Baltimore Ecosystem Study Institute of Ecosystem Studies

2011 BES Annual Meeting Presentation and Poster Abstracts

Spectral analysis of long-term and high-frequency chloride and nitrate stream chemistry data in the Baltimore Ecosystem Study LTER watersheds
VerHoef, Jason
Co-Authors: Claire Welty, Sujay Kaushal, and Andrew Miller

Abstract: Spectral analytical methods have been set forth by others to advance the understanding of watershed processing of chemical species through evaluation of time series data (chemical constituents in precipitation and stream flow) in the frequency domain. These methods have not been applied to urban areas or to nested watersheds spanning a gradient of urbanization. Using NADP (National Atmospheric Deposition Program) precipitation chloride data from Beltsville, MD, 13 years of chloride and nitrate data from 8 BES long-term stream sites, and recent nitrate and specific conductance high-frequency sensor data from Dead Run Franklintown, we calculated power spectra of stream nitrate and chloride and of precipitation chloride data. We carried out computations by applying a Date-Compensated Discrete Fourier Transform algorithm to time series data followed by processing with a variable triangular smoothing window and aliasing filter. Analysis of BES stream chemistry data showed that all streams exhibit fractal behavior, and that fractal slopes of nitrate and chloride spectra decrease with increasing watershed impervious surface area. Analysis of high-frequency sensor data showed fractal scaling for both nitrate and chloride spectra for Dead Run Franklintown, where fractal slopes were greater and shifted upward in comparison to the slopes of the spectra for the long-term BES data at the same site. Watershed mean transit times, calculated by taking the ratio of the spectral power of streamflow chloride to that of precipitation chloride, were found to range from 0.125 to 4 years for the 8 BES watersheds, with an average value of 1 year.