Date of Award

Fall 2019

Document Type

Thesis Restricted

Degree Name

Master of Science (MS)

Department

Mathematics

Committee Chairperson

Andrew Crossett, Ph.D.

Committee Member

Scott McClintock, Ph.D.

Committee Member

Timothy Lutz, Ph.D.

Abstract

I examined the memory of streamflow using two well established methods of time series analysis: detrended fluctuation analysis and cross-correlation. With data from gauges located in southeast Pennsylvania and northern Delaware, DFA measures scaling behavior by removing localized trends. Trends, while nearly unavoidable in nature, can allow for false detection of memory. Application of DFA shows streams often display long memory or non-stationary processes. Using differencing to prewhiten data, I compared how the streams interacted with each other by crosscorrelation analysis. This resulted in correlations greater than 0.15 at a lag of zero days for all pairs of streams. Only one set of streams had higher cross-correlation at a lag of one day rather than zero; further analysis showed the true lag lies at about 8.5 hours rather than one day.

Share

COinS