H13A-1028:
How should we build a generic open-source water management simulator?

Monday, 15 December 2014
Majed Khadem1, Philipp Meier2, David E Rheinheimer3, Silvia Padula4, Evgenii Matrosov5, Philipp Selby5, Stephen Knox5 and Julien J. Harou5, (1)University of Manchester, Manchester, M13, United Kingdom, (2)EAWAG Swiss Federal Institute of Aquatic Science and Technology, Duebendorf, Switzerland, (3)University of California Merced, Merced, CA, United States, (4)University College London, London, United Kingdom, (5)University of Manchester, Manchester, United Kingdom
Abstract:
Increasing water needs for agriculture, industry and cities mean effective and flexible water resource system management tools will remain in high demand. Currently many regions or countries use simulators that have been adapted over time to their unique system properties and water management rules and realities. Most regions operate with a preferred short-list of water management and planning decision support systems. Is there scope for a simulator, shared within the water management community, that could be adapted to different contexts, integrate community contributions, and connect to generic data and model management software? What role could open-source play in such a project? How could a genericuser-interface and data/model management software sustainably be attached to this model or suite of models? Finally, how could such a system effectively leverage existing model formulations, modeling technologies and software? These questions are addressed by the initial work presented here. We introduce a generic water resource simulation formulation that enables and integrates both rule-based and optimization driven technologies. We suggest how it could be linked to other sub-models allowing for detailed agent-based simulation of water management behaviours. An early formulation is applied as an example to the Thames water resource system in the UK. The model uses centralised optimisation to calculate allocations but allows for rule-based operations as well in an effort to represent observed behaviours and rules with fidelity. The model is linked through import/export commands to a generic network model platform named Hydra. Benefits and limitations of the approach are discussed and planned work and potential use cases are outlined.