Emerging Contaminants Occurrence and Streamflow Responses to Extreme Climate Conditions in an Agricultural Watershed

Abstract

Runoff from agricultural fields is a significant non-point source of pollution to water bodies, as it transports sediments, nutrients, pesticides, and veterinary pharmaceuticals (VA). Climate change intensifies the hydrological cycle, generating more extreme hydrometeorological and climate conditions (EHCC) that lead to floods and droughts. However, there is limited information regarding those impacts on water quality in agricultural areas. The aim of this dissertation is to evaluate the occurrence of VA and streamflow response to EHCC in the Shell Creek (SC) watershed, Nebraska. Streamflow and water quality are simulated using the Soil and Water Assessment Tool (SWAT). VA are detected using Polar Organic Chemical Integrative Samplers (POCIS) and simulated using the pesticide subroutine in SWAT. Model performance is measured by the Nash-Sutcliffe Efficiency (NSE) coefficient. Streamflow response is assessed through statistical analysis of flows and loading pollutants in periods of EHCC. VA concentrations range from 0.0003 to 68 ng/L and some display significant temporal variation. The hydrologic model reproduces monthly flows with NSE of 0.40-0.92. It overestimates lower flows by 1-2 m3/s in years with very wet and dry summer. A rain gage-forced model simulates medium-flow conditions (10-90th percentiles) closer to observations, although, it overestimates lower flows (≤ 10th) and underestimates higher flows (≥ 90th) up to 0.1 and 1 m3/s, respectively. The water quality model adequately reproduces monthly flows, sediments, and nutrients (NSE = 0.61-0.82). However, it poorly reproduces atrazine and VA (NSE ≤ 0.01). This study demonstrates the utility of POCIS for monitoring ambient levels of pharmaceuticals, and their occurrence confirms that agricultural activities influence surface water quality. Rain gages located outside SC incorporate the spatiotemporal variation of precipitation. Their integration with radar precipitation data can support hydrologic models and bring quantitative information to watershed managers. Continuous water quality monitoring is needed for an adequate implementation of hydrological models in small agricultural watersheds.

La escorrentía superficial de campos agrícolas es fuente de contaminación. El caudal y calidad del agua en “Shell Creek” se simularon utilizando el software SWAT. Los antibióticos veterinarios (VA) se detectaron utilizando muestreadores POCIS. El rendimiento del modelo se midió mediante el coeficiente “Nash-Sutcliffe” (NSE), y la respuesta del caudal a eventos hidrometeorológicos extremos (EHCC) se evaluó mediante análisis estadístico. Las concentraciones de VA oscilaron entre 0,0003 a 68 ng/L con variaciones temporales significativas. El modelo reprodujo caudales mensuales con NSE de 0,40-0,92, pero sobreestima caudales inferiores de 1 a 2 m³/s en años con veranos muy húmedo y seco. El modelo forzado por datos de pluviómetros simula condiciones de flujo medio cercanas a las observadas, aunque sobreestima flujos bajos (≤ percentil 10°) y subestima flujos altos (≥ percentil 90°) hasta 0,1 y 1 m³/s, respectivamente. El modelo de calidad del agua reprodujo adecuadamente caudales mensuales, sedimentos y nutrientes (NSE = 0,61-0,82), pero pobremente atrazina y VA (NSE ≤ 0,01). Este estudio evidencia la presencia de productos farmacéuticos en ambientes acuáticos. El monitoreo continuo de la calidad del agua es necesario para el manejo adecuado del recurso hídrico.