文章基本信息

标题：Testing empirical and synthetic flood damage models: the case of Italy
本地全文：下载
作者：Amadio, Mattia ; Scorzini, Anna Rita ; Carisi, Francesca 等
期刊名称：Natural Hazards and Earth System Sciences
电子版ISSN：2195-9269
出版年度：2019
卷号：19
期号：3
页码：661-678
DOI：10.5194/nhess-19-661-2019
语种：English
出版社：Copernicus Publications
摘要：Flood risk management generally relies on economic assessments performed byusing flood loss models of different complexity, ranging from simpleunivariable models to more complex multivariable models. The latter account for alarge number of hazard, exposure and vulnerability factors, beingpotentially more robust when extensive input information is available. Wecollected a comprehensive data set related to three recent major flood eventsin northern Italy (Adda 2002, Bacchiglione 2010 and Secchia 2014), includingflood hazard features (depth, velocity and duration), buildingcharacteristics (size, type, quality, economic value) and reported losses.The objective of this study is to compare the performances of expert-basedand empirical (both uni- and multivariable) damage models for estimating thepotential economic costs of flood events to residential buildings. Theperformances of four literature flood damage models of different natures andcomplexities are compared with those of univariable, bivariable andmultivariable models trained and tested by using empirical records fromItaly. The uni- and bivariable models are developed by using linear,logarithmic and square root regression, whereas multivariable models arebased on two machine-learning techniques: random forest and artificial neural networks. Results provide important insights about the choice of thedamage modelling approach for operational disaster risk management. Ourfindings suggest that multivariable models have better potential forproducing reliable damage estimates when extensive ancillary data for floodevent characterisation are available, while univariable models can beadequate if data are scarce. The analysis also highlights that expert-basedsynthetic models are likely better suited for transferability to other areascompared to empirically based flood damage models.