Oral Presentation Australasian Groundwater Conference 2017

Is model calibration alone good enough? A case study of the Chowilla Floodplain modelling demonstrates the benefits of model post-audits. (#112)

Chris Li 1 , Juliette Woods 1 , Matt Gibbs 1 , Jan Whittle 1 , Tony Herbert 1
  1. DEWNR, Adelaide, SA, Australia

The Chowilla Floodplain is one of the most important floodplains in South Australia due to its ecological and cultural significance. However, due to River Murray regulation and the consequent reduced flood frequency, the floodplain suffers greatly from a lack of soil moisture and salt accumulation. To address this problem, a regulator has been constructed to enable the floodplain anabranch water levels to be temporarily raised, and large areas of the floodplain to be inundated. However, this will also raise the naturally-saline groundwater, and subsequently increase the saline groundwater discharge to the floodplain anabranches, ultimately to the River Murray. There are policy obligations for the salinity impact of such an action to be assessed via modelling.

A coupled surface water–groundwater modelling approach is used to simulate the salinity impact of regulator operations. The surface water model (MIKEFLOOD) estimates the changes in surface water levels and inundation areas caused by regulator operations. The outputs are then fed into the groundwater model (MODFLOW) to simulate the resulting groundwater discharge to the floodplain anabranches and River Murray. The groundwater model was constructed and calibrated in 2012, with the riverbed conductance being the most sensitive and uncertain parameter.

A number of regulator-operation trials have occurred in recent years, which enabled additional data such as salt load to be collected. This study uses these data to undertake a model post-audit, which proves to be beneficial as it further constrains the riverbed conductance, identifies processes that are not captured in the model, and allows the model’s predictive capability to be assessed. It is recommended that when new data becomes available, model post-audits should be proactively carried out, especially on models used for management decisions, to provide additional insights and establish greater confidence in prediction outcomes.

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