Graph-Based Model for Efficient Data Retrieval in Incomplete Stormwater Networks

Modelling urban stormwater networks (USNs) provides valuable insights into their performance and assists in improving management strategies. However, a common and significant challenge arises from incomplete information about USNs, where essential network data (e.g., sewer diameters) are unavailable, hindering reliable hydrodynamic analysis. To address this issue, we propose an efficient and fully automated graph-based data retrieval model for USNs with incomplete information. The model automatically infers missing physical attributes, including sewer diameter and slope, by considering the topological features (e.g., connectivity) and hierarchical patterns observed in sewer diameter variations. The framework was tested on a real-world USN with a complete dataset, where data gaps were artificially introduced by randomly removing sewer diameter and slope information ranging from 10% to 90%. Each data gap scenario was repeated 100 times, resulting in 900 incomplete USN configurations. The results demonstrated that the model efficiently retrieved missing data with high accuracy, achieving up to 90% data recovery while accurately reproducing hydrodynamic attributes, such as flow rates. This model provides an efficient tool for water utilities managing incomplete USNs, enabling them to conduct various hydrodynamic analyses.