Adapting hydrological modeling for Atlantic Canada's climate, landscape, and vegetation conditions: from field to small and large watersheds

Abstract

Pollution from nonpoint sources poses a significant threat to aquatic ecosystems. Best management practices (BMPs) can be developed to control soil erosion and reduce the movement of nutrients and sediments from agricultural lands to streams. Direct assessments of the impact of land use change and BMPs on water quantity and quality through field experiments are time-consuming and costly and, as a result, model simulations of hydrological processes and BMPs impacts can serve as a complementary approach to field measurements. However, model simulations require detailed inputs and complex calibration procedures, which may delay their acceptance among decision makers. Central to this thesis is development of a simple decision-support tool for decision makers and economists to evaluate multi-year impacts of land use change and BMPs on water quantity and quality for large ungauged watersheds. The ArcGIS-based tool (i.e., the land use and BMPs assessment tool, LBAT) uses statistical models derived from simulations generated with the Soil and Water Assessment Tool (SWAT). To provide reliable simulations for Atlantic Canada, SWAT was modified to address maritime-winter climate conditions of high snow accumulation. New physically-based soil-temperature and snowmelt modules were developed and incorporated in SWAT to account for snow-insulation effects and rain-on-snow events on the seasonal evolution of soil temperature. It was hypothesized that modification of SWAT would provide superior predictions of water flow and nutrient loadings for Atlantic Canada. With appropriate calibration, the modified version of SWAT was validated against field data collected from a small experimental watershed in northwest New Brunswick, Canada, i.e., the Black Brook watershed (BBW). Once finalized, LBAT and SWAT were applied to a large watershed consisting of the BBW (i.e., Little River watershed). Results suggested that LBAT and a calibrated version of SWAT performed equally well in simulating annual stream flow and sediment and nitrate-N loadings, with LBAT performing slightly better for annual soluble-P loading. In addition, LBAT performed much better than an uncalibrated version of SWAT for sediment and nutrient loadings. The LBAT has a unique role in ungauged watershed management in New Brunswick for its simplicity and flexibility compared with process-based hydrological models. Keywords: best management practices; decision support tool; hydrology; soil and water assessment tool; soil temperature; snowmelt; water pollution