Matthais,
You may be right that ANOVSA is not the best choice here. I am using Peranso with has a wide variety of methods to analyze periodicity.
The author has this to say about which method to choose:
======================================================
Peranso offers a wide variety of methods to analyse periodicities. An obvious question therefore is : which method should I use for what type of object (e.g., variable star). Are some period analysis methods better suited than others for specific types of variable stars or asteroids ?
This is a very difficult question to answer, and unfortunately there is no such thing as the “universal” period analysis method, that is the best choice for whatever type of object. Below, we present some simple guidelines, that may be helpful in answering the question.
There’s an excellent paper by M.J. Graham et al, which compares several period finding algorithms, and addresses the above question too. It analyses the accuracy, reliability, performance etc of the algorithms.
Whatever method of Peranso you decide to use, always take following two remarks into consideration:
all period determination methods are dependent on the quality of the observations. Low quality light curves (few data points, noisier data) will produce results with lower reliability.
Use your eyes and your brain first before starting a period analysis. Draw a light curve of the observations, and inspect its shape, signature, characteristics, etc. We call this visual inspection. That by itself usually reveals very significant information. Unfortunately, our brain is not faultless, so we have to rely on other approaches too, using mathematical techniques. That’s where Peranso enters the picture.
The selection of a period analysis method may be influenced by many things :
the amount of observations
their spread in time (equally or unequally spaced)
type of variations (regular-shaped or not)
expected physical properties of the system (can it be multi-periodic)
etc
As a rule of thumb :
Delta Cepheids and RR Lyrae variables in general can be quite well analysed with the Lafler-Kinman method
Eclipsing binaries may produce dominant periods at half-period values. Lomb-Scargle and Generalized Lomb-Scargle are more sensitive to this, and hence are not recommended for such variables. CE is very successful in handling these stars.
Regular periodic variables (T Tau stars, Cepheids, Delta Scuti stars, Mirae, RR Lyrae, …) are well suited for CE.
If you expect the system to be multi-periodic, use CLEANest.
If the light curve is highly non-sinusoidal, use ANOVA. Otherwise, you may consider DCDFT or CLEANest.
PDM is also well suited for highly non-sinusoidal data with only a few observations over a limited period of time.
FALC is a de facto standard for asteroid period analysis. Try that one first.
If you’re studying exoplanet transits, use EEBLS.
If you suspect that the period of your variable is not constant over time, then use WWZ
In developing Peranso, I have studied hundreds of light curves of many different objects. Although there is no “universal” period analysis method, there is one that - in my humble opinion - comes pretty close, and that’s ANOVA. I have been amazed by its power to improve peak detection sensitivity and to damp alias periods. Try it out yourself, and see if it suits your data. If not, there are many others to experiment with.
======================================================
Note the final paragraph where he says ANOVA is the most universal of analysis methods.
I tried all the methods in Peranso using the last 10 years of V data from AAVSO. Here are the ones that give an ~164 day period and those that give an ~328 day period.
164 Day
Lomb-Scargle
Generalized Lomb-Scargle
Bloomfield
DFT (Deeming)
DCDFT (Ferraz-Mello)
CLEANest (Forster)
Jurkewich
WWZ
PDM
Phase-binned AoV
CE
328 Day
ANOVA
FALC (Harris)
Dworetsky (both peaks almost even)
Renson (both peaks almost even)
Lafler-Kinman (both peaks almost even)
So yeah, more methods come up with a 164 day period. But still the author seems to think ANOVA is a good first choice. In general, it works well for me.
