Evaluation of the effectiveness of BMPs for improving water quality in a pasture dominated watershed

Li-Chi Chiang, Purdue University

Abstract

Due to intensive farm practices, non-point source (NPS) pollution has become one of the most challenging environmental problems in agricultural and mixed land use watersheds. The NPS pollution problem can be controlled by implementing various best management practices (BMPs) in watersheds. The objectives of this research were to: (1) evaluate comprehensive BMP scenarios in a pasture dominated watershed for controlling nutrient losses at various temporal and spatial scales; (2) evaluate the impacts of weather variation, land use changes and pasture management on improving water quality; (3) hindcast the watershed responses under a best and worst suite of BMPs implemented in the watershed; (4) optimize the selection and placement of management practices to both minimize the BMP-implemented area and pollutant loadings from the pasture lands. A total of 171 BMP combinations incorporating grazing and pasture management, riparian and buffer zones, and poultry litter applications were evaluated for their effectiveness using the Soil and Water Assessment Tool (SWAT) model. The SWAT model was parameterized using detailed farm and watershed scale data. The stochasticity in weather was captured by generating 250 various possible weather realizations for a 25-year period, using measured historical climate data for the watershed. More than 43,000 runs of the SWAT model were needed to evaluate the impacts of 172 different watershed management decisions under stochastic weather conditions. Therefore, the SWAT model was run in the Condor environment, a distributed high performance computing framework. This framework significantly reduced the model run time from 2.5 years to 18 days and enabled us to perform comprehensive BMP analyses that may not have been possible with traditional model runs on a few desktop computers. The model results indicated that losses of total nitrogen (N), mineral phosphorus (P) and total P increased with an increase in litter application rates. For the same application rates, greatest losses were predicted for fall application timings compared to spring and summer applications. Overgrazing resulted in greater nutrient losses compared to baseline conditions for all application rates, timings, and litter characteristics indicating that overgrazing of pasture areas must be avoided if any improvement in the water quality is to be expected. Variability in weather conditions significantly affected BMP performance; under certain weather conditions, increases in pollutant losses can be greater than reductions due to BMPs implemented in the watershed. Buffer strips were the most important BMPs affecting the losses of total N and total P from the pasture areas. Considerable land use changes have been observed in the study area during 1992-2004 with an increase in urban lands from 3% to 12% of the watershed and a decrease in pasture lands from 48% to 37%. Besides land use changes, pasture management also changed during the study period. An increase in total nitrogen losses in the Beatty Branch subwatershed was mainly due to an increase in nutrient inputs in the pasture areas, and increases in total sediment and nitrogen losses in the Moores Creek subwatershed were mainly due to an increase in urban lands. Therefore, the individual impacts of land use changes and conservation practices should be quantified to get a true picture of the success of Conservation Effects Assessment Program (CEAP) in watersheds experiencing significant land use changes. An optimization model which combined a multiobjective genetic algorithm (NSGA-II) and a watershed model (SWAT) was used to evaluate the selection and placement of BMPs among different sets of BMP options. Two objective functions were established for the optimization process, which were both minimizing pollutant loadings and the BMP placement area. Various optimal BMP combinations derived from a multiobjective genetic algorithm (GA) optimization tool with a different set of BMP options were further compared with the solution from the targeting method, in which the pasture lands having greater pollutant losses were selected for BMP implementation. The results showed that the optimization is less effective when certain BMPs are not considered (e.g. vegetated filter strips), and it requires much longer computation time than the targeting method to search for optimal BMPs. Although the targeting method is effective for selecting and placing an optimal BMP, larger areas need to have BMPs implemented to achieve the same pollutant reductions as using the optimization tool. (Abstract shortened by UMI.)

Degree

Ph.D.

Advisors

Chaubey, Purdue University.

Subject Area

Agricultural engineering|Water Resources Management|Range management

COinS