10.26190/5D534D6145702
Barker, A
A
Barker
Pitman, A
A
Pitman
Evans, J
J
Evans
Spaninks, F
F
Spaninks
Uthayakumaran, L
L
Uthayakumaran
Probabilistic forecasts for water consumption in Sydney, Australia from stochastic weather scenarios and a panel data consumption model.
UNSWorks, UNSW
2019
Report
Urban water consumption
Stochastic weather generation
Climate change
UNSW Sydney
http://handle.unsw.edu.au/1959.4/unsworks_60200
UNSWorks Copyright
Medium-term (1-10 year) probabilistic forecasts of urban water consumption can be useful in providing a range of possible outcomes for input into the budget and infrastructure planning of a water utility. A stochastic weather generator is developed in this study to generate multiple weather scenarios spanning the financial years 2014/15 to 2024/25 using a daily time step. These weather scenarios are then used as inputs to an existing panel data water consumption model. The resulting water demand forecasts form a probabilistic forecast of water consumption with an average range of 7.3%. In addition, the weather scenarios are used to examine the weather sensitivity of forecast consumption. We demonstrate the importance of accurate simulation of interannual variability, intersite correlation and intervariable correlation of the simulated weather variables in obtaining a realistic range of probabilistic consumption forecasts.