Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.

Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.

Document Type

Open Access Dissertation

Degree Name

Doctor of Philosophy (PhD)

Degree Program

Civil Engineering

Year Degree Awarded

2016

Month Degree Awarded

September

First Advisor

Mi-Hyun Park

Second Advisor

John Tobiason

Third Advisor

Nicholas Reich

Subject Categories

Environmental Engineering

Abstract

This research was principally concerned with the task of quantifying dissolved and suspended constituents carried in river water when direct measurements are not available. This is a question of scientific and societal relevance, and one with a long history of study and a great deal of remaining difficulty. The traditional approach to estimating these quantities, linear regression models (LMs), suffers from poor flexibility and high subsequent bias in many applications. This research applied semiparametric generalized additive models (GAMs), a more flexible class of regression models, evaluated their performance in various locations and conditions, and applied them in a proactive modeling effort in a major water-supply reservoir. Chapter 1 compared GAMs to LMs for estimating nutrient and organic carbon loads in three major tributaries of the Wachusett Reservoir in central Massachusetts. The relative performance of each model was determined using cross-validation. GAMs outperformed LMs in most cases, explaining an additional 2% of load variance and 5% of concentration variance in validation data on average. Relative differences between the two modeling approaches exceeded 100% depending on the time interval of the load estimate. Chapter 2 assessed the applicability of GAMs to the prediction of riverine solute concentrations during extreme high-flow events when such events are absent from the models' calibration data. The models tended to overpredict extreme-event concentrations, with increasing bias and variance for increasingly extreme hydrologic conditions. Despite an overall increase in uncertainty for extreme-event concentration estimates, estimates under extreme hydrologic conditions could be improved by taking into account the observed bias in the aggregated regional database. Chapter 3 developed and applied a methodology to generate reservoir tributary discharge and constituent concentration time-series for an imposed extreme-event scenario. A multivariate probability model was developed for constituent concentration in an arbitrary number of tributaries and water-quality constituents, conditional on time and hydrologic condition. Two separate historical storm events were modified using 3 extreme precipitation depths on tributaries of the Wachusett Reservoir Watershed in Massachusetts, U.S. Quasi-Monte Carlo was used to propagate this uncertainty to a process-based model of the receiving water body.

Share

COinS