Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.


Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  Pp. 011-018 in Total Maximum Daily Load (TMDL) Environmental Regulations–II Proceedings of the 8-12 November 2003 Conference (Albuquerque, New Mexico USA), Publication Date 8 November 2003.  .(doi:10.13031/2013.15532)
Authors:   A. Bekele and A. McFarland
Keywords:   Water quality, Trend analysis, TMDL, Flow adjustment, LOWESS, OLS, Smoothing factor

Trend analysis is an important component of most total maximum daily load (TMDL) projects in assessing the adequacy of implementation plans on water quality improvement. One problem in trend analysis of water quality data is the confounding effect of varying streamflow conditions at the time of sampling. Commonly used methods of flow adjustment are ordinary least squares (OLS) regression and locally weighted regression and smoothing scatterplots (LOWESS) with a pre-specified smoothing factor (f) of 0.5. We compared trend analysis results using these two methods of flow adjustment as well as LOWESS flow adjustment approach with the optimal f value obtained based upon the bias corrected Akaike Information Criterion (AICc1). The objective of the study was to evaluate the importance of flow adjustment in trend analysis and determine the best flow adjustment method among these three procedures. Monthly concentrations from grab samples of soluble orthophosphate-phosphorus (PO4-P), total P (TP), and total suspended solid (TSS) were evaluated for three stream sites within the North Bosque River watershed in Central Texas. The nonparametric Kendall test was employed to test for the presence of trend. As expected, the detection of trend was greatly influenced by flow adjustment. The LOWESS procedure was considered more appropriate than OLS, because the relationships between flow and constituent concentration were nonlinear. The AICc1 based f value and f=0.5 in the LOWESS procedure gave similar trend results indicating that the default f value of 0.5 may be adequate for reducing variability in constituent concentrations due to flow.

(Download PDF)    (Export to EndNotes)