Informing water resource analyses with instrumental and palaeoclimate records — Australian Meteorological and Oceanographic Society
  1. Schedule
  2. 25. The science of drought
  3. Informing water resource analyses with instrumental and palaeoclimate records

Informing water resource analyses with instrumental and palaeoclimate records (#262)

Natasha Ballis 1 , Murray C Peel 1 , Rory J Nathan 1 , Benjamin J Henley 2 , David J Karoly 3
  1. Environmental Hydrology and Water Resources, Department of Infrastructure Engineering, University of Melbourne, Parkville, Victoria, Australia
  2. School of Earth Sciences, University of Melbourne, Parkville, Victoria, Australia
  3. NESP Earth Systems and Climate Change Hub, CSIRO, Aspendale, Victoria, Australia

A number of water resource systems across Australia have been tested by, or are at risk of being tested by, periods of protracted drought. However, assessment of these water resource systems are hindered by short instrumental records.

Water resource analyses are highly sensitive to the hydrologic variability at a site, and to understand this variability, hydrologists need sufficiently long records. For storage yield analyses, Pretto et al. (1997) show that sequence lengths of 1000 years or more are required to identify a steady-state estimate of storage capacity. For stochastic generation of hydrologic data, Thyer et al. (2006) demonstrate that depending on the persistence structure of annual rainfall, record lengths of at least a hundred and possibly more than 500 years are needed to identify a stochastic model and estimate its parameters. However, in Australia, the longest annual instrumental records of rainfall and streamflow are approximately 160 years long, with most being much shorter.  

Palaeoclimate reconstructions, which can be substantially longer than instrumental records, provide a potential opportunity to better understand long-term hydrologic variability at a site. Published seasonally- and annually-resolved reconstructions of hydrologic variables in Australia have a median length of approximately 350 years, and a maximum length of over 1000 years. However, there are aspects of published palaeoclimate reconstructions that impact their use in water resource analyses. These include the amount, type and treatment of uncertainty, the reporting of model parameter metrics that are relevant to a water resource analysis, and also the direct relevance of the reconstructed variable.

Considering the potential value and also the challenges of using palaeoclimate reconstructions in water resource analyses, we investigate the properties of reconstructions that are suitable to be preserved in stochastic water resource analyses. Further, we investigate how to inform a stochastic model with properties of both instrumental and palaeoclimate records.

  1. Pretto, P.B., Chiew, F.H., McMahon, T.A., Vogel, R.M. and Stedinger, J.R., 1997. The (mis) behavior of behavior analysis storage estimates. Water Resources Research, 33(4), pp.703-709.
  2. Thyer, M., Frost, A.J. and Kuczera, G., 2006. Parameter estimation and model identification for stochastic models of annual hydrological data: Is the observed record long enough?. Journal of Hydrology, 330(1-2), pp.313-328.
#amos2020