten. Die dargestellten Unabhängigkeiten veranschaulichen, wie sich die gemeinsame Wahrscheinlichkeitsverteilung in ein Produkt lokaler, bedingter Wahrscheinlichkeitsverteilungen zerlegen lässt. Im Verlauf dieser Arbeit werden verschiedene Naturgefahren (Erdbeben, Hochwasser und Bergstürze) betrachtet und mit Bayesschen Netzen modelliert. Dazu wird jeweils nach der Netzwerkstruktur gesucht, welche die Abhängigkeiten der Variablen am besten beschreibt. Außerdem werden die Parameter der lokalen, bedingten Wahrscheinlichkeitsverteilungen geschätzt, um das Bayessche Netz und dessen zugehörige gemeinsame Wahrscheinlichkeitsverteilung vollständig zu bestimmen. Die Definition des Bayesschen Netzes kann auf Grundlage von Expertenwissen erfolgen oder - so wie in dieser Arbeit - anhand von Beobachtungsdaten des zu untersuchenden Naturereignisses. Die hier verwendeten Methoden wählen Netzwerkstruktur und Parameter so, dass die daraus resultierende Wahrscheinlichkeitsverteilung den beobachteten Daten eine möglichst große Wahrscheinlichkeit zuspricht. Da dieses Vorgehen keine Expertenwissen voraussetzt, ist es universell in verschiedenen Gebieten der Gefährdungsanalyse einsetzbar. Trotz umfangreicher Forschung zu diesem Thema ist das Bestimmen von Bayesschen Netzen basierend auf Beobachtungsdaten nicht ohne Schwierigkeiten. Typische Herausforderungen stellen die Handhabung stetiger Variablen und unvollständiger Datensätze dar. Beide Probleme werden in dieser Arbeit behandelt. Es werden Lösungsansätze entwickelt und in den Anwendungsbeispielen eingesetzt. Eine Kernfrage ist hierbei die Komplexität des Algorithmus. Besonders wenn sowohl stetige Variablen als auch unvollständige Datensätze in Kombination auftreten, sind effizient arbeitende Verfahren gefragt. Die hierzu in dieser Arbeit entwickelten Methoden ermöglichen die Verarbeitung von großen Datensätze mit stetigen Variablen und unvollständigen Beobachtungen und leisten damit einen wichtigen Beitrag für die wahrscheinlichkeitstheoretische Gefährdungsanalyse. KW - Bayes'sche Netze KW - Naturgefahren KW - Gefahrenanalyse KW - Unsicherheiten KW - Hochwasser KW - Bayesian networks KW - natural hazards KW - hazard assessments KW - uncertainties KW - flood events Y1 - 2013 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:517-opus-69777 ER - TY - JOUR A1 - Sieg, Tobias A1 - Vogel, Kristin A1 - Merz, Bruno A1 - Kreibich, Heidi T1 - Tree-based flood damage modeling of companies: Damage processes and model performance JF - Water resources research N2 - Reliable flood risk analyses, including the estimation of damage, are an important prerequisite for efficient risk management. However, not much is known about flood damage processes affecting companies. Thus, we conduct a flood damage assessment of companies in Germany with regard to two aspects. First, we identify relevant damage-influencing variables. Second, we assess the prediction performance of the developed damage models with respect to the gain by using an increasing amount of training data and a sector-specific evaluation of the data. Random forests are trained with data from two postevent surveys after flood events occurring in the years 2002 and 2013. For a sector-specific consideration, the data set is split into four subsets corresponding to the manufacturing, commercial, financial, and service sectors. Further, separate models are derived for three different company assets: buildings, equipment, and goods and stock. Calculated variable importance values reveal different variable sets relevant for the damage estimation, indicating significant differences in the damage process for various company sectors and assets. With an increasing number of data used to build the models, prediction errors decrease. Yet the effect is rather small and seems to saturate for a data set size of several hundred observations. In contrast, the prediction improvement achieved by a sector-specific consideration is more distinct, especially for damage to equipment and goods and stock. Consequently, sector-specific data acquisition and a consideration of sector-specific company characteristics in future flood damage assessments is expected to improve the model performance more than a mere increase in data. Y1 - 2017 U6 - https://doi.org/10.1002/2017WR020784 SN - 0043-1397 SN - 1944-7973 VL - 53 SP - 6050 EP - 6068 PB - American Geophysical Union CY - Washington ER - TY - GEN A1 - Sieg, Tobias A1 - Vogel, Kristin A1 - Merz, Bruno A1 - Kreibich, Heidi T1 - Seamless Estimation of Hydrometeorological Risk Across Spatial