Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.
Hydrologic and Water Quality Models: Sensitivity
Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org
Citation: Transactions of the ASABE. 58(6): 1721-1744. (doi: 10.13031/trans.58.10611) @2015
Authors: Yongping Yuan, Yogesh Khare, Xiuying Wang, Prem B. Parajuli, Isaya Kisekka, Stefan Finsterle
Keywords: Hydrologic and water quality modeling, One-at-a-time, Parameters, Sensitivity analysis, Sensitivity measures, Sensitivity methods.
Abstract. This article is part of a collection of articles that provides a comprehensive description of hydrologic and water quality (H/WQ) model calibration and validation concepts and processes. Sensitivity analysis (SA), which is often used to quantify the strength of relationships between model inputs and outputs, is an essential evaluation of any kind of modeling. SA is crucial in H/WQ models due to various aspects involved in H/WQ modeling processes, such as empiricism, spatiotemporal scales, and complexity, that require an assessment of parameters‘ influence on the model‘s prediction. This study synthesized SA applications for 25 H/WQ models in the special collection on model use, calibration, and validation published in 2012 and provides guidance on their future applications. Commonly used SA methods are summarized along with tools to implement them. While SA was not employed for all 25 models in the special collection, a wide range of SA methods (from partial derivatives to variance-based global methods) and sensitivity measures (from scatter plots to variance decomposition measures) were used in the literature. Some model parameters were found to be important in most sensitivity applications performed for the models; however, their relative importance varied from study to study, underscoring the necessity of SA for every new model application. Nevertheless, summarizing important model parameters can still serve as a starting point for model users. Since most studies concentrated on model parameters alone; future SA applications in H/WQ modeling should also consider other inputs (climate data, boundary conditions, etc.) and non-parametric aspects, such as features and processes considered in the model.
(Download PDF) (Export to EndNotes)