Biology Reference
In-Depth Information
FIGURE 11-25.
PHASEREF analysis of the noisy nonstationary cosine wave originally presented in Figure 11-15, again
calculated with 6-hour and 24-hour filters. However, whereas clean period estimates were obtained
from the analysis shown in Figure 11-24, the present case shows sufficient noise is present so as to
confound a clean assessment of period, arising primarily at the long-time end of the data series (the
details of which are presented in Figure 11-26).
to remove noise prior to analysis. Although this is not a good idea
in general because removing noise structure prior to analysis will
typically introduce (sometimes considerable) bias into analytical
results, it may prove beneficial to an analysis like that performed
by PHASEREF.
E
XERCISE
11-2
Figures 11-27 through 11-30 present the outputs of the procedures
COSIN2NL, FFT-NLLS, and PHASEREF performed on time series from
Figure 11-15 after preprocessing by ARFILTER. Compare the results with
those presented in Figures 11-21, 11-23, 11-25, and 11-26 showing the
outputs of the same procedures performed on the original time series.
Comment on the differences and whether or not, in your opinion, using
filtering was beneficial and/or necessary in each case.
