More open clusters under a near-full moon

[Martin Meredith]

 

In spite of a 97% illuminated moon which gave the sky a milky washed out look and killed off all but the brightest stars visually, I continued looking at faint open clusters in Cassiopeia. Although gradients were sometimes a struggle to suppress, it never ceases to amaze me how much is visible using EEVA techniques under such conditions (and indeed how much colour can be extracted). Here are a few captures live at the scope. N is up, all were composed of 15s L, R, G, B subs, live-combined in LAB colour.

 

NGC 436. This is apparently a young cluster similar in age to Alpha Persei, full of blueish stars, and very different from the other clusters I observed.

 

 

 

Berkeley 64. I'm not sure of the current status of this cluster as it is not mentioned in a recent (2018) GAIA-based survey of likely clusters , so it might be a chance grouping. Earlier data give an age of 1 billion years and a distance of 12-13000 light years. There is in fact little to see at the listed coordinates, but I can see a potential cluster nearby, and centred in my image. (The listed position is well to the right of centre, near the red star). There is still quite a lot of work to be done finding correct coordinates for some of these less-visited clusters. Having plotted all the GAIA data for the Berkeleys, it seems that 10-20% are not in the catalogued position!

 

 

 

 

Berkeley 2. This one seems more definite with 93 members (as of the 2018 study), a distance of 22000 light years and an age of around 800 million years. When this was building up on the screen the most notable features were the 2 arcs of stars rather like the Antennae galaxy tails, merging at the orange star in the centre of the cluster (the bright foreground stars are not part of the cluster).

 

 

 

 

The most enjoyable sight for me was Berkeley 8, both for the cluster itself and the field of foreground stars is is embedded in. Some of those foreground stars form a beautiful curve and display a remarkable range of colours which inspired me to look up their types. Unfortunately, very little data appears to be available on a cursory search.

Having observed 37 Berkeley clusters in the last week, I'm beginning to notice some common visual elements (although they are probably ilusory!). Many of them possess arcs of faint stars. For Berkeley 8, these arcs emanate from the centre like the legs of a malnourished starfish (although I count 6 of them).

Berkeley 8 is a very old cluster, estimated to be 3.2 billion years old, rich in members (252), but relatively close (for a Berkeley) at around 11,600 light years.

 

 

 

cheers

Martin

 

[vlaiv]

Have you tried flat calibration as well for this?

I'm just asking because in these images it's obvious that flats have not been applied, and I believe they help more than just dust shadows and vignetting. At least for CMOS sensors, as they seem to suffer more from pixel to pixel QE variations.

[Martin Meredith]

On this occasion no calibration was applied (just hot pixel removal). I normally do darks at least (but that part has not yet been re-implemented since I added colour) and will do flats when that aspect is implemented for individual filters.

Re flat/QE differences, bear in mind that these views are effectively zoomed in views using a tiny sensor (it's a guide cam after all) so some of the visible artefacts are no doubt due to interpolation, and there is only 1 minute max of each of R G B in all of these shots so plenty of colour noise there too. Also, there is automatic gradient removal being applied which is based on a planar 2D fit -- pretty rough but works well much of the time (but can struggle in strong moonlight). But this is EEVA not AP so nobody minds 😉. The Lodestar is a CCD rather than a CMOS sensor.

Martin

[vlaiv]

39 minutes ago, Martin Meredith said:

Also, there is automatic gradient removal being applied which is based on a planar 2D fit

Here is suggestion for additional background removal technique - do "upper" sigma reject couple of times and do linear fit on remaining pixels to do background removal.

Simply do few iterations of pixel selection - calculate average and sigma, and discard all pixels above some multiplier of sigma - for example all pixels with value higher than 3 sigma. Do linear fit on remaining pixels and remove it from the image.

You can do couple of rounds of this removal. If you do color - use same settings and you can use "AND" operator on pixel mask for each channel - that improves chance that you have background selected.

This method works very good as long as you have pure background framing the object - I suspect it will not work good on images where you have nebulosity in whole frame (there is no background which you can try to isolate in that case).

[Martin Meredith]

That's essentially what I do, actually. I take a set of random points in the image (500), remove the top and bottom 20% percentiles, then do the linear fit. Super-fast and works pretty well most of the time. (29ms or so for Lodestar 580x760 on MAcBook Air c 2015).

Example of before/after for a single 15s sub:

 

Another more typical case

Edited February 8, 2020 by Martin Meredith
Added image