Machine-learning blends of geomorphic descriptors: value and limitations for flood hazard assessment across large floodplains

Abstract

Abstract. Recent literature shows several examples of simplified approaches that perform flood hazard (FH) assessment and mapping across large geographical areas on the basis of fast-computing geomorphic descriptors. These approaches may consider a single index (univariate) or use a set of indices simultaneously (multivariate). What is the potential and accuracy of multivariate approaches relative to univariate ones? Can we effectively use these methods for extrapolation purposes, i.e., FH assessment outside the region used for setting up the model? Our study addresses these open problems by considering two separate issues: (1) mapping flood-prone areas and (2) predicting the expected water depth for a given inundation scenario. We blend seven geomorphic descriptors through decision tree models trained on target FH maps, referring to a large study area (∼ 105 km2). We discuss the potential of multivariate approaches relative to the performance of a selected univariate model and on the basis of multiple extrapolation experiments, where models are tested outside their training region. Our results show that multivariate approaches may (a) significantly enhance flood-prone area delineation (accuracy: 92 %) relative to univariate ones (accuracy: 84 %), (b) provide accurate predictions of expected inundation depths (determination coefficient ∼ 0.7), and (c) produce encouraging results in extrapolation.

Estimations of Crop Losses Due to Flood Using Multiple Sources of Information and Models: The Case Study of the Panaro River

“…The level of detail required by the hazard simulation is a topic that deserves further exploration, since the selection of a specific methodology depends on the compromise between its accuracy and the data requirement [22]. Nowadays, many different methodologies are available for the reproduction and observation of hazardous variables such as satellite technologies, simplified approaches [23][24][25], etc. The adoption of hydraulic models represents a classic and well-consolidated approach.…”

Geomorphic Flood Hazard Mapping: From Floodplain Delineation to Flood-Hazard Characterization

“…The present research shows the application, comparison and discussion of two DEM-based models for flood-hazard mapping, one univariate (i.e. a single DEM-based geomorphic index is used) and one multivariate (a variety of indices is used; see also Magnini et al 2022). We select Italy as the study area, given its remarkable variability of hydrological, climatic, and geomorphological characteristics.…”

“…Manfreda et al 2015), and we combine them to model flood hazard through a decision tree classifier (see e.g. Magnini et al 2022). After the calibration over specific fractions of the study area, the two models are applied to the whole region.…”

“…The methods adopted for setting up the models are inspired by previous literature (see e.g. Tavares da , Magnini et al 2022), but the nature and goal of our application and discussion are new. Differently from what has been done in relevant previous studies (see e.g.…”

“…Differently from what has been done in relevant previous studies (see e.g. Tavares da Costa et al 2020, Magnini et al 2022, we do not limit our assessment of models' performance to the similarity of their output to the target flood map in test areas that differ from training ones. Instead, we perform a detailed assessment of flood DEM-based susceptibility maps in areas where the calibration flood map may be inhomogeneous or inaccurate -which is a common situation in practice.…”