Denitrification Modeling Workshop, November 2006
Summary of Soil Water Assessment Tool (SWAT)
SWAT is a process based hydrology and water quality model, designed to estimate impacts
of land management practices on water quantity and quality in complex watersheds. The model
runs on a daily time step, and designed to utilize readily available input data. The model
provides the user with several options for calculating components of a water balance (e.g., runoff
curve number, Green and Ampt, equation, Penman Montieth, Priestly-Taylor, etc.). Complex
watersheds are divided into sub-basis which have relatively uniform land cover and soil
properties. Each sub-basin is represented by a one dimensional soil profile, which is connected
to down stream profiles by overland flow channels.
Denitrification losses in each soil layer is estimated based on first order decay kinetics, with
the rate coefficient based on temperature and total organic carbon content in a soil layer, if the
soil moisture content is greater than 95% of soil porosity. When the soil moisture content is less
than 95% of porosity, denitrification is estimated to be zero. In our simulations in Illinois, we
used the denitrification rate coefficient as one of several calibration parameters.
SWAT has been applied to a wide range of watersheds throughout the world, with sizes
ranging from a few hectares to the continental US. Simulations are generally between 5 and 20
years in length. The model is primarily used to address questions related to the impacts of
agricultural management practices on water quantity and quality. Emphasis is on comparisons of
land management effects over multiple years rather than during individual events.
The empirical basis of the denitrification equation and parameters is not specified in SWAT
documentation, but Pohlert et al. (2006) indicate that SWAT employs the approach used by the
EPIC model. Our SWAT simulations produced estimates of denitrification that were within the
realm of plausibility. However, Pohlert et al. (2006) recently reported that the model produced
unrealistically large estimates of denitrification in a German watershed, which prompted them to
incorporate DNDC routines into SWAT and to call the new model SWAT-N.
SWAT code and documentation are readily accessible <http://www.brc.tamus.edu/swat>.
Development is supported by USDA ARS. There are a large number of developers and users.
Training workshops are offered periodically, and there have been several international SWAT
Conferences convened since 2000.
Our simulations of the Embarras River watershed in Illinois indicated that after calibration
SWAT accounted for 51% of the variation in observed monthly nitrate flux in drainage waters.
The average annual denitrification was estimated to be 22 kg N ha-1 yr-1, but we did not have
direct measurements of in-field denitrification for comparison. The greatest deviations between
observed and simulated nitrate flux in drainage waters were associated with unusually large
rainfall events in the spring. The model overestimated N fixation in soybeans by a factor of
approximately 1.8. Simulated reductions in fertilizer input resulted in reductions in
denitrification and nitrate flux in drainage waters, but increases in N fixation maintained a
positive soil N balance, which is not consistent with empirical observations.
Arnold, J. G. and N. Fohrer. 2005. SWAT2000: current capabilities and research opportunities
in applied watershed modelling, Hydrol. Process. 19: 563–572, doi:10.1002/hyp.5611.
Hu, X., G. F. McIsaac, M. B. David, and C. B. Louwers. 2006. Modeling Riverine Nitrate
Export from East-Central Illinois Using SWAT. In Revision.
Pohlert, T., L. Breuer, J. A. Huisman, and H.-G. Frede. 2006. Assessing the model performance
of an integrated hydrological and biogeochemical model for discharge and nitrate load
predictions Hydrol. Earth Syst. Sci. Discuss. 3: 2813–2851.