Date of Award


Degree Type


Degree Name

Master of Science in Agricultural and Biological Engineering


Agricultural and Biological Engineering

First Advisor

Jane R. Frankenberger

Committee Chair

Jane R. Frankenberger

Committee Member 1

Indrajeet Chaubey

Committee Member 2

Kevin W. King

Committee Member 3

Eileen J. Kladivko


Simulation models such as the Soil and Water Assessment Tool (SWAT) have become widely used in determining the water quality impacts of various management practices. Ensuring that the algorithms accurately represent the processes simulated has become an important goal. Tile drainage is a standard practice in the Midwest, US in order to reduce risk of yield loss due to excess water. Multiple tile drainage and water table algorithms have been available in the SWAT model between the initial SWAT release and revision 638 used in this study. Testing of those algorithms is often limited. Furthermore, algorithms in the current version have not been tested using small scale measured tile discharge.

To better represent the hydrologic processes related to subsurface drainage, four modifications were made to the SWAT model subsurface hydrology routines in order to increase the physical basis of these algorithms. First, percolation through the soil profile was altered to be based on Darcy’s Law and the Buckingham-Darcy Law. Second, the restrictive layer of the soil profile was redefined to be the bottom of the soil profile and an additional variable was added to control the seepage through the restrictive layer. Third, the water table height algorithm, which was based on an algorithm applicable at only one site, was redefined to be within the lowest unsaturated layer. Lastly, the lag through the tile drains, which caused an unrealistic delay under default conditions was removed and flow is delayed by only the drainage coefficient.

These changes were evaluated at the experimental tile drained field at the Southeast Purdue Agricultural Center (SEPAC). The model was developed with a single hydrologic response unit (HRU) and calibrated for both tile flow and nitrate. The modifications improved the performance of SWAT for water table and tile flow predictions, although the nitrate was more severely under-predicted.

The modifications were tested on a small watershed located in Central Ohio monitored by the USDA-ARS. Each tile output in this watershed was monitored allowing for each tile to be individually modeled and analyzed with SWAT. This watershed was also calibrated for tile flow and nitrate. Here again, the modifications showed an improvement for tile flow but a reduction in performance for nitrate. Phosphorus was also looked at but not calibrated for, and an extreme under-prediction issue was observed.

These modifications improved the physical basis or simplified the process representation in the SWAT model, and showed improvement to the tile flow model predictions. The model should be further tested and further developments, specifically for nitrogen and phosphorus, should continue.