Uncertainty in streamflow rating curves: Methods, controls and consequences

Discharge time series' are one of the core data sets used in hydrological investigations. Errors in the data mainly occur through uncertainty in gauging (measurement uncertainty) and uncertainty in determination of the stage-discharge relationship (rating curve uncertainty). Thirty-six flow gauges from the Namoi River catchment, Australia, were examined to explore how rating curve uncertainty affects gauge reliability and uncertainty of observed flow records. The analysis focused on the deviations in gaugings from the rating curves because standard (statistical) uncertainty methods could not be applied. Deviations of greater/lesser than 10% were considered significant to allow for a measurement uncertainty threshold of 10%, determined from quality coding of gaugings and operational procedures. The deviations in gaugings were compared against various factors to examine trends and identify major controls, including stage height, date, month, rating table, gauging frequency and quality, catchment area and type of control. The analysis gave important insights into data quality and the reliability of each gauge, which had previously not been recognized. These included identification of more/less reliable periods of record, which varied widely between gauges, and identification of more/less reliable parts of the hydrograph. Most gauges showed significant deviations at low stages, affecting the determination of low flows. This was independent of the type of gauge control, with many gauges experiencing problems in the stability of the rating curve, likely as a result of sediment flux. The deviations in gaugings also have widespread application in modelling, for example, informing suitable calibration periods and defining error distributions. This paper demonstrates the value and importance of undertaking qualitative analyses of observed records.