# How should I interpret this result obtained using lomb scargle?

I used lomb scargle for periodic analysis.

The time interval used is constant at 500 seconds.

#My data (x=time, y=intensity values) is :

x= ( 00:28, 00:37, 00:45, 00:53, 01:02, 01:10, 01:19, 01:27, 01:36, 01:45, 01:53, 02:01, 02:11, 02:19 )

y=(18264.682317 ,18720.708081 ,18476.409461 ,19592.046320

,19166.021184 , 18984.852405 , 20971.112057 ,20946.024653

21590.194169 , 21771.175308 , 21474.776018 , 20156.969939

21688.756080 , 22570.152358 )

--> The time interval of x data is equal to 500 seconds.

and lomb scargle code is:

frequency,power = LombScargle(x,y).autopower(nyquist_factor=1)
ax.plot(frequency,power,'-')
#ax4.axvline(x=best_frequency)
ax.set_ylabel('(4-2) band R₁(2.5) Intensity - Lomb Scargle Power')

1. How should I interpret the lomb scargle results shown in the picture(below picture)?

2. Why does the maximum appear near zero(below picture)?

• What is lomb scargle? For readers like me who have no idea, could you add a link or reference? Thanks!
– uhoh
Feb 28 at 8:29
• Welcome to astronomy SE! As @uhoh states, a bit of editing your question would help. Feb 28 at 11:10

Not being familiar with Lomb-Scargle (never having used it myself) the first thing I would say is that your graph is not really suitable for this kind of analysis. If I graph it in my spreadsheet I get what looks more like a "bumpy line" not a periodic pattern :

The purpose of this algorithm is to take data that has a periodic pattern and find try and determine the frequency spectrum of that periodic data. If you use data that is not periodic or does not properly exhibit a range of data values that allow discovery of the period then you unsurprisingly do not get sensible results. So what I'm saying is that I suspect your data is not going to give you useful results with this algorithm. I think you need more data points over a longer timescale.

Note my limited understanding is based on how similar algorithms are used for analyzing light-curve data. This is well explained in some detail in this paper on Arxiv. THe first section has no mathematics at all and illustrates a typical use case of the algorithm. Make particular note that you need data covering more than than the period of the light curve to get a sensible result. Note also the way the first "maximum peak" in the power spectrum is the period. Figure (1) is that paper shows the raw data (a jumbled mess covering multiple periods) and figure (2) shows the way the period is identified and how the raw data can be graphed using this discovered period to properly align out of phase values. I am dubious that your data is suitable, possibly because the underlying data is not periodic or is only a partial window of a period.

Section 8 of that paper has several pieces of advice about effectively using this algorithm, so I would recommend you read that even if you have to skip the mathematics in the middle of the paper.