...
Background
Model calibration analysis is not available for calibrating Urban scenarios.is used to ensure that modelled outputs sufficiently represent available observed values. The extent to which modelled values match observed values is quantified using an objective function such as the Nash Sutcliffe Efficiency (NSE) coefficient. Source has several objective functions to choose from - see the UG chapter Calibration Wizard for catchments for a description of available objective functions. The process of calibration involves adjusting model parameters through multiple model runs until a satisfactory value of the chosen Objective Function is obtained. The choice of parameters and appropriate Objective Function is key to this process.
Calibration of urban demand models can be challenging as there are hundreds of parameters that can be varied. Fortunately, the values of many parameters are well informed by the available input data. Due to the large number of parameters and the resulting high degree of parameter non-uniqueness, calibration using standard automatic optimisation algorithms is unlikely to be successful. It is proposed that a process of manual parameter adjustment is adopted, focussing on sensitive parameters that are known to have a high degree of uncertainty.
The Urban Developer (UD) plugin gives the user access to an Urban Developer Scenario in Source enabling the user to model urban demands at household level. One of the main advantages of the Urban Developer plugin is to facilitate the upscaling of household scale end-use demands to city- or water authority-scale bulk water demands in a Source catchments or river system model. To achieve this, outputs from an Urban Developer Scenario run can be transferred to a Source scenario where they can be used to represent bulk urban demands at a Source Water User node. Calibrating these urban demands requires an understanding of the representation of demands at the household (end-use) level within the Urban Developer scenario as well as at the bulk water demand level within the Source scenario.
Figure 1 uses an average of the default values[1] applied in a UD Plugin Urban Behavioural Demand node to illustrate the components of urban demand over a two-year period. Constant non-time-varying end-use efficiencies are applied and therefore indoor demand remains constant. Indoor demands will be more consistent, with long term changes occurring mostly as a result of population growth and also the implementation of changes in end-use efficiencies. The introduction of rainwater Tanks and greywater use will also impact on this long-term trend. Outdoor demands, however, are more variable, reflecting climate variations which can be the result of seasonal changes as well changes in use due to restrictions introduced during drought periods.
Key to urban demand calibration is understanding this split between Indoor and Outdoor demands in a specific study area. Indoor demand calibration will determine the consistent, base component of a demand time series while outdoor demand calibration will determine the ‘peaks’ and ‘troughs’ of a demand time series.
Figure 1 Urban demand components
[1] Default end-use values are based on those reported by Roberts (2005) as the result of a study involving Yarra Valley Water residential customers in Melbourne, Australia
| Page Tree | ||
|---|---|---|
|