Integrated catchment-reservoir modeling to support smart management of reservoir resources in Singapore

Yang Peipei, Lloyd Chua and Kim N. Irvine


Runoff from urban catchments is one of the four national taps, or sources of municipal water, for Singapore. Currently, Singapore captures runoff from two-thirds of its surface and routes this runoff via 8,000 km of waterways to be stored in 17 reservoirs throughout the island. An average of 2,400 mm of rainfall per year provides abundant supply of runoff to the reservoirs, but it is important to consider both water quality and flooding potential in managing the water resources. To date, we have integrated the application of PCSWMM to model catchment runoff discharging to three reservoirs and the ELCOM-CAEDYM (Estuary, Lake and Coastal Ocean Model - Computational Aquatic Ecosystem DYnamics Model) model to three-dimensionally represent reservoir hydrodynamics and water quality. The applications have been diverse, including evaluating the impact of floating solar panels on reservoir quality; enhanced consideration of the spatial variability of wind on a downtown reservoir using a combination of computational fluid dynamics and probability algorithms to assess the impact on reservoir hydrodynamics; and assessment of rain gauge density on catchment runoff and reservoir water quality evaluations. Singapore has been identified as the greenest city in the world, based on an MIT assessment of remotely sensed canopy cover (approximately 30% cover). Vancouver was second at nearly 26% cover. While the greenspace enhances urban liveability, the forested land uses surrounding the reservoirs present some particular challenges to modelling runoff.  Examples of these different projects are reviewed. Singapore has established a smart nation policy and as such, water management is moving towards real time assessment and decision-making which requires rapid execution of the integrated catchment-reservoir models. Our approach to adapting the integrated models to run on the cloud as a means to reduce computational time also is discussed.

Permanent link: