Rashid, I. O. (2014). Evaluation of SWAT Model Applicability in a First-Order Agricultural Watershed in Southern Ontario. Geography & Environmental Studies Theses and Dissertations. 1. https://scholars.wlu.ca/geog_etd/1

The primary objective of most agricultural operations including those at the Strawberry Creek Watershed is to maximize crop yield. This often requires the input of additional nutrients to enhance soil fertility, and installation of artificial subsurface drainage to curtail water-logging of soils with very low hydraulic conductivity. However, the combination of additional nutrients and subsurface drainage often leads to a disruption of the existing hydrochemical regime. Excess nitrate and phosphorus export to surface water bodies may yield negative environmental impacts, including the eutrophication of downstream areas. To address this issue, a modeling technique was deployed to quantify and assess these processes under various seasonal scenarios.

The Soil and Water Assessment Tool (SWAT) model was modified (SWATtile) and parameters defined to simulate the effects of tile drainage on flow and nitrate (NO3⁻) export from small watersheds during the four seasons characteristic of southern Ontario. This study compares differences and similarities between observed watershed processes against model output by: (1) utilizing the SWATtile model for comparison of simulated to measured discharge from a watershed, and a tiled field, from several years of data, (2) utilizing the SWATtile model for comparison of simulated to measured NO3⁻ from a watershed and tiled field, and (3) several scenarios are presented on how modifications to tile spacing (density) can be manipulated to achieve a balance between improving soil drainage while minimizing NO3⁻ export. The effects of tile density changes were evaluated to determine the impact of moisture availability (for tile flow) as precipitation cycled from rain to snow and back to rain.

In the first part of this study, comparison of detailed simulations of seasonal flow patterns from both the gauged watershed and a gauged tiled field for winter 2007 to winter 2008, reveals similarities and contrasts in flow patterns for daily time scales. Due to its distributed nature, the SWAT model is subdivided into fundamental units of analysis designated as Hydrologic Response Units (HRUs). Each HRU consists of a unique soil and landuse type and is capable of autonomous analysis and result generation. The gauged subwatershed area drained by the Below Middle Road (BMR) tile has been continuously monitored for more than six years. This subwatershed was defined in the SWAT model setup as an independent HRU so that results generated from simulations can be directly compared to measured values from the same area. In terms of landuse, soil type and tile spacing, the BMR – HRU is representative of other tiled fields in the watershed. Simulated stream and tile flow for each season were comparable to that of the observed. Linear trends between measured main channel flow and that of measured tile flow was stistically significant. However, trend agreement between simulated main channel discharge and BMR tile was not statistically significant, although it demonstrated a general linear pattern.

For the second part of the study, comparison of observed/measured watershed NO3⁻ concentration against results generated by SWATtile model were quantified across all seasons with the contrast being greatest for the spring season. The general trend in modeled NO3⁻ is for more of it to be exported during low flows. NO3⁻ then increases with volume of flow. The tile outlet yields a higher NO3⁻ load per unit area, as this contribution originates from a much smaller area (0.43 km2) compared to main stream outlet with contribution from the entire watershed which is a much large area (2.86 km2). For both the first and second scenarios, the tile drainage component was also disabled to enable observation of dominance of overland flow as a result of an elevated water table. Consequently, there was an observable reduction in crop NO3⁻ uptake which is largely due to an increased rate of denitrification under anaerobic conditions.

The third part of the study introduces variability in density between feeder tiles and thus altered the drainage intensity. The drainage intensity is the rate at which water is removed from a field and is thus proportional to tile density. As the intensity is increased, drainage and NO3⁻ export also increases proportionally. On the other hand, as the lateral distance is increased above 50 ft. (15.24 m), tile drainage and NO3⁻ export from the field are reduced. Crop NO3⁻ uptake was also reduced (decreased productivity) with an increase in tile density (from 50ft. to 35ft.). This was also characterised by increased NO3⁻ export. The anoxic conditions might also favour denitrification which may lead to further NO3⁻ loss. For the watershed simulation, although decreasing tile density helped reduce NO3⁻ mass export (density reduced from 50ft [90 kg/ha] to 65ft [82 kg/ha]), it was still not enough to attain the required drinking water standard of 10 mg/L (and the limit of 12.8 mg/L for aquatic species). However, when the tile density was reduced to 85ft. (30m), the concentration of NO3⁻ decreased to 25 kg/ha. The farmers will need to reduce fertilizer usage by an average of 25 kg/ha to attain the drinking water standard. In reality, such low tile density is rare as installation may not substantially enhance drainage and therefore not be beneficial to crop production. Other forms of intervention, such as regulating fertilizer application and timing, employing crop rotation with crops that collectively utiize the full spectrum of nutrients, and preservation of riparian buffer strips, will be required to complement tile drain density manipulation to ensure economic productivity and pollution abatement.