Water quality modelling for the Gardens by the Bay, Singapore
by Gerard Pijcke
Water quality modelling for Gardens by the Bay, Singapore
The water quality of a 5 hectare lake system in Gardens by the Bay, Singapore is modelled using a combined hydrodynamic (Delft3D) and water quality model (DELWAQ). The lake system comprises two lakes, Dragonfly and Kingfisher Lake, connected by a 5 metre wide stream. Inflow is from runoff from the 50 hectare catchment and from pumping from the neighbouring Marina Bay Reservoir at a rate of 20 Ls-1, 24/7. Outflow is through the irrigation offtake in Dragonfly Lake and by excess overflow from Dragonfly Lake to Marina Bay Reservoir. Monitoring and modelling of the water quality of the lakes is important for reasons of ecological sustainability, public health concerns and aesthetic purposes.
For twelve inflow locations, during three wet and three dry events, the total phosphors concentrations were determined by colorimetry and total nitrogen, nitrite, nitrate and phosphate concentrations by ion chromatography. Nutrient inflows into the lake system in Gardens by the Bay range between 2 and 20 mgL-1 for total nitrogen and 0.05 and 1.7 mgL-1 for total phosphorus and are well above threshold values for total nitrogen (1.0 mgL-1) and total phosphorus (0.06 mgL-1) by Singapore’s Public Utility Board (PUB). The inflow of nutrients is approximately equally divided over inflow from the Marina Bay Reservoir and catchment inflows. Non-point sources contribute approximately 40% of the total phosphorus and total nitrogen load in [kg yr-1] and [kg yr-1ha-1] associated with the catchment runoff (that is excluding inflow from Marina Bay Reservoir).
The water quality model represent the water column temperature well with Root Mean Squared Error of 0.38 oC in Dragonfly Lake and 0.63 oC in Kingfisher Lake and Nash-Sutcliffe Efficiency of 0.73 and 0.72 respectively. Due to the use of average concentrations as model input values, fluctuations in the concentrations for dissolved oxygen, chlorophyll-a, total nitrogen and total phosphorus are not well captured by the model. Concentrations for water quality variables in Kingfisher Lake and Dragonfly Lake are in space. Vertical concentration differences for the modelled variables at the deepest location (4 metres) in Dragonfly Lake are highest for dissolved oxygen, 0.20 mgL-1, and chlorophyll-a, 0.37 µgL-1. Kingfisher Lake strongly responds to changes in the characteristics of inflow from Marina Bay Reservoir; the water quality in Dragonfly Lake is influenced more by catchment inflows. The combination of observation data and model data show there is a net inflow of nutrients in the lake system in Gardens by the Bay and show nutrients are accumulating in the biomass and sediment in Gardens by the Bay.
Further studies should verify the nutrient loads from subcatchments through comparison of land-use types and fertilizer application in the Gardens. Sediment sampling will be required to verify the magnitude and rate of nutrient accumulation in the sediment layer. The sensitivity of the model to time-varying input data can be assessed by using the data collected for individual events rather than the average values. The three dimensional model can be simplified to a two-dimensional model. More extensive studies should indicate whether a further simplification of the model in which Dragonfly and Kingfisher Lake are represented as two fully mixed reservoirs is possible.
Student: G. Pijcke
Committee: Prof. Dr. Ir. N.C. van de Giesen, A.W. Heemink, G.H.W. Schoups, V. Babovic, S. Bayen, K.C. Mynampati