Deep Learning Methods for Flood Mapping: A Review of Existing Applications and Future Research Directions

Abstract

Abstract. Deep Learning techniques have been increasingly used in flood management to overcome the limitations of accurate, yet slow, numerical models, and to improve the results of traditional methods for flood mapping. In this paper, we review 58 recent publications to outline the state-of-the-art of the field, identify knowledge gaps, and propose future research directions. The review focuses on the type of deep learning models used for various flood mapping applications, the flood types considered, the spatial scale of the studied events, and the data used for model development. The results show that models based on convolutional layers are usually more accurate as they leverage inductive biases to better process the spatial characteristics of the flooding events. Models based on fully-connected layers, instead, provide accurate results when coupled with other statistical models. Deep learning models showed increased accuracy when compared to traditional approaches and increased speed when compared to numerical methods. While there exist several applications in flood susceptibility, inundation, and hazard mapping, more work is needed to understand how deep learning can assist real-time flood warning during an emergency, and how it can be employed to estimate flood risk. A major challenge lies in developing deep learning models that can generalize to unseen case studies. Furthermore, all reviewed models and their outputs, are deterministic, with limited considerations for uncertainties in outcomes and probabilistic predictions. The authors argue that these identified gaps can be addressed by exploiting recent fundamental advancements in deep learning or by taking inspiration from developments in other applied areas. Models based on graph neural networks and neural operators can work with arbitrarily structured data and thus should be capable of generalizing across different case studies and could account for complex interactions with the natural and built environment. Physics-based deep learning can be used to preserve the underlying physical equations resulting in more reliable speed-up alternatives for numerical models. Similarly, probabilistic models can be built by resorting to Deep Gaussian Processes or Bayesian neural networks.

A Machine Learning-Based Surrogate Model for the Identification of Risk Zones Due to Off-Stream Reservoir Failure

“…The results in terms of class probability provide additional information that can be considered for decision making. In this regard, other algorithms more specifically developed for considering uncertainty may also be useful, such as deep Gaussian processes [67,68].…”

Unraveling the complexities of urban fluvial flood hydraulics through AI

“…From Scikit-Learn, Decision Tree Classifier is used to perform the classification task on flooding location. Gini Impurity is used as a loss function of the DT classifier 44 .…”

“…Neural network models are highly sensitive to the initial randomization of weights, number of layers, number of neurons, activation functions and algorithm to choose (e.g., gradients descent) 41 – 43 . In the traditional ML and DL methods, a major challenge lies in developing models that can generalize to unseen case studies and sites 44 . This investigation overcomes this obstacle by leveraging a two-stage approach with a set of ML classifiers and a DNN-based regression model used to predict the flooded extends and magnitudes with a comprehensive set of urban hydraulic features.…”

Connecting hydrological modelling and forecasting from global to local scales: Perspectives from an international joint virtual workshop

“…Lack of appropriate training data for ML and the incorporation of physical principles within ML networks were identified as open challenges, requiring routine evaluation, diagnosis, and domainknowledge integration to deliver more skillful predictions globally. Furthermore, the large quantities of training data necessary for Deep Learning (DL) could additionally be sourced from smartphone camera pictures and videos, or from social media, along with leveraging generative models, such as Generalised Adversarial Networks or GANs, to produce synthetic data for datascarce regions (Bentivoglio et al, 2022). Moreover, Physics Informed Neural Networks (PINNs) also hold promise for flood modelling in combination with methods from deep Gaussian processes or Bayesian neural networks to evaluate model and data uncertainties through probabilistic hazard mapping (Mahesh et al, 2022).…”