Glenn S. Warner

Application of a steady-state nutrient model and inferences for load reduction strategy in two public water supply reservoirs in eastern Connecticut

Abstract Mansfield Hollow Lake (MHL) and Willimantic Reservoir (WR) are two reservoir lakes located in eastern Connecticut in the northeastern United States. MHL formed behind the Mansfield Hollow Dam constructed by the U.S. Army Corps of Engineers in 1952 and is primarily fed by the Fenton, Mount Hope and Natchaug Rivers. The WR lies approximately 1-km downstream from the Mansfield Hollow Dam. Total dissolved nitrogen, phosphorus and chlorophyll a measurements indicate the water bodies could be classified as borderline mesotrophic/eutrophic. A steady-state numerical software package (Bathtub) designed to facilitate application of empirical eutrophication models to morphometrically complex reservoirs was used to determine the trophic status in MHL and WR based on different phosphorus and nitrogen loading budgets. The short hydraulic residence times and rapid flushing rates in MHL and WR are directly related to the flow rates in the streams discharging into MHL. The low flow period could significantly increase the hydraulic residence times of these two reservoirs. Therefore, the sampling design emphasized periods of low flow in late August and early September to assess the impact of nutrient inputs to MHL and WR during dry periods. The results of a low flow sampling period (August 2002) were used to calibrate and test the Bathtub model developed for these water bodies. Application of the Bathtub model to differing flow regimes, notably average flows, suggested that nitrogen or phosphorus could limit the productivity and cause eutrophication in the two lakes. Results of this study indicated that the Bathtub model could be used to predict total nitrogen and total phosphorus concentrations with reasonable accuracy, but it might not be a suitable tool for predicting organic nitrogen or algae in rapidly flushing lake systems. To further investigate and validate the assumptions made in this study, more sampling data are needed, especially during high intensity storm events to investigate possible sources of nutrient flow into the two lake system and further calibrate the Bathtub model for the MHL-WR watershed.