Oral Presentation Australasian Groundwater Conference 2017

Statistical analysis and interpolation of water level data at the Bureau of Meteorology (#210)

Brendan Dimech 1 , Dr. Elisabetta Carrara 1 , Dr. Tim Peterson 2
  1. Bureau of Meterology, Melbourne
  2. University of Melbourne, Melbourne

Under the Commonwealth Water Act 2007 the Bureau of Meteorology is tasked with collecting groundwater level time series data from states and territories on a regular basis. Currently groundwater level data has several shortcomings, including inconsistent information on quality of data as well as the intermittent and non-continuous readings for most bores across Australia.

To alleviate these shortcomings HydroSight (http://peterson-tim-j.github.io/HydroSight/) was created through an ARC linkage grant supported by the Bureau.  Now the Bureau is operationalising several of the modules from this tool including:

  • Statistical analysis for data quality assessment – for automating the numerical identification of water level data monitoring errors and outliers.
  • Create monthly times series – for automating the time series infilling of groundwater level data using rainfall and a simplified soil moisture transfer function noise model.

The goal for this project is to identify observations that are statistically anomalous and/or erroneous.  To achieve this goal, the Bureau needs to run the tool for over 100,000 bores that have water level observations. HydroSight's code has been extracted and modified to read and write directly to the Bureau's databases and fit with the automated processes for ingesting state data.

The analysis uses a set of hydrogeologically relevant heuristics to identify physically questionable observation (e.g. long periods of constant head). Observations are then checked for outliers by iteratively applying a globally calibrated double exponential smoothing time-series model. The remaining data is then interpolated to a regular month observation frequency using a calibrated nonlinear transfer function noise model. The outcome is groundwater hydrographs interpolated to a regular timestep that accounts for between observation forcing dynamics.

The outputs will be made available within the Bureau’s suite of groundwater products and will be used to inform the groundwater processes in the Water Resources Assessment Landscape (AWRA-L) model.

  • We are offering awards for Career and Early Career presentations and posters. Please indicate length of time since highest degree completed.: 0 to 5 Years or currently studying