Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Environmental Microbial & Food Safety Laboratory » Research » Publications at this Location » Publication #289101

Title: Modeling runoff and microbial overland transport with KINEROS2/STWIR model: Accuracy and uncertainty as affected by source of infiltration parameters

item GUBER, ANDREY - Michigan State University
item Pachepsky, Yakov
item YAKIREVICH, ALEXANDER - Ben Gurion University Of Negev
item Shelton, Daniel
item WHELAN, GENE - Us Environmental Protection Agency (EPA)
item Goodrich, David - Dave
item Unkrich, Carl

Submitted to: Journal of Hydrology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/3/2014
Publication Date: 12/7/2014
Citation: Guber, A., Pachepsky, Y.A., Yakirevich, A., Shelton, D.R., Whelan, G., Goodrich, D.C., Unkrich, C.L. 2014. Modeling runoff and microbial overland transport with KINEROS2/STWIR model: Accuracy and uncertainty as affected by source of infiltration parameters. Journal of Hydrology: 519.644-655.

Interpretive Summary: Runoff of microorganisms from field-applied manure and animal waste is dependent on how rainfall or irrigation water partition between overland flow and infiltration into soil. Infiltration depends on both the ability of soil to transmit water and to retain water. The effect of the latter factor on microbial transport from pasture/fields has attracted surprisingly little attention. We compared three methods to evaluate this effect and found that soil survey data can provide a good estimate of the water retention-related infiltration parameter when used with an ensemble of predictive relationships rather then a single one. These results are useful for predicting microbial fate and transport of microbes in land-applied or deposited animal waste. It also demonstrates the feasibility of using soil survey data in providing key modeling parameters.

Technical Abstract: Overland transport of microorganisms from fields and pastures causes concerns for the quality of surface water used for irrigation, recreation, and other activities. This transport to large extent is controlled by the ability of soils to conduct infiltration water. Compared with soil saturated hydraulic conductivity, there has been relatively little research on how the shape of the infiltration front, that depends on soil water retention, may affect the transport of dissolved and suspended material in runoff. Our objectives were to evaluate the uncertainty of the net capillary drive associated with (i) sources and methods of the parameter estimation and (ii) spatial averaging of soil properties used for parameter estimation; and (iii) to examine how this uncertainty translates into uncertainty of the runoff and microbial overland transport predictions. The overland flow and transport model KINEROS2/STWIR was used with experimental data collected in runoff experiments at the manured field at the OPE3 USDA-ARS research site. Three capillary drive estimation methods were compared, namely: fit to the measured cumulative runoff hydrograph, estimation based on soil texture class using KINEROS2 manual, and estimation from basic soil properties using pedotransfer functions (PTFs). Prediction with individual inputs first then averaging, and input averaging then prediction was used with PTFs in multimodel simulations to generate capillary drive values. Also the ensemble modeling was applied when runoff and fecal coliform predictions were obtained with individual PTFs and then were averaged. The overall model accuracy and uncertainty was affected by the accuracy and uncertainty of the parameter G estimation. The KINEROS2-estimated parameter yielded slightly less accurate prediction of runoff, FC concentrations and total FC, than calibrated model. The spatial variability of soil properties used as PTF inputs resulted in highly variable estimates. In general, the parameter variability was smaller for the ensemble of PTFs compared to the multimodel approach. The smaller uncertainty in capillary drive estimations produced by the ensemble method translated into much more certain KINEROS2/STWIR predictions of cumulative runoff and total FC in runoff water. The uncertainty of predicting the FC concentration was much smaller compared with the runoff and total FC predictions, and multimodeling could provide robust prediction results when FC concentration in runoff water is a concern. Overall, ensemble PTF-based modeling of the capillary drive estimation can be efficient for runoff and the bacteria overland transport simulations when a single effective value used across the study area