Complexity vs. Simplicity: Tradeoffs in Integrated Water Resources Models

Wednesday, 17 December 2014
Jordan Gonda, Amin A Elshorbagy, Howard S Wheater and Saman Razavi, University of Saskatchewan, Saskatoon, SK, Canada
Integrated Water Resources Management is an interdisciplinary approach to managing water. Integration often involves linking hydrologic processes with socio-economic development. When implemented through a simulation or optimization model, complexities arise. This complexity is due to the large data requirements, making it difficult to implement by the end users. Not only is computational efficiency at stake, but it becomes cumbersome to future model users. To overcome this issue the model may be simplified through emulation, at the expense of information loss. Herein lies a tradeoff: Complexity involved in an accurate, detailed model versus the transparency and saliency of a simplified model. This presentation examines the role of model emulation towards simplifying a water allocation model. The case study is located in Southern Alberta, Canada. Water here is allocated between agricultural, municipal, environmental and energy sectors. Currently, water allocation is modeled through a detailed optimization model, WRMM. Although WRMM can allocate water on a priority basis, it lacks the simplicity needed by the end user. The proposed System Dynamics-based model, SWAMP 2.0, emulates this optimization model, utilizing two scales of complexity. A regional scale spatially aggregates individual components, reducing the complexity of the original model. A local scale retains the original detail, and is contained within the regional scale. This two tiered emulation presents relevant spatial scales to water managers, who may not be interested in all the details of WRMM. By evaluating the accuracy of SWAMP 2.0 against the original allocation model, the tradeoff of accuracy for simplicity can be further realized.