Under global climate change, urban flooding occurs frequently, leading to huge economic losses and human casualties. Extreme rainfall is one of the direct and key causes of urban flooding, and accurate rainfall estimates at high spatiotemporal resolution are of great significance for real-time urban flood forecasting. Using existing rainfall intensity measurement technologies, including ground rainfall gauges, ground-based radar, and satellite remote sensing, it is challenging to obtain estimates of the desired quality and resolution. However, an approach based on processing distributed surveillance camera network imagery through machine learning algorithms to estimate rainfall intensities shows considerable promise. Here, we present a novel approach that first extracts raindrop information from the surveillance camera images (rather than using the raw imagery directly), followed by the use of convolutional neural networks to estimate rainfall intensity from the resulting raindrop information. Evaluation of the approach on 12 rainfall events under both daytime and nighttime conditions shows that generalization ability, and especially nighttime predictive performance, is significantly improved. This represents an important step toward achieving real-time, high spatiotemporal resolution, measurement of urban rainfall at relatively low cost.

An early contamination warning system with deployed water quality sensors is often used to enhance the safety of a water distribution system (WDS). While algorithms have been developed to select an optimal water quality sensor placement strategy (WQSPS) for WDSs, many of them do not account for the influences caused by future uncertainties, such as sensor failures and system changes (e.g., demand variations and configuration/expansion changes in the WDS). To this end, this paper proposes a comprehensive framework to evaluate the robustness of WQSPSs to these possible uncertainties. This is achieved by considering five different performance objectives of WQSPSs as well as possible future demand and typology variations of WDSs under a wide range of sensor failure scenarios. More specifically, an optimization problem is formulated to evaluate the robustness of the WQSPSs, in which an evolutionary-based optimization approach coupled with an efficient data-archive method is used to solve this optimization problem. The framework is demonstrated on two real-world WDSs in China. The results show that: (1) the WQSPS's robustness can be highly dependent on the performance objectives considered, implying that an appropriate objective needs to be carefully selected for each case driven by practical needs, (2) the WDS's demand and configuration changes can have a significant influence on the WQSPS's robustness, in which the solution with more sensors in or close to the affected area is likely to better cope with these system changes, and (3) the proposed framework enables critical sensors to be identified, which can then be targeted for prioritizing maintenance actions.

Most of the contamination source localization methods for water distribution systems (WDSs) assume the availability of accurate water quality models and multi-parameter online sensors, which are often out of reach of many water utilities. To address this, a novel manual grab-sampling method (MGSM) is developed to effectively and efficiently locate continuous contamination sources in a WDS using a dynamic and cyclical sampling strategy. The grab samples are collected at a pre-specified number of hydrants by the corresponding teams followed by laboratory tests. The MGSM optimizes the sampling plan at each cycle by making the probability of contamination source(s) in each sub-network as equal as possible, where sub-networks are determined by the selected hydrants and current flow pipe directions. The CS's size is reduced at each cycle by exploiting sample testing results obtained in the previous cycle until there are no further hydrants to sample from. Two real-world WDSs are used to demonstrate the effectiveness of the proposed MGSM. The results obtained show that the MGSM can significantly reduce the spatial range of the CS (to about 5% of the entire WDS) for a range of scenarios including multiple contamination sources and pipe flow direction changes. We found that an optimal number of sampling teams exists for a given WDS, representing a balanced trade-off between detection efficiency and sampling/testing budgets. Due to its relative simplicity, the proposed MGSM can be used in engineering practice straightaway and it represents a viable alternative to the methods associated with water quality models and sensors.

Urban flooding is a major issue worldwide, causing huge economic losses and serious threats to public safety. One promising way to mitigate its impacts is to develop a real-time flood risk management system; however, building such a system is often challenging due to the lack of high spatiotemporal rainfall data. While some approaches (i.e., ground rainfall stations or radar and satellite techniques) are available to measure and/or predict rainfall intensity, it is difficult to obtain accurate rainfall data with a desirable spatiotemporal resolution using these methods. This paper proposes an image-based deep learning model to estimate urban rainfall intensity with high spatial and temporal resolution. More specifically, a convolutional neural network (CNN) model called the image-based rainfall CNN (irCNN) model is developed using rainfall images collected from existing dense sensors (i.e., smart phones or transportation cameras) and their corresponding measured rainfall intensity values. The trained irCNN model is subsequently employed to efficiently estimate rainfall intensity based on the sensors’ rainfall images. Synthetic rainfall data and real rainfall images are respectively utilized to explore the irCNN's accuracy in theoretically and practically simulating rainfall intensity. The results show that the irCNN model provides rainfall estimates with a mean absolute percentage error ranging between 13.5% and 21.9%, which exceeds the performance of other state-of-the-art modeling techniques in the literature. More importantly, the main feature of the proposed irCNN is its low cost in efficiently acquiring high spatiotemporal urban rainfall data. The irCNN model provides a promising alternative for estimating urban rainfall intensity, which can greatly facilitate the development of urban flood risk management in a real-time manner.

Hydraulic modeling of a foul sewer system (FSS) enables a better understanding of the behavior of the system and its effective management. However, there is generally a lack of sufficient field measurement data for FSS model development due to the low number of in-situ sensors for data collection. To this end, this study proposes a new method to develop FSS models based on geotagged information and water consumption data from smart water meters that are readily available. Within the proposed method, each sewer manhole is firstly associated with a particular population whose size is estimated from geotagged data. Subsequently, a two-stage optimization framework is developed to identify daily time-series inflows for each manhole based on physical connections between manholes and population as well as sewer sensor observations. Finally, a new uncertainty analysis method is developed by mapping the probability distributions of water consumption captured by smart meters to the stochastic variations of wastewater discharges. Two real-world FSSs are used to demonstrate the effectiveness of the proposed method. Results show that the proposed method can significantly outperform the traditional FSS model development approach in accurately simulating the values and uncertainty ranges of FSS hydraulic variables (manhole water depths and sewer flows). The proposed method is promising due to the easy availability of geotagged information as well as water consumption data from smart water meters in near future.

This paper proposes a multistage method for burst leak localization through valve operations (VOs) and smart demand metering in district meter areas (DMAs) of water distribution systems (WDSs). Each stage includes partitioning of the DMA into two subregions using VOs and identification of potentially leaking pipes within the subregions through water balance analysis based on smart demand meters. Such a process is performed repeatedly (multiple stages) to narrow down the spatial range for pinpointing leak locations. To improve efficiency, a bisection optimization problem is formulated to localize the minimum leak areas using the lowest number of VOs, which is solved by a graph theory-based method. The utility of the proposed method is demonstrated using two DMAs (DMA1 and DMA2) of a real WDS with different topological properties. Results show that the proposed method can efficiently localize artificial burst leaks in DMA1 within 7–15% of the total pipe length, implying that the proposed method is theoretically effective in localizing pipe burst leaks. The real application to DMA2 has identified two leak regions with 2.3 and 4.2 km of pipe length (around 3–6% of the entire DMA2) using 18 VOs. These two burst leaks have been subsequently confirmed and pinpointed using listening rods by practitioners of the local water utility. These results indicate that the proposed multistage method is effective and efficient for burst leak localization, which can be promising for wide practical applications due to rapid developments of smart WDSs (e.g., smart demand meters or control valves).