Towards Urban Flood Susceptibility Mapping Using Data-Driven Models

Abstract

<p>Both frequency and severity of urban pluvial floods have been increasing due to rapid urbanization and climate change. Hydrological and two dimensional (2D) hydrodynamic models are still too computationally demanding to be used for real-time applications for large urban areas (i.e. flood management scale). As an alternative, data-driven models could be used for flood susceptibility mapping. This study evaluated and compared the performance of image-based models represented by a convolutional neural network (CNN) and point-based models represented by an artificial neural network (ANN), a random forest (RF) and a support vector machine (SVM) with regard to the spatial resolution of the input data. We also examined model transferability. Eleven variables representing topography, anthropogenic aspects and precipitation were selected to predict flood susceptibility mapping. The results showed that: (1) all models were skilful with a minimum area under the curve AUC = 0.87. (2) The RF models outperformed the other models for all spatial resolutions. (3) The CNN models were superior in terms of transferability. (4) Aspect and elevation were the most important factors for flood susceptibility mapping for image-based and point-based models respectively.</p>

Transferability of data-driven models to predict urban pluvial flood water depth in Berlin, Germany

“…The overall design of this study was as follows: firstly, we selected three areas (Figure . 1) that have frequently been flooded in the last decades based on a flood inventory (Seleem et al, 2022) gathered between 2005 and 2017. 2D hydrodynamic simulations were carried out in these areas.…”

“…In summary, deep learning was consistently superior to shallow machine learning in literature but recent studies showed the contrary (Seleem et al, 2022;Grinsztajn et al, 2022). However, shallow machine learning algorithms have not been systematically challenged in terms of transferability for urban flood modelling.…”

“…Point-based data-driven models such as the random forest (RF) algorithm have been widely used in the literature to map susceptibility for pluvial flooding (Lee et al, 2017;Chen et al, 2020;Zhao et al, 2020;Seleem et al, 2022). RF models outperformed convolutional neural networks (CNN) to map flood susceptibility in Berlin at various spatial resolutions, and showed promising results outside the training domain (Seleem et al, 2022).…”

“…Point-based data-driven models such as the random forest (RF) algorithm have been widely used in the literature to map susceptibility for pluvial flooding (Lee et al, 2017;Chen et al, 2020;Zhao et al, 2020;Seleem et al, 2022). RF models outperformed convolutional neural networks (CNN) to map flood susceptibility in Berlin at various spatial resolutions, and showed promising results outside the training domain (Seleem et al, 2022). Hou et al (2021) trained RF and K-nearest neighbour (KNN) algorithms to predicate urban pluvial flood water depth using only the rainfall characteristics as inputs, and Zahura et al ( 2020) trained a RF model to predict flood water depth in an urban coastal area using three topographic predictive features.…”

“…CNNs have recently demonstrated the potential to map urban pluvial flood susceptibility (Zhao et al, 2020(Zhao et al, , 2021Seleem et al, 2022) and flood hazard (Löwe et al, 2021;Guo et al, 2022).They are designed to extract spatial information from the input data and to handle image (raster) data without an unwarranted growth in the model complexity. Löwe et al (2021) trained a CNN model based on the U-Net architecture (Ronneberger et al, 2015) to predict urban pluvial flood water depth.…”