Author : Philip Halteman
Publisher :
Page : 420 pages
File Size : 11,94 MB
Release : 2015
Category : Bayesian statistical decision theory
ISBN :
The effective management of non-point source nutrient pollution continues to prove elusive. Though the scientific literature is unequivocal that all anthropogenic land uses contribute to non-point source (NPS) pollution, variable levels of contribution over time and across location and complex relationships between cost and effect make finding technologically effective management solutions difficult. In addition, these solutions are implemented in a world of scarce resources, diverse and often competing concerns and values, and intense public scrutiny. Clearly, making the best possible decision about how to manage NPS pollution under these conditions is not simple. My overarching goal was to develop and test several practical approaches that provide insight into the implications of management decisions and the trade-offs facing water quality managers using the challenges of restoring Lake Champlain as a test case. I first demonstrate a simple spreadsheet-based method for (1) identifying the areas of greatest potential for further phosphorus reductions, (2) estimating the potential scale of those reductions, and (3) identifying the severe tradeoffs that exist between cost and effectiveness at high levels of management. Results of this method suggest that better and more extensive management of developed impervious surfaces and annual cropland and hayland represent the greatest potential for phosphorus reductions. Farmstead management, combined sewer overflows, and wastewater treatment present little opportunity under the current regulatory environment. Results also suggest that due to order-of-magnitude differences in cost-effectiveness between management practices for developed and agricultural lands, substantial tradeoffs exist between cost-efficiency and equity in the distribution of responsibility for management. Second, in an effort to quantify the variability of NPS contributions over time and space, I developed and applied a Bayesian hierarchical modeling approach to incorporate annual hydrologic variability and uncertainty about land use areas into estimates of land-use specific phosphorus loading rates and watershed-scale residual loading. The model was able to replicate both average load and the variability around that average with an acceptable degree of precision. The results of this approach suggest that for some watersheds, unmanageable sources of phosphorus are dominant. Third, I applied a Bayes network to predict the effects of alternative management scenarios on phosphorus loads. Using evolutionary optimization and a multiple-criteria decision analysis, I explored the tradeoffs between cost, effectiveness, and distributional equity in the burden of management. Results of this study indicate that the probability that phosphorus loads will comply with regulatory targets is, in some watersheds, small under any management scenario. More interestingly, it also appears that there are large differences between watersheds in the ability of management actions to raise those probabilities, and the significant and non-linear tradeoffs between cost, effectiveness, and equity will make decision-making - and achieving restoration targets - difficult. Together, these approaches provide a foundation for a fuller and more completely informed decision-making process that incorporates uncertainty and identifies key trade-offs for the State of Vermont as it implements a new management plan for Lake Champlain.