Multi-variable flood damage modelling with limited data using supervised learning approaches

Wagenaar, Dennis; Jong, Jurjen de; Bouwer, Laurens M.

doi:10.5194/nhess-17-1683-2017

Cited by 86 publications

(106 citation statements)

References 25 publications

(31 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Publicly available data sets, such as from openstreetmap.org, can provide specific information such as the occupancy or height of buildings at the object level (Figueiredo & Martina, 2016). This kind of information could be used to assess damage caused to individual buildings by means of, for example, engineering-based hurricane damage models (Pita et al, 2013;Vickery et al, 2006) or multivariable flood damage models (Merz et al, 2013;Sieg et al, 2017;Wagenaar et al, 2017). So far, state-of-the-art damage models (Prahl et al, 2016;Sieg et al, 2017;Wagenaar et al, 2017) are not able to keep up with this development causing a gap between what is currently done and what improvements would theoretically be feasible.…”

Section: 1029/2018ef001122mentioning

confidence: 99%

“…Considering only one object, it can be seen that the point estimate, that is, the mean, is a rather unlikely value within the skewed damage distribution (Figure 2, k = 1, and Figure S1). In most cases, the point estimate for single objects will overestimate or underestimate the flood damage, which results in large errors as observed in Seifert et al (2010), Sieg et al (2017), and Wagenaar et al (2017). The higher the number of objects, by summing up their damage estimates, the better the empirical distribution fits to a normal distribution (Central Limit Theorem).…”

Section: The Role Of Vulnerability Uncertaintymentioning

confidence: 99%

“…Despite the progress in damage estimation in recent years (Figueiredo et al, 2018;Kreibich et al, 2016;Schröter et al, 2014;Sieg et al, 2017;Wagenaar et al, 2017), the available damage models are not able to describe the damage process reliably, and the process is stochastic to a substantial extent. Therefore, as implemented in this approach, the distribution of the damage estimate should be assessed at every spatial scale in order to report the lack of knowledge about the damage estimates.…”

Section: The Role Of Vulnerability Uncertaintymentioning

confidence: 99%

See 2 more Smart Citations

Seamless Estimation of Hydrometeorological Risk Across Spatial Scales

et al. 2019

View full text Add to dashboard Cite

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.

show abstract

Section: 1029/2018ef001122mentioning

confidence: 99%

Section: The Role Of Vulnerability Uncertaintymentioning

confidence: 99%

Section: The Role Of Vulnerability Uncertaintymentioning

confidence: 99%

See 1 more Smart Citation

Seamless Estimation of Hydrometeorological Risk Across Spatial Scales

et al. 2019

View full text Add to dashboard Cite

show abstract

“…The conventional approaches for modeling flood risk are mostly dependent on the flood water depth to estimate the associated damage (Aerts et al 2014, Velasco et al 2014. Several recent studies have shown that considering multivariate data will improve the damage estimates (Wagenaar et al 2017). Therefore, over the past few years, several studies evaluated flood risk in various regions of the globe (de Moel et al 2015, Arnell andGosling 2016, van Berchum et al 2018) using a multitude of variables representing hazard, vulnerability, and exposure.…”

Section: Introductionmentioning

confidence: 99%

Leveraging machine learning for predicting flash flood damage in the Southeast US

Alipour

Ahmadalipour

Abbaszadeh

et al. 2020

Environ. Res. Lett.

View full text Add to dashboard Cite

Flash flood is a recurrent natural hazard with substantial impacts in the Southeast US (SEUS) due to the frequent torrential rainfalls that occur in the region, which are triggered by tropical storms, thunderstorms, and hurricanes. Flash floods are costly natural hazards, primarily due to their rapid onset. Therefore, predicting property damage of flash floods is imperative for proactive disaster management. Here, we present a systematic framework that considers a variety of features explaining different components of risk (i.e. hazard, vulnerability, and exposure), and examine multiple machine learning methods to predict flash flood damage. A large database of flash flood events consisting of more than 14 000 events are assessed for training and testing the methodology, while a multitude of data sources are utilized to acquire reliable information related to each event. A variable selection approach was employed to alleviate the complexity of the dataset and facilitate the model development process. The random forest (RF) method was then used to map the identified input covariates to a target variable (i.e. property damage). The RF model was implemented in two modes: first, as a binary classifier to estimate if a region of interest was damaged in any particular flood event, and then as a regression model to predict the amount of property damage associated with each event. The results indicate that the proposed approach is successful not only for classifying damaging events (with an accuracy of 81%), but also for predicting flash flood damage with a good agreement with the observed property damage. This study is among the few efforts for predicting flash flood damage across a large domain using mesoscale input variables, and the findings demonstrate the effectiveness of the proposed methodology.

show abstract

“…Vulnerability is often represented in terms of stage-damage functions, also called damage functions in the literature, which are associated with a flood parameter, commonly water depth, a relative amount of loss or an absolute monetary loss. Vulnerability functions are often developed based on expert judgment [4][5][6], and sometimes they are derived from observed losses [7][8][9]. Recently also some physical laboratory models have been developed to better investigate the key damage mechanisms [10,11].…”

Section: Introductionmentioning

confidence: 99%

Empirical flash flood vulnerability functions for residential buildings

Arrighi

Mazzanti²,

Pistone³

et al. 2020

SN Appl. Sci.

View full text Add to dashboard Cite

The estimation of flood losses is crucial for the quantitative estimation of risk and the cost-benefit analysis of risk mitigation and prevention measures. Flood losses are usually estimated by means of vulnerability functions, which are often developed in a synthetic way by experts, since loss datasets collected after events are sparse or lacking and mostly refer to slow riverine inundations. Simple univariable empirical vulnerability functions are recognized as a more valuable tool with respect to synthetic curves developed for different contexts. The aim of this work is to derive empirical vulnerability functions for a flash flood event which occurred in Livorno (Italy) in September 2017, based on the hydrologic-hydraulic reconstruction of the flood and observed losses to residential properties. The hydrologic processes are reproduced with a distributed hydrologic model, and the flood propagation is simulated with a full 2D model. Many open GIS data are used and combined with virtual on-site inspection in the study area to analyze building exposure. Different regression curves are tested to fit the data and obtain damage curves. The results show (i) a poor correlation between relative or absolute losses and flow velocity, (ii) a low correlation between relative losses and water depth, thus confirming the existing literature. More significant correlation is found between absolute losses and water depths. The curves here derived provide a first attempt to develop damage functions for flash flood in Italy, which could be applied in similar urban and morphologic contexts.

show abstract

Multi-variable flood damage modelling with limited data using supervised learning approaches

Cited by 86 publications

References 25 publications

Seamless Estimation of Hydrometeorological Risk Across Spatial Scales

Seamless Estimation of Hydrometeorological Risk Across Spatial Scales

Leveraging machine learning for predicting flash flood damage in the Southeast US

Empirical flash flood vulnerability functions for residential buildings

Contact Info

Product

Resources

About