Use of AstroImageJ to improve general LCs

[SCO]

AstroImageJ specializes in improving LCs for an assumed standard transit structure, or flat line structure in the case of comps.  However, there appear to be some functions of the software that could probably improve the LC of an undetermined shape and periodicity.  For example, on the Data set 3 Fit Settings window, the use of Outlier Removal, Comp Star selection, and Detrend Parameter selection might improve the reliability of the raw data.

Could you please provide some advice on how best to modify the use of the Multi-plot Main, Multi-plot Y Data, and Data set 3 Fit Settings screens to improve raw data of an LC as described above?

What I mean by that is, for example, on the Multi-plot Y Data screen, I would assume that fit mode would be set off and on the Data set 3 Fit Settings window LDCs would not be used, etc.  What other differences would you recommend?

[karenacollins]

I think the best you can do in AIJ at the moment is to detrend using airmass, or some other common detrend parameters such as Sky/Pixel_T1, Width_T1, X(FITS)_T1, X(FITS)_T1, and/or time (JD_UTC or BJD_TDB). To do that, I recommend enabling the fitting panel in the Fit Mode setting of Multi-plot Y-data, but in the fitting panel, disable "Enable Transit Fit" near the top left. This leaves only detrending active in the middle of the fit panel.

Now select each of the detrend parameters you want to test in different detrend parameter rows. If you need more detrend rows, increase "Detrend Vars" in Multi-plot Main. Then use the detrend optimizer to see what AIJ finds (Click start in that sub-panel). I'd start with "Max Detrend Pars" = 1. If that doesn't distort your light curve too much, then maybe try "2", etc (press "Start" again after changing the setting).

You can also manually test different detrending combinations by hand by using the checkboxes next to them.

For these cases where AIJ has no model to fit, you'll have to judge what detrending is acceptable by eye.

Once you are satisfied that the detrending is doing all it can do, then you might try data cleaning. I'd recommend using "N x σ" = 3 and use the <Shift>-click on Start option to remove data points beyond 3σ. You might <Shift>-click on Start a few times until you start seeing data that you want to keep being removed (or you reach a point where 0 new points are cleaned). If you get too far, you can back up and restore the previous points (up to 10 times) using the undo button.

We are looking to see if we can implement general fitting to polynomials etc, but if we find a way to implement with reasonable work, that would still be at least a few weeks away.

Karen

[SCO]

Thanks Karen,  This is extremely helpful.

Ed