Relationships Between Long-term Atmospheric Wet Deposition and Stream Chemistry in Mid-Appalachian Forest Catchments

Monday, 15 December 2014
David R DeWalle1, Elizabeth W Boyer1 and Anthony R Buda2, (1)Pennsylvania State University Main Campus, University Park, PA, United States, (2)USDA ARS, University Park, PA, United States
Improved understanding of the link between atmospheric deposition and surface water quality is critical to assessing the degree to which forested watersheds have recovered from acidification. This presentation draws upon long-term (1982-2013) atmospheric wet deposition and stream chemistry time series to study how changes in atmospheric chemical inputs have been propagated to stream waters. We used autocorrelation and lagged cross-correlation techniques to analyze monthly time series describing variations of chloride, sulfate and inorganic nitrogen concentrations for four pairs of stream/deposition monitoring sites. Autocorrelation analysis revealed that individual atmospheric input time series of sulfate and inorganic nitrogen were strongly seasonal, while chloride inputs exhibited little seasonality. Stream chemistry time series exhibited gradually declining autocorrelation trends with increasing lag times suggesting that atmospheric input signals were variably damped by the forest ecosystems . Lagged cross-correlation between raw atmospheric and stream chemistry time series indicated gradual decreases in correlation within superimposed regular annual cycles of correlation over 10- 15 years of lag time. Pre-whitening of each atmospheric and stream time series using regression or ARIMA models removed the influence of long-term trends, seasonal cycles and other factors and revealed occurrence of relatively few and highly variable lag times with significant correlations. While lagged cross-correlation of raw time series data provided some useful insights into the long-term trend and seasonal nature of the linkages between atmospheric deposition and stream chemistry, cross-correlation of shorter-term residual variations after prewhitening did not produce a consistent pattern of lag times with significant correlations in our monthly time series data.