Impact of Stage Measurement Errors on Streamflow Uncertainty. Issue 3 (23rd March 2018)

• Record Type: Journal Article
• Authors: Horner, I.; Renard, B.; Le Coz, J.; Branger, F.; McMillan, H. K.; Pierrefeu, G.
• Is Part Of: Water resources research. Volume 54:Issue 3(2018)
• Page Start: 1952
• Page End: 1976
• Publication Date: 2018-03-23
• Subjects: streamflow -- uncertainty -- stage -- error -- hydrometry -- rating curve; Hydrology -- Periodicals; 333.91
• DOI: 10.1002/2017WR022039
• Languages: English
• ISSNs: 0043-1397

Abstract

Abstract: Stage measurement errors are generally overlooked when streamflow time series are derived from uncertain rating curves. We introduce an original method for propagating stage uncertainties due to two types of stage measurement errors: (i) errors of the stage read during the gauging and (ii) systematic and nonsystematic (independent) errors of the recorded stage time series. The error models are generic and can be used for any probabilistic rating curve estimation method that provides an ensemble of rating curves. The new method is applied to a range of six contrasting hydrometric stations in France. Uncertainty budgets quantifying the contribution of various error sources to the total streamflow uncertainty are computed and compared for streamflow time series averaged at time intervals from hour to year. A sensitivity analysis is conducted on the stage time series error model to identify the most sensitive parameters. The results are site specific, which illustrates the key role played by the properties of both the hydrometric site and the gauged catchment. Across the range of sites, stage errors of the gaugings are found to have limited impact on rating curve uncertainty, at least for gaugings performed in fair conditions. Nonsystematic errors in the stage time series have a negligible effect, generally. However, systematic stage errors should not be neglected. Over the six hydrometric stations in this study, the 95% uncertainty component reflecting stage systematic errors (from ±0.5 cm to ±6.8 cm) alone ranged from 4% to 12% of daily average streamflow, and from 1% to 3% of yearly average streamflow as sensors were assumed to be recalibrated every 30 days. Perspectives for improving and validating the streamflow uncertainty estimation techniques are eventually discussed. Plain Language Summary: Water discharges in rivers are monitored by measuring the water level (known as stage) every 15 minutes or so and by converting it to discharge with a calibrated stage‐discharge model (known as rating curve). We introduce an original method for computing the uncertainty of water discharge time series due to the measurement errors of the stage records used as inputs of the stage‐discharge model. Systematic errors in stage measurements, for instance due to sensor drift or calibration errors, were found to produce non negligible discharge uncertainties. Over six hydrometric stations with varied conditions, the corresponding discharge uncertainty ranged from 4% to 12% of daily average discharge, and from 1% to 3% of yearly average discharge. Key Points: Stage errors in gaugings and in stage time series affect streamflow computation Specific error models are introduced and used to compute streamflow time series uncertainties Effect of stage errors depends on site characteristics and averaging time intervals, systematic stage errors being the most impactful