You are here:
Time-Frequency Analysis of Beach Bacteria Variations and its Implication for Recreational Water Quality Modeling
Citation:
GE, Z. AND W. E. FRICK. Time-Frequency Analysis of Beach Bacteria Variations and its Implication for Recreational Water Quality Modeling. ENVIRONMENTAL SCIENCE & TECHNOLOGY. American Chemical Society, Washington, DC, 43(4):1128-1133, (2009).
Impact/Purpose:
The overall objective of the proposed study is to evaluate the loadings, fate and transport of bacterial contaminants from agricultural non-point sources in surface waters through the use of DNA-based technology that can quantify and track fecal contamination back to its source.
Description:
This paper explores the potential of time-frequency wavelet analysis in resolving beach bacteria concentration and possible explanatory variables across multiple time scales with temporal information still preserved. The wavelet scalograms of E. coli concentrations and the explanatory variables observed at Huntington Beach, Ohio in 2006 clearly reveal well-defined short-time patterns of different periods, timings, phases, and durations, exhibiting nonstationarity and nonlinearity in all time series. If linear regression models must be used rather than more advanced nonlinear models, these observed characteristics in the time series favor modeling data over shorter durations such that nonstationarity can be reduced and recent temporal patterns in the variations adequately represented. In comparison, a model based on multiple years of data may perform poorly in response to local patterns and cause excessive model error. To further justify the statistical robustness of short-period regression models (typically weekly models), we discuss the stability of the first and second order moments. We found that models based on a duration that is 5/3 times the period of the local largest-scale time pattern can generally attain stable moments.