A multiplex network approach to road flooding prediction

Urban flooding poses risks to life, property, and health every year in the United States. Although accurate models of road, channel, and storm sewer dynamics exist, they are often not deployable at a short time scale suitable for prediction and emergency response. Using a multiplex network model of the road, channel, and storm sewer networks and the Height Above Nearest Drainage (HAND) method, urban flood prediction can be addressed with a network interaction perspective. By redefining the nodal activity during a storm event, critical nodes of the network can be identified using the network betweenness centrality on a larger scale. Here, the multiplex network is constructed on the University of Texas campus, and modeled through the severe Memorial Day 2015 storms. Critical areas of roadway flooding are identified throughout the UT Campus, corresponding to hotspots of high active betweenness centrality throughout the storm. The multiplex network approach serves as an emergency-response-oriented prediction tool for urban flooding.