Improving Watershed Models to Achieve a Better Prediction of Water Quantity and Quality

Watershed models are powerful tools for simulating different scenarios to understand the impact of management practices and are used to support and guide decision-making. However, there are often challenges and limitations to using watershed models in some areas of watershed modeling; 1) model calibration in the areas with data limitations; 2) acquiring complete weather data that accurately reflect watershed model responses; 3) accurate representation of manure operation in watershed models. This dissertation addresses each of the aforementioned challenges using new approaches and tools in three studies with the main objective of achieving a better prediction of water quality and quantity and enhancing watershed models. Chapter 2 presents a method (multi-basin calibration (MBC)) to estimate watershed model parameters that lack long-term streamflow records. In the MBC method, first, the Soil and Water Assessment Tool (SWAT) models are initialized individually for several similar neighboring watersheds with a short period of measured streamflow. Then, we aggregate the simulated and observed flows from each initialization with short histories to generate a combined observed-simulated streamflow record that is longer than the initial length of each individual member in order to increase the information content. The Nash-Sutcliffe efficiency (NSE) from this merged time series was used as the basis for calibrating using a differential evolution algorithm. To evaluate the MBC, SWAT models for three newly instrumented USGS gages in Lake Champlain Basin of Vermont, USA, were compared to the commonly used similarity-based regionalization (SBR) approach. Results demonstrate that short periods of hydrological measurement from multiple locations in a basin can represent a system similar to long-term measurements. Chapter 3 develops a method to generate a complete weather data time series with the integration of multiple Global Historical Climatology Network (GHCN) stations and to assess the benefit of much higher density, lower reliability precipitation measurements from private citizens collected by the Community Collaborative Rain, Hail, and Snow (CoCoRaHS) network data that was integrated into the GHCN. To evaluate the performance of the methodology, generated weather data is used to force the Soil and Water Assessment Tool (SWAT) models of 21 United States Department of Agriculture (USDA)-Agricultural Research Service (ARS)-Natural Resource Conservation Service (NRCS)-Conservation Effects Assessment Project (CEAP) watersheds to simulate daily streamflow. The results demonstrated that integration of multiple GHCN stations including higher-density, but perhaps lower-quality weather data can enhance model performance. A comparison with published SWAT model results further corroborated improved model performance using newly combined GHCN data. Chapter 4 develops a hybrid SWAT model, SWAT-Dairy, to accurately represent the impact of manure operation on nutrient transport. The SWAT-Dairy model incorporates process-based livestock routines, developed in the R platform, which quantify daily manure production, stored manure, daily total nitrogen (N) and phosphorus (P), organic and mineral N and P, and dynamic manure nutrient fractions based on animal characteristics, feed characteristics, and environmental conditions. Outputs are then used in SWAT to simulate the impact of livestock manure production. The new model, with simulated manure application management, is applied to a farm in the Little Otter Creek Basin in Vermont, US. Subbasin- and farm-level N and P losses from manure management using the new model were compared for different feed management scenarios.