Scales T2 - Postprints der Universität Potsdam Mathematisch-Naturwissenschaftliche Reihe N2 - Hydrometeorological hazards caused losses of approximately 110 billion U.S. Dollars in 2016 worldwide. Current damage estimations do not consider the uncertainties in a comprehensive way, and they are not consistent between spatial scales. Aggregated land use data are used at larger spatial scales, although detailed exposure data at the object level, such as openstreetmap.org, is becoming increasingly available across the globe.We present a probabilistic approach for object-based damage estimation which represents uncertainties and is fully scalable in space. The approach is applied and validated to company damage from the flood of 2013 in Germany. Damage estimates are more accurate compared to damage models using land use data, and the estimation works reliably at all spatial scales. Therefore, it can as well be used for pre-event analysis and risk assessments. This method takes hydrometeorological damage estimation and risk assessments to the next level, making damage estimates and their uncertainties fully scalable in space, from object to country level, and enabling the exploitation of new exposure data. T3 - Zweitveröffentlichungen der Universität Potsdam : Mathematisch-Naturwissenschaftliche Reihe - 743 KW - spatial scales KW - risk assessment KW - hydro-meterological hazards KW - object-based damage modeling KW - uncertainty KW - probabilistic approaches Y1 - 2019 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:517-opus4-435341 SN - 1866-8372 IS - 743 SP - 574 EP - 581 ER - TY - JOUR A1 - Sieg, Tobias A1 - Vogel, Kristin A1 - Merz, Bruno A1 - Kreibich, Heidi T1 - Seamless Estimation of Hydrometeorological Risk Across Spatial Scales JF - Earth's Future N2 - Hydrometeorological hazards caused losses of approximately 110 billion U.S. Dollars in 2016 worldwide. Current damage estimations do not consider the uncertainties in a comprehensive way, and they are not consistent between spatial scales. Aggregated land use data are used at larger spatial scales, although detailed exposure data at the object level, such as openstreetmap.org, is becoming increasingly available across the globe.We present a probabilistic approach for object-based damage estimation which represents uncertainties and is fully scalable in space. The approach is applied and validated to company damage from the flood of 2013 in Germany. Damage estimates are more accurate compared to damage models using land use data, and the estimation works reliably at all spatial scales. Therefore, it can as well be used for pre-event analysis and risk assessments. This method takes hydrometeorological damage estimation and risk assessments to the next level, making damage estimates and their uncertainties fully scalable in space, from object to country level, and enabling the exploitation of new exposure data. KW - spatial scales KW - risk assessment KW - hydro-meterological hazards KW - object-based damage modeling KW - uncertainty KW - probabilistic approaches Y1 - 2019 U6 - https://doi.org/10.1029/2018EF001122 SN - 2328-4277 VL - 7 IS - 5 SP - 574 EP - 581 PB - Wiley-Blackwell CY - Hoboken, NJ ER - TY - GEN A1 - Sieg, Tobias A1 - Shinko, Thomas A1 - Vogel, Kristin A1 - Mechler, Reinhard A1 - Merz, Bruno A1 - Kreibich, Heidi T1 - Integrated assessment of short-term direct and indirect economic flood impacts including uncertainty quantification T2 - Postprints der Universität Potsdam : Mathematisch-Naturwissenschaftliche Reihe N2 - Understanding and quantifying total economic impacts of flood events is essential for flood risk management and adaptation planning. Yet, detailed estimations of joint direct and indirect flood-induced economic impacts are rare. In this study an innovative modeling procedure for the joint assessment of short-term direct and indirect economic flood impacts is introduced. The procedure is applied to 19 economic sectors in eight federal states of Germany after the flood events in 2013. The assessment of the direct economic impacts is object-based and considers uncertainties associated with the hazard, the exposed objects and their vulnerability. The direct economic impacts are then coupled to a supply-side Input-Output-Model to estimate the indirect economic impacts. The procedure provides distributions of direct and indirect economic impacts which capture the associated uncertainties. The distributions of the direct economic impacts in the federal states are plausible when compared to reported values. The