A New Graph-Based Deep Learning Model to Predict Flooding with Validation on a Case Study on the Humber River

Abstract

Floods are one of the most lethal natural disasters. It is crucial to forecast the timing and evolution of these events and create an advanced warning system to allow for the proper implementation of preventive measures. This work introduced a new graph-based forecasting model, namely, graph neural network sample and aggregate (GNN-SAGE), to estimate river flooding. It then validated the proposed model in the Humber River watershed in Ontario, Canada. Using past precipitation and stage data from reference and neighboring stations, the proposed GNN-SAGE model could estimate the river stage for flooding events up to 24 h ahead, improving its forecasting performance by an average of 18% compared with the persistence model and 9% compared with the graph-based model residual gated graph convolutional network (GNN-ResGated), which were used as baselines. Furthermore, GNN-SAGE generated smaller errors than those reported in the current literature. The Shapley additive explanations (SHAP) revealed that prior data from the reference station was the most significant factor for all prediction intervals, with seasonality and precipitation being more influential for longer-range forecasts. The findings positioned the proposed GNN-SAGE model as a cutting-edge solution for flood forecasting and a valuable resource for devising early flood-warning systems.

Graph-Based Deep Learning Model for Forecasting Chloride Concentration in Urban Streams to Protect Salt-Vulnerable Areas

“…This ultimately prevented the model's generalization of the problem, returning inferior results than the GNN-SAGE. The proposed model, however, could extract and identify the spatiotemporal relationship between input and output variables, improving its generalization and, consequently, its forecasting due to its better understanding of the graph-structured data [37], as verified in previous studies [41][42][43]. Figure 12 shows a scatter plot for the GNN-SAGE model.…”

“…This approach can achieve good results for short forecasting horizons. However, its performance deteriorates for further future horizons as the model cannot track the influence of the dynamics of external factors impacting future values [42,43].…”

“…The GNN-SAGE model's structure enables it to capture complex spatiotemporal patterns between a node and its neighbors, enhancing its forecasting performance compared to traditional ML and DL methods. This results in cutting-edge outcomes when applied to various time series problems [42,43]. The GNN-SAGE model was implemented using Pytorch and Scikit-Learn for Python, and its structure in the context of this study is presented in Figure 6.…”

“…The SHAP analysis, based on game theory, calculated the contribution of each input parameter used by the model for forecasting and was implemented into the present work after the model had been trained using the SHAP library (the SHAP library documentation can be read at https://shap.readthedocs.io/en/latest/, accessed on 28 August 2023) by evaluating the model for each situation where one of the independent variables is not used. This way, SHAP can identify relationships among the input data, identifying their influence, importance, and correlation over the model's output [41][42][43]55]. The determination of the influence of each variable provides deeper insight into how the model provides its results, being a viable option to explain the analyzed ML paradigm locally.…”

“…This lag is known in the literature and can be attributed to the need for more spatiotemporal data for extended time windows as the leading forecasting time increases [59][60][61]. However, the proposed model reduced this gap when applied to different time series forecasting problems, providing more accurate and reliable predictions in longer-horizon forecasts [42,43].…”