Improving rating curves with 2D hydrodynamic modelling

The use of gauged flows for flood hydrology depends on rating curves, relating flow rate to gauge height. Rating curves are developed by undertaking a series of discrete flow measurements during flood events. When a flow occurs that exceeds previous gauging records, particularly widespread overbank flow conditions, an extrapolation of the rating curve is required to allow flow rate estimation. At numerous gauges across Victoria, the January 2011 gauge height exceeded the highest rated level by a significant margin. In these cases, extrapolated rating curves are developed by Thiess to allow flow estimation for these extreme events. However, in a number of flood studies undertaken by Water Technology, calibrated hydraulic models have shown that rating curve extrapolation does not accurately predict peak flows for the January 2011 event for many gauge locations. For some gauges the extrapolated rating curve underestimated flow and on others (such as Murchison) significantly over estimated flow. The peak flows of these significantly large events within the extrapolated range of rating curves are critical in the estimation of design flood peak flows using flood frequency analysis. Taking the Murchison gauge on the Goulburn River as an example, a recent revision to the extrapolated section of the rating curve by Thiess resulted in a change in the estimated peak flow of the largest flow on record from 195,000 ML/d to 311,000 ML/d, a 60% increase. This significant change in peak flow means that the revised 1% AEP design peak flow would be larger than the previous 0.2% AEP design peak flow. This extrapolated section of the rating curve therefore could result in dramatic changes in design flood levels, influencing planning decisions. This paper discusses three examples of current rating curves, with verification, revision and extrapolation of the curves undertaken to estimate peak flows at Murchison, Quambatook and Skipton. For each of these gauges, a 2D hydrodynamic model was calibrated to known flood levels and flows, and used to verify, revise or extend the high-flow or extrapolated section of the rating curve. The 2D modelling approach is superior to traditional 1D modelling, Manning's equation or extrapolation based methods of extending a rating curve. The 2D hydraulic modelling is less sensitive to model schematisation and parameterisation and takes into account the spatial variability of overbank flow conditions. This paper highlights the need to consider uncertainty in rating curves, particularly in their extrapolated section, as a standard component of the flow estimation process. It also demonstrates the role 2D hydrodynamic modelling can play in the verification, revision and extrapolation of rating curves. The method used in this study is widely applicable to the verification of other gauges with extrapolated rating curves.