Data Uncertainty Estimation Tool for Hydrology and Water Quality (DUET-H/WQ): Estimating Measurement Uncertainty for Monitoring and Modelling Applications

In spite of advanced modelling capabilities, hydrologic and water quality data remain vital for scientific assessment, management, decision-making, and modelling. Although uncertainty in measured data affects all of these applications, measurement uncertainty is typically ignored in monitoring projects. To change this, we published an uncertainty estimation framework for measured discharge and water quality data in 2006. From this framework, the Data Uncertainty Estimation Tool for Hydrology and Water Quality (DUET-H/WQ) was designed as a user-friendly tool to facilitate uncertainty estimation. DUET-H/WQ provides published uncertainty estimates for data collection procedures and then estimates the uncertainty within each procedural category as well as the cumulative uncertainty. The software estimates uncertainty for individual measured values as contributed by measurement and data processing and management. It does not account for uncertainties associated with spatial variability or influences of scale. The broad applicability of DUET-H/WQ was established by its application to data collected in five monitoring projects from a variety of watershed conditions. Results indicated that uncertainty in individual values was typically least for discharge, higher for sediment and dissolved N and P, and higher yet for total N and P. The uncertainty inherent in measured data has numerous economic, societal, and environmental implications; therefore, scientists can no longer ignore measurement uncertainty in data collection and reporting. It is our hope that DUET-H/WQ will contribute to making uncertainty estimation a routine component in hydrologic and water quality monitoring projects. Measurement uncertainty also has important implications in modelling applications. The impact of uncertainty in model calibration and validation data is commonly discussed, but rarely included, in the evaluation of model accuracy. In order to change this oversight, we recently modified several goodness-of-fit indicators to incorporate measurement uncertainty into model calibration and validation. A similar method is currently being tested that incorporates both measurement and model uncertainty into model goodness-offit evaluation.