Modeling optimal design parameters of constructed wetlands receiving agricultural runoff

Abstract

Constructed wetlands intercepting surface runoff from agricultural fields have been shown to act as a nutrient buffer, providing water quality improvements in the runoff water to be deposited safely downstream. Three surface flow wetlands were constructed in the Deer Creek-Clinton Lake sub watershed of the Upper Wakarusa Watershed, in Douglas County, Kansas, as a part of a pilot project to retain and improve runoff water quality from terraced fields with tile outlet runoff. A monitoring study conducted from 2014-15 showed that the wetlands were acting as efficient solids-removal systems. Nitrogen removal results, however, were on the lower end of the performance spectrum with high variation. This study was carried out to model the hydraulic, hydrological and water treatment parameters of the wetlands and to better understand their performance capacity.A custom curve number prediction model incorporating antecedent precipitation and vegetation patterns was developed to estimate the runoff produced from the terraced agricultural watersheds. The standard SCS unit hydrograph method was modified to match the fast-moving tile drained surface runoff. The model was adapted to account for changes in vegetation patterns onsite, which were observed to have a significant impact on the runoff volume. A dynamic wetland water budget model, also developed as part of this work, showed the variability in the performance of the wetlands between dry and wet weather conditions. During wet weather periods of high flows (early May to mid-June), the wetlands acted as flow-through systems with water levels at or above the weir outflow height. Contrarily, during dry weather periods, the water levels fall, and the outflow decreases significantly. These changes in water levels had a direct influence on the hydraulic retention times. A Monte Carlo simulation was developed to evaluate the likely distribution of retention times under different antecedent seasonal conditions. Based on the Monte Carlo analysis, the median retention time was calculated to be less than 4hrs in both wetlands. The 90% retention time was determined to be around 10 hrs, demonstrating the dramatic effect of wet and dry weather conditions. Nutrient removal calculations were performed using a relaxed tank in series model due to the tendency of constructed wetlands to behave as something in between plug flow and well mixed systems. This assessment was combined with Bayesian analysis to evaluate the likely ranges of the predicted nutrient and sediment removal from the wetlands. The calculated reaction rate values were found to be on the lower end of those observed in published literature for similar studies. Modeling results suggest that nitrogen concentrations are likely to decrease by 10-30%, compared to 30-50% for phosphorus and 65-85% for suspended solids, as runoff moves through the wetlands during and after a storm event. Future implementation of the constructed wetlands would significantly benefit from improvements in the wetland design. The presence of tile outlet terraced watersheds with saturated soil conditions demands special considerations for optimal wetland performance. These considerations include smaller than usual watershed to wetland ratio to increase retention time of the runoff water, higher aspect ratio to facilitate the removal of low velocity zones and more aggressive planting practices. Scenarios modeled for the study sites showed that the increase in removal efficiency potential would be the most evident for phosphorus and solids and less for nitrogen concentrations. A tenfold increase in reaction rate parameter was able to achieve an upwards of 50% nitrogen removal but would require significant changes in wetland design and maintenance. Similar improvements can be achieved by decreasing the watershed to wetland ratio to 5:1. Increasing the aspect ratio of the wetlands was not enough to achieve high nitrogen removal efficiencies. This study demonstrates the potential for coupled watershed-wetland models to aid in constructed wetland design. Future improvements in the modeling work could provide even deeper insights into improving wetland performance but would require extensive data collection.