0
$\begingroup$

I'm observing weird symmetry and repeating pattern on my unevenly sampled time series data after Lomb-Scargle transformation. I used astropy lomb-scargle.

# convert time to frequency
t_delta = [(t[i + 1] - t[i]).total_seconds() for i in range(len(t) - 1)]
t_sec = np.cumsum(t_delta)
f = 1 / t_sec

from astropy.stats import LombScargle

frequency, power = LombScargle(t_sec, inj).autopower()

After Lomb-Scargle transformation, the plot looks like this:

enter image description here

Upper one is the transformed data, and the bottom one is the original data. In the transformed data plot, the point where y-value is negative is just an outlier (there was a problem with my data set at that point).

I have two other plots:

enter image description here

enter image description here

As you can see, something is happening around x=0.017 ~~ every 33 sec and x=0.033 ~~ every 66 sec. And the patterns occurring around x=0, x=0.017, and x=0.033 seems to be all identical, having similar height and symmetrical. I don't understand why this is happening.

I have another data set that has a little different patterns:

enter image description here

This one also has very small local peaks around x=0, x=0.017, and x=0.033, but major peaks are taking place around x=0.014 and x=0.020.

Based on these patterns, my conclusion is that, the larger the frequency of the oscillation, the farther apart the symmetrical peaks are from each other. If there no oscillations like the first plot, the peaks are just one clean spikes.

I have three questions:

  1. Why is the symmetrical and repeating pattern observed every 33 sec (x=0.017) and 66 sec (x=0.033), after Lomb-Scargle transformation?

  2. Why are the symmetrical peaks farther apart from each other when the oscillation has high frequency?

  3. What is the significance of the y-axis in Lomb-Scargle transformation?

Please note that the samples are unevenly sampled, either every 1 min or 2 min (sampling rate switched between the two quite frequently)

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.