Files in this item



application/pdfZHAO-DISSERTATION-2017.pdf (5MB)
(no description provided)PDF


Title:A service-driven approach to assist water management during extreme events
Author(s):Zhao, Tingting
Director of Research:Minsker, Barbara
Doctoral Committee Chair(s):Minsker, Barbara
Doctoral Committee Member(s):Cai, Ximing; Srikant, Rayadurgam; Uster, Halit
Department / Program:Civil & Environmental Eng
Discipline:Civil Engineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
Abstract:Water shortages and flooding have caused large property losses and endangered human lives in many areas. Rapid and informed response is needed to ensure effective water management, including reliable and immediate data synthesis, near-real-time forecasting, and model-based decision support for water operations. A structure to rapidly process heterogeneous information and models needed for near-real-time water management is critical for decision makers. This dissertation develops a service-driven approach to decision support in water management, focusing on case studies related to drought and flooding. For flood management, real-time reservoir management is a critical component of decision support. Estimating and predicting reservoir inflows is particularly essential for water managers, given that flood conditions change rapidly. We propose a data-driven framework for real-time reservoir inflow prediction, using a service-oriented approach, that enables ease of access through a Web browser. We have tested the services using a case study of the Texas flooding events in the Lower Colorado River Basin in November 2014 and May 2015, which involved a sudden switch from drought to flooding. We have constructed two prediction models: a statistical model for flow prediction and a hybrid statistical and physics-based model that estimates errors in the flow predictions from a physics-based model. The performances of these two models are compared for short-term prediction. In addition, both the statistical and hybrid models have been published as Web services through Microsoft’s Azure Machine Learning (AzureML) service, and are accessible through a browser-based Web application. The study demonstrates that the statistical flow prediction model can be automated and provides acceptably accurate short-term forecasts. However, for longer-term prediction (2 hours or more), the hybrid model fits the observations more closely than the purely statistical or physics-based prediction models alone. The second case study focuses on droughts, developing methods to better manage significant imbalances between water supply and demand. A service-driven approach is used to couple river modeling services with optimization services for determining optimal water allocation strategies under daily drought scenarios in a permit system. An accurate and computationally efficient meta-model approach is then developed to relieve the computational burden of the simulation-optimization model. This work uses a drought event in the Upper Guadalupe River Basin, Texas, in April 2015 as a case study to illustrate the benefits of the approach. Weather and water demand uncertainty are considered through scenario-based optimization. The results have demonstrated that the simulation-optimization model services can easily be coupled using DataWolf workflow tool and AzureML service, providing improved water allocation strategies relative to the current approach. The scenario analysis shows that the permit grouping system, which organizes water right permit holders into groups rather than considers each water user individually, is an easy and manageable approach for water allocation. In addition, the adaptive meta-model approach is efficient to relieve the computational burden in simulation-optimization model, thereby enabling large-scale real-time Web services for decision support.
Issue Date:2017-06-21
Rights Information:Copyright 2017 Tingting Zhao
Date Available in IDEALS:2017-09-29
Date Deposited:2017-08

This item appears in the following Collection(s)

Item Statistics