|Talebizadeh, Mansour - Orise Fellow|
|Starks, Patrick - Pat|
Submitted to: Journal of the American Water Resources Association
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/19/2018
Publication Date: 9/27/2018
Citation: Talebizadeh, M., Moriasi, D.N., Steiner, J.L., Gowda, P.H., Tadesse, H.K., Nelson, A.M., Starks, P.J. 2018. APEXSENSUN: An open-source package in R for sensitivity analysis and model performance evaluation of APEX. Journal of the American Water Resources Association. p. 1-15. https://doi.org/10.1111/1752-1688.12686.
DOI: https://doi.org/10.1111/1752-1688.12686 Interpretive Summary: Hydrologic and water quality models are increasingly used to evaluate the impacts of climate, land use, and land management practices on quantity and quality of land and water resources. A new software packaged called APEXSENSUN was developed to perform sensitivity and uncertainty analysis. The software was applied to Rock Creek watershed, a small agricultural watershed located in northern Ohio, to detect most sensitive parameters that affect APEX’s performance regarding streamflow and sediment yield. In the end, the APEXSENSUN was successfully used for extracting simulations that met a set of predefined criteria and then used for constructing uncertainty bands for streamflow and sediment load predictions. Overall, the APEXSENSUN proved to be a very flexible and easy-to-use software for performing sensitivity analysis and model calibration. This computer program will contribute to the goal of USDA to parameterize and validate APEX to support nation-wide deployment of the nutrient tracking tool and will also contribute to other APEX modeling studies.
Technical Abstract: The Agricultural Policy/Environmental eXtender (APEX) model is used to evaluate the impact of different land management strategies with respect to water availability, soil and water quality, plant growth, and economics. In this study, a flexible software package named APEXSENSUN was developed in R to streamline the sensitivity analysis and calibration of the APEX model. The Rock Creek watershed in northern Ohio was used as the study area and parameter sensitivities were calculated for streamflow and sediment load predictions. Comparing the static and dynamic sensitivity results indicated that there could be a significant variation between the number and ranking of APEX parameters depending on input conditions. The simulations used for sensitivity analysis were used by APEXSENSUN for identifying those simulations that meet certain performance criteria. The parameter sets for accepted simulations were then used for constructing an empirical uncertainty band which provides a better way of communicating uncertainty with model predictions.