Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Hydrology and Remote Sensing Laboratory » Research » Publications at this Location » Publication #320897

Research Project: Leveraging Remote Sensing, Land Surface Modeling and Ground-based Observations ... Variables within Heterogeneous Agricultural Landscapes

Location: Hydrology and Remote Sensing Laboratory

Title: Improving model prediction reliability through enhanced representation of wetland soil processes and constrained model auto calibration – A paired watershed study

Author
item SHARIF, AMIR - University Of Maryland
item LANG, M.W. - University Of Maryland
item McCarty, Gregory
item Sadeghi, Ali
item LEE, SANGCHUL - University Of Maryland
item YEN - Texas A&M University
item JEONG, J. - Texas A&M University

Submitted to: Journal of Hydrology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/11/2016
Publication Date: 10/1/2015
Citation: Sharif, A., Lang, M., Mccarty, G.W., Sadeghi, A.M., Lee, S., Yen, Jeong, J. 2016. Enhancing model prediction reliability through improved soil representation and constrained model auto calibration - A paired waterhsed study. Journal of Hydrology. 541:1088-1103.

Interpretive Summary: In this study, we presented modification to the Soil and Water Assessment Tool (SWAT), a watershed water quality model, for better representation of soils with various degrees of hydricity, and a new calibration technique that has the capacity to calibrate paired watersheds simultaneously on a single framework, so called “parallel calibration.” Modification was primarily aimed to expand spatial representation for denitrification, by assigning different denitrification rates for different soils within the basin, as opposed to one rate for the whole watershed. Model simulation results showed that when both proposed methodologies were applied jointly to paired sub-watersheds (Greensboro and Tuckahoe) within the larger Choptank watershed, the performance of the model improved significantly and simulations converged to realistic sums. Applying the parallel calibration scheme improved model performance, hence advocating the use of parallel calibration as a reliable technique, in paired watersheds. The modified SWAT model showed the capacity to spatially distinguish areas of high denitrification potential (hot spots), that has implications in identifying prominent areas for wetland restoration by maximizing denitrification rates and reducing nitrate loads into the watershed estuaries.

Technical Abstract: Process based and distributed watershed models possess a large number of parameters that are not directly measured in field and need to be calibrated through matching modeled in-stream fluxes with monitored data. Recently, there have been waves of concern about the reliability of this common practice of calibration, stemming from the fact that the so called “adequately calibrated models” may contain input data errors not readily identifiable by the model users. Such shortcomings, however, stem from the use of an evaluation criterion that only concerns the global in-stream responses of the model without paying attention to the intra-watershed responses. In this study, we introduced a modification to Soil and Water Assessment Tool (or SWAT) model, and a new calibration technique that collectively reduce the chance of misrepresenting intra-watershed responses. The modification was to better represent nitrate cycling in soils with various degrees of hydricity and the new calibration technique was used to increase the model capacity to calibrate paired watersheds simultaneously on a single framework. It was found that when both proposed methodologies were applied jointly to a system of paired watersheds on the eastern shores of Chesapeake Bay (namely Greensboro and Tuckahoe watersheds), the performance of the models suffered, however, the intra-watershed responses (mass of nitrate lost to denitrification) in the Greensboro and Tuckahoe models automatically converged to realistic sums. The modified SWAT model showed the capacity to spatially distinguish areas of high denitrification potential (hot spots), an ability that has implications in identifying prominent areas for wetland restoration and controlled drain structures that minimize ditch flow for maximizing denitrification.