Gradual cooperative coverage models for optimally locating rain gauges on an urban transportation network.

Transportation systems can be severely impacted by heavy rainfall and floods, resulting in damages to infrastructure, disruptions in the traffic flow, and compromising public safety and local economies. Most of the weather-related disruptions in transportation networks are caused by flash flooding due to heavy precipitation, resulting in a rapid accumulation of water, particularly in urban areas. Consequently, transportation managers need tools to monitor point-specific rainfall data in order to forecast flash floods and make timely decisions to minimize negative impacts. This study presents three facility location models (Maximum Coverage, P-median and Gradual-cooperative) to optimize rain gauge networks based on a flash-flood risk map that accounts for transportation. The models are compared in terms of their capacities to detect flash flood events using data from Tuscaloosa County, AL, USA. The best results were obtained by using the gradual-cooperative model, because of its propensity to allocate the sensors closer to each other in more risky areas. Location models can help to optimize rain gauge networks by incorporating distance and cooperation functions by covering areas at risk of flooding. Conclusions are drawn with respect to the fitness of each model to this specific problem.