Development of Data-Driven Approaches to Forecasting Water Quality Impairment Risk
Despite many years of work, a variety of pollutants continue to reach our waters with resulting deleterious effects. On an annual basis, the economic impact of water quality impairments runs into the billions of dollars including: cost of water treatment; clean-up costs; costs to human health; and, losses in revenues from tourism and fishing. Historically, such incidences and the resulting state of affairs have been difficult to predict. Today a substantial amount of water quality data exists, much of which can be used for forecasting trends such that conditions can be predicted and adaptive management solutions implemented. Furthermore, computational capacity is now much advanced, allowing comprehensive predictions over larger scales. The overall goal of this proposed project is to determine the viability of Support Vector Machines (SVMs) for forecasting water quality trends and associated risks of water body impairments. Specifically, we will: 1) Conduct prediction experiments on selected water quality determinants using SVMs; 2) Evaluate SVM performance by comparing results with those obtained from classical time series forecasting methodologies; and, 3) Test the robustness of the SVMs based on their ability to capture seasonal variations.
July 1, 2019
Ph.D. in Agricultural, Biological, Civil/Environmental , or Industrial Engineering, or Computer Science, or a closely related discipline. Demonstrated knowledge and expertise with programming, machine learning, statistics. Publication and presentation record.
Margaret W. Gitau, Ph.D. email@example.com. Agricultural and Biological Engineering. URL: web.ics.purdue.edu/~mgitau
David R. Johnson, Ph.D. firstname.lastname@example.org. Industrial Engineering/Political Science.
1. Sekaluvu, L., L. Zhang, and M.W. Gitau. 2017. Evaluation of constraints to water quality improvements in response to conservations measures in the Western Lake Erie Basin. Journal of Environmental Management. https://doi.org/10.1016/j.jenvman.2017.09.063.
2. Guo, T., S. Mehan, M. Gitau, Q. Wang, T. Kuczek, and D. Flanagan. 2017. Impact of number of realizations on the suitability of simulated weather data for hydrologic and environmental applications. Stochastic Environmental Research and Risk Assessment. DOI 10.1007/s00477-017-1498-5.
3. Gitau, M.W., J. Chen, and Z. Ma. 2016. Water quality indices as tools for decision making and management. Water Resources Management, DOI: 10.1007/s11269-016-1311-0.
4. David R. Johnson, Jordan R. Fischbach, and David S. Ortiz (2013) Estimating Surge-Based Flood Risk with the Coastal Louisiana Risk Assessment Model. Journal of Coastal Research: Special Issue 67 - Louisiana′s 2012 Coastal Master Plan Technical Analysis: pp. 109 – 126.
5. Fischbach, Jordan R., David R. Johnson, David S. Ortiz, Benjamin P. Bryant, Matthew Hoover, and Jordan Ostwald, Coastal Louisiana Risk Assessment Model: Technical Description and 2012 Coastal Master Plan Analysis Results. Santa Monica, CA: RAND Corporation, 2012. https://www.rand.org/pubs/technical_reports/TR1259.html.