ratio between indirect and direct economic impacts shows that the sectors Manufacturing, Financial and Insurance activities suffered the most from indirect economic impacts. These ratios also indicate that indirect economic impacts can be almost as high as direct economic impacts. They differ strongly between the economic sectors indicating that the application of a single factor as a proxy for the indirect impacts of all economic sectors is not appropriate. T3 - Zweitveröffentlichungen der Universität Potsdam : Mathematisch-Naturwissenschaftliche Reihe - 708 KW - June 2013 KW - Damage KW - Model KW - Inoperability KW - Disasters KW - Hazards KW - Germany KW - Losses KW - Event KW - Costs Y1 - 2019 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:517-opus4-429119 IS - 708 ER - TY - JOUR A1 - Sieg, Tobias A1 - Schinko, Thomas A1 - Vogel, Kristin A1 - Mechler, Reinhard A1 - Merz, Bruno A1 - Kreibich, Heidi T1 - Integrated assessment of short-term direct and indirect economic flood impacts including uncertainty quantification JF - PLoS ONE N2 - Understanding and quantifying total economic impacts of flood events is essential for flood risk management and adaptation planning. Yet, detailed estimations of joint direct and indirect flood-induced economic impacts are rare. In this study an innovative modeling procedure for the joint assessment of short-term direct and indirect economic flood impacts is introduced. The procedure is applied to 19 economic sectors in eight federal states of Germany after the flood events in 2013. The assessment of the direct economic impacts is object-based and considers uncertainties associated with the hazard, the exposed objects and their vulnerability. The direct economic impacts are then coupled to a supply-side Input-Output-Model to estimate the indirect economic impacts. The procedure provides distributions of direct and indirect economic impacts which capture the associated uncertainties. The distributions of the direct economic impacts in the federal states are plausible when compared to reported values. The ratio between indirect and direct economic impacts shows that the sectors Manufacturing, Financial and Insurance activities suffered the most from indirect economic impacts. These ratios also indicate that indirect economic impacts can be almost as high as direct economic impacts. They differ strongly between the economic sectors indicating that the application of a single factor as a proxy for the indirect impacts of all economic sectors is not appropriate. KW - June 2013 KW - Damage KW - Model KW - Inoperability KW - Disasters KW - Hazards KW - Germany KW - Losses KW - Event KW - Costs Y1 - 2019 U6 - https://doi.org/10.1371/journal.pone.0212932 SN - 1932-6203 VL - 14 IS - 4 PB - Public Library of Science CY - San Francisco ER - TY - JOUR A1 - Schroeter, Kai A1 - Kreibich, Heidi A1 - Vogel, Kristin A1 - Riggelsen, Carsten A1 - Scherbaum, Frank A1 - Merz, Bruno T1 - How useful are complex flood damage models? JF - Water resources research N2 - We investigate the usefulness of complex flood damage models for predicting relative damage to residential buildings in a spatial and temporal transfer context. We apply eight different flood damage models to predict relative building damage for five historic flood events in two different regions of Germany. Model complexity is measured in terms of the number of explanatory variables which varies from 1 variable up to 10 variables which are singled out from 28 candidate variables. Model validation is based on empirical damage data, whereas observation uncertainty is taken into consideration. The comparison of model predictive performance shows that additional explanatory variables besides the water depth improve the predictive capability in a spatial and temporal transfer context, i.e., when the models are transferred to different regions and different flood events. Concerning the trade-off between predictive capability and reliability the model structure seem more important than the number of explanatory variables. Among the models considered, the reliability of Bayesian network-based predictions in space-time transfer is larger than for the remaining models, and the uncertainties associated with damage predictions are reflected more completely. KW - floods KW - damage KW - model validation KW - Bayesian networks KW - regression tree Y1 - 2014 U6 - https://doi.org/10.1002/2013WR014396 SN - 0043-1397 SN - 1944-7973 VL - 50 IS - 4 SP - 3378 EP - 3395 PB - American Geophysical Union CY - Washington ER -