Abstract
Many hydrologic studies that are the basis for water resources planning and management rely on streamflow information. Calibration and use of hydrologic models to extend flow series based on rainfall data, perform flood frequency analysis, or develop flood maps for land use planning and design of engineering works, such as channels, dams, bridges, and water intake, are examples of such studies. In most real-world engineering applications, errors in flow data are neglected or not adequately addressed. However, because flows are estimated based on the water level measurements by fitted rating curves, they can be subjected to significant uncertainties. How large these uncertainties are and how they can impact the results of such studies is a topic of interest for researchers, practitioners, and decision-makers of water resources. The quantitative assessment of these uncertainties is important to obtain a more realistic description of many water resources related studies. River restoration in many areas is limited by data availability and funding. A means to assess the uncertainty of flow data to be used in the design and analysis of river restoration projects that is cost effective and has minimal data requirements would greatly improve the reliability of river restoration design. This paper proposes an assessment of how uncertainties related to rating curves and frequency analysis may affect the results of flood mapping in a real-world application to a small watershed with limited data. A Bayesian approach was performed to obtain the posterior distributions for the model parameters and the HEC-RAS (Hydrologic Engineering Center-River Analysis System) hydraulic model was used to propagate the uncertainties in the water surface elevation profiles. The analysis was conducted using freely available data and open source software, greatly reducing traditional analysis costs. The results demonstrate that for the study case the uncertainty related to the frequency analysis study impacted the water profiles more significantly than the uncertainty associated with the rating curve.