Modeling Stormwater Runoff for Reservoir Management in a Mixed Land Use Urban Watershed: Challenges of the Tropical Monsoon Climate of Singapore

K.N. Irvine, National Institute of Education, Nanyang Technological University, Singapore, Le S.H and Chau H.C.L, School of Civil and Environmental Engineering, Nanyang Technological University, Singapore


Under its Four National Taps policy to provide water security, stormwater runoff is now captured from two-thirds of Singapore’s land area and stored in 17 reservoirs for subsequent use. All major estuaries have been dammed to create reservoirs, and the Public Utilities Board (PUB, the water resource management agency for Singapore) intends to capture water from remaining streams near the shoreline, which will increase Singapore’s water catchment area to 90% by 2060. Ultimately, stormwater is expected to meet 30% of the country’s water demand. Management of the reservoirs to help ensure water security clearly is a priority and an important aspect of this management is to understand the water quantity and quality dynamics of the stormwater runoff inputs. This study examines runoff to the Kranji Reservoir which was constructed in 1975 and has a surface area of 450 ha with a storage capacity of 15,850,000 m3. The catchment area for the reservoir is approximately 5,700 ha, with land use being 53% urban (high density and low density residential, commercial and service, light industry), 33.5% forest (principally a military training area), 5% agriculture (vegetables, horticulture and ornamental fish), and 8.5% water. The catchment was discretized based on land use and 13 automated flow and water quality sampling stations were installed, together with 6 tipping bucket rain gauges. Runoff was modeled with PCSWMM using a discrete event and continuous approach. Model performance was quite good, with the Nash-Sutcliffe Coefficient and r2 between observed and modeled flow ranging between 0.599-0.733 and 0.601-0.735, respectively, for continuous calibration runs. To date, while event-based water quality modeling has provided reasonable representations of the observed data, continuous modeling has been less accurate.

Several challenges were encountered while calibrating the model for runoff and water quality. The climate of Singapore is a tropical rainforest with no true dry season. Annual mean rainfall is approximately 2,343 mm and annual maximum 60 minute rainfall intensity ranges between 70 and 130 mm hr-1. In keeping with its tropical rainforest classification, the mean number of days per month with rain ranges between 11 and 19. Although data for up to 6 rain gauges were available, it was clear from the monitored flow, model results for some events, and weather radar, that there is considerable spatial variability in Singapore’s rainfall patterns. Even 6 rain gauges within the Kranji watershed were not capable of entirely capturing this spatial variability for stormwater modeling purposes. Infiltration measurements with a double O-ring infiltrometer showed that infiltration rates were well represented by the Horton equation, but exhibited considerable spatial variability (even in a small park area), with maximum initial infiltration capacity ranging between 150 and 800 mm hr-1. Despite relatively high infiltration capacities, because of the high intensity rains in Singapore, Hortonian overland flow on pervious surfaces frequently is observed and this may have an important impact on water quality. The large forested subcatchment has dense, rainforest vegetation which we suspect has high interception storage capacity that was difficult to capture realistically within the model. Finally, because it rains, on average, every other day, there essentially is no buildup period between events and therefore the traditional buildup-washoff approach to water quality modeling is not appropriate in the context of Singapore’s climate

Permanent link: