A Particle Filter Estimation of Taylor’s Law Fluctuation Scaling in Water Demand

Kurzfassung

Uncertainty of water demand is a major consideration for the resilience and risk management of water distribution systems, which are critical infrastructures. This demand uncertainty has been shown to exhibit heavy-tailed characteristics, i.e., extreme deviations from expected behavior are more common than conventionally anticipated. Taylor’s Law of fluctuation scaling has been used for the risk management of systems subject to similar heavy-tailed uncertainty. Using the example of urban water demand time series, we show that a Particle Filter-type Bayesian estimation can be used to uncover fluctuation scaling in infrastructure demand time series following Taylor’s Law with time-variant scaling parameters. Furthermore, we demonstrate that the observed fluctuation scaling exceeds what is assumed by conventional water demand models. We then discuss how this insight impacts the management of critical infrastructures and how future research can address the arising challenges from these results.