Environmental and Water Resources Decision-Making Under Uncertainty

mathematical programming;optimization;sampling;monitoring;decision analysis

Harrison, Kenneth Watson ; H. Christopher Frey, Committee Member,E. Downey Brill, Committee Member,Kenneth Reckhow, Committee Member,Morton A. Barlaz, Committee Member,S. Ranji Ranjithan, Committee Chair,Harrison, Kenneth Watson ; H. Christopher Frey ; Committee Member ; E. Downey Brill ; Committee Member ; Kenneth Reckhow ; Committee Member ; Morton A. Barlaz ; Committee Member ; S. Ranji Ranjithan ; Committee Chair

University:North Carolina State University

关键词: mathematical programming;    optimization;    sampling;    monitoring;    decision analysis;   

Others  :  https://repository.lib.ncsu.edu/bitstream/handle/1840.16/5564/etd.pdf?sequence=1&isAllowed=y

美国|英语

来源: null

【 摘 要 】

"Decision-making under uncertainty" is an important area of study in numerous disciplines.The variety of quantitative methods that have been proposed to address environmental and water resources problems reflects the importance of this subject.In a review of the literature, methods were compared and contrasted and promising areas for future research were identified.Conclusions drawn from the review were that 1) large gains may be realized from cross-disciplinary research, 2) significant benefits may be realized from considering uncertainty, 3) advanced algorithms—probabilistic search methods and efficient methods for Bayesian analysis—and increased computing power should greatly extend the applicability of existing methods, and 4) in particular, decision-theoretic methods that have wide application for sequential decision-making.A new decision-theoretic method, Bayesian programming (BP), was developed that takes advantage of the increased computing power and improvements in Bayesian analysis methods.The method has wide applicability, suitable for problems in which there is 1) uncertainty in the modeling, 2) stochastic behavior in the systems that are modeled, 3) the possibility to reduce uncertainty through data collection, and 4) the opportunity for a recourse decision after a period of data collection.The approach combines systematic search methods (mathematical programming) and Bayesian statistical analysis techniques (Markov chain Monte Carlo) in a decision analysis framework. The BP method is tested with application to a hypothetical, but realistic river basin management problem, using real data from the much-studied Athabasca River in Alberta, Canada.The management problem involves balancing the objectives of pulp mill development and water quality protection (dissolved oxygen).Results from application of the BP method were compared with those applying other methodologies.Examination of the results indicated that the BP method is a practical method worthy of additional research.Ultimately, it is hoped, this research will lead to computer-based tools that will improve environmental and water resources decision-making.