Time and Frequency Domain Analyses of Hydrologic and Stream Water Quality Data
Date of Award
Doctor of Philosophy
Randall W. Gentry
John S. Schwartz, Edmund Perfect, Patrick J. Mulholland
The issue of scaling of deterministic properties at various domain levels has become an area of interest for active research. Interpretation of observational data, and similarly hydrologic simulation, require an understanding of how certain deterministic modeling parameters scale from the smaller to larger domains spatially and temporally. Novel approaches to better understand scaling behavior in hydrologic systems and that will result in improved deterministic modeling are necessary to advance the state-of-the-science in hydrologic transport processes, particularly in the areas of microbial concentration and inorganic chemical constituent dynamics. Two primary objectives of this research were i) to determine the statistical relationship of microbial concentrations, and other inorganic chemical water quality parameters (nitrate, chloride, sulfate and calcium) to hydrologic variables (i.e. precipitation, discharge) and ii) to observe the presence of persistence (long-term and short-term if any) and the strength of the persistence for better defining the behavior of watershed scale responses of those parameters. Time and frequency domain analyses were performed to observe the short-term and long-term persistence of the hydrologic variables and the parameters’ time series data from two watersheds in East Tennessee. Discharge and other water quality parameters showed short-term as well as long-term persistence ranging from about two weeks to approximately a year. Hurst analysis was done for the time series data to find the strength of the persistence (in terms of Hurst coefficient) both for the untransformed as well as deseasonalized data using spectral and rescaled range analysis. All parameters, except rainfall, showed some level of persistence ranging from 0.62 to 0.96. The Hurst coefficient could be used to generate fractional Brownian motion or fractional Gaussian noise, which helps to develop the forward predication model in a stochastic way for the time series data. It can also be used to design and evaluate the sampling frequency for the hydrologic and water quality data. Time and frequency domain approaches used in this research are particularly important in evaluating watershed-scale water quality changes or microbial persistence, which will be helpful to develop the improved watershed management strategies.
Koirala, Shesh Raj, "Time and Frequency Domain Analyses of Hydrologic and Stream Water Quality Data. " PhD diss., University of Tennessee, 2007.