Towards the quantification of water quantity and quality impacts of rainwater tanks in South East Queensland

Abstract

To help ensure the security of water supply to urban catchments, a popular policy choice is to promote the installation of domestic rainwater tanks. While yield is primarily considered, this policy also causes a change in urban-runoff stormwater volumes and consequentially, nutrient export. Modelling tools are used to predict the yield, volumetric reliability, overflow volume, and nutrient export that the deployment of rainwater tanks will cause. These modelling tools commonly utilise an up-scaling approach to analyse the behaviour of multiple tanks, where the performance of a single tank with average characteristics is linearly scaled up to represent a larger cluster of tanks. Previous research has shown that this up-scaling method significantly overestimates the yield and volumetric reliability (Mitchell et al. 2008), and underestimates the overflow volume of the cluster (Neumann et al. 2011). A sensitivity analysis of the parameters used to represent a rainwater tank (roof catchment areas, tank storage capacities, demand) was carried out, using on-hand data for Melbourne water demand, rainfall and maximum temperature (Neumann et al. 2011). It identified that the non linearity of the tank yield and overflow in relation to some of the model parameters means that the adoption of an "average" (i.e. spatially lumped) tank to represent the behaviour of the entire cluster is subject to significant errors. This paper establishes that, accordingly, nutrient export loads are also underestimated, and describes a stochastic water balance and quality model to effectively quantify the overflow nutrient load from a cluster of rainwater tanks at the catchment-scale. In 2008, to reduce mains water consumption the South East Queensland region made rainwater tanks mandatory in new detached dwellings. It is important to accurately estimate overflow nutrient loads discharged to stormwater to identify changes at the catchment scale caused by this wide implementation of rainwater tanks. The purpose of this paper is to apply the method described in Neumann et al (2011) and Maheepala et al (2011) to South East Queensland data, avoiding the limitations of spatial lumping of the performance of an average rainwater tank. This study examines the impact on overflow loads and volumes, and potable water savings at a catchment scale of a simulated cluster of RWTs in the Brisbane region, and also analyses the model's sensitivity to geographic climate parameters. By using Brisbane water demand, rainfall, and maximum temperature data, this paper illustrates the effect that the varied climates of Melbourne and Brisbane have on the results of the model, via different rainfall and usage patterns. The results of this study for the Brisbane Local Government Area indicate an overestimation of volumetric reliability and yield, and an underestimation of the overflow volume and inflow load as well as the outflow load, which varies between 15% and 27%, depending on the nutrient. Therefore it is not recommended to use an average tank to predict the performance of a cluster of household rainwater tanks to find their contribution to potable water savings, overflow volumes and subsequent nutrient and sediment export to stormwater. Instead, we recommend using stochastic simulation of rainwater tanks, which will include the use of probability distributions to represent tank characteristics and stochastic representation of end use water demands, calibrated using local climate and observed demand and rainwater tank data.