Abstract : Accurate maps of wastewater networks in cities are mandatory for an integrated management of water resources. However, in many countries around the world this information is unavailable or inaccurate. A new mapping method is put forward to create wastewater network maps using manhole cover locations as a prime information source. These locations could be available via ground surveys , remote sensing techniques or stakeholder's databases. A new algorithm is developed which considers manhole covers as the nodes of the network and connects them automatically. It minimizes cost functions defined by industry rules thus generating an optimized network. The various input data and the rules used to build the deterministic tree-shaped graph being uncertain, a stochastic version of the algorithm is also put forward to generate a set of probable networks in addition to the optimized one. The method is tested on the wastewater networks of Prades-le-Lez and validated on the town of Ramonville Saint Agne. Both towns are located in Southern France, are part of the two most dynamic metropolitan areas of France and are under constant urban pressure due to their proximity with the cities of Montpellier and Toulouse. The shape and topology of the mapped networks are compared to the actual ones. The results indicate an overall good agreement between the layouts of the real and generated networks. The proposed algorithm may thus be used to map wastewater networks from sampled georeferenced manhole covers, elevation and street network databases. Additional sources of information are however necessary to recreate the networks full geometry and insure proper conveyance. The low error values and high scores for completeness, correctness and quality indicate that the method is robust and may be adapted and tested on other study zones.
