Graph neural network-based anomaly detection for river network systems

Abstract

<ns3:p> <ns3:bold>Background:</ns3:bold> Water is the lifeblood of river networks, and its quality plays a crucial role in sustaining both aquatic ecosystems and human societies. Real-time monitoring of water quality is increasingly reliant on in-situ sensor technology. </ns3:p> <ns3:p>Anomaly detection is crucial for identifying erroneous patterns in sensor data, but can be a challenging task due to the complexity and variability of the data, even under typical conditions. This paper presents a solution to the challenging task of anomaly detection for river network sensor data, which is essential for accurate and continuous monitoring.</ns3:p> <ns3:p> <ns3:bold>Methods:</ns3:bold> We use a graph neural network model, the recently proposed Graph Deviation Network (GDN), which employs graph attention-based forecasting to capture the complex spatio-temporal relationships between sensors. </ns3:p> <ns3:p>We propose an alternate anomaly threshold criteria for the model, GDN+, based on the learned graph. To evaluate the model's efficacy, we introduce new benchmarking simulation experiments with highly-sophisticated dependency structures and subsequence anomalies of various types. We also introduce software called gnnad.</ns3:p> <ns3:p> <ns3:bold>Results: </ns3:bold> We further examine the strengths and weaknesses of this baseline approach, GDN, in comparison to other benchmarking methods on complex real-world river network data. </ns3:p> <ns3:p> <ns3:bold>Conclusions:</ns3:bold> Findings suggest that GDN+ outperforms the baseline approach in high-dimensional data, while also providing improved interpretability. </ns3:p>