Time-Series Analyses of Water-Quality Data

Abstract
The basic concepts of Fourier and power spectrum analyses are presented in terms of their application to dissolved oxygen and temperature data of the Delaware Estuary. The results of Fourier analyses computations performed on hourly data from water-quality monitoring stations are presented. For temperature, more than 92% of the total variance is accounted for by the 365-day periodicity. For dissolved oxygen, more than 90% of the total variance is accounted for by the annual periodicity and the next four higher harmonics (semi-annual; tri-annual, etc.). The dissolved oxygen variability remaining after removal of the principal Fourier components was found to be approximately normally distributed with an average standard deviation of 0.85 mg per 1. Power spectrum analyses of the 24-hr averaged DO residual seconds indicated a peak in the region of 30 days to 60 days and for temperature a peak is dominant at 30 days. The analyses all indicate a predominance of low frequency (periods of from 7 days to 1 yr and longer) types of phenomena. The implications of the results from a sampling and control point of view are evaluated.