Comparing Narrower and Wider Narrowband Filters

In my recent narrowband images, I needed very long exposure times to get a decent signal in OIII and SII. And I was wondering if the small window of the Custom Scientific filters (Ha 4.5nm, SII 3nm, OIII 3nm) are maybe too small. And that they remove a lot of the background noise but also leave less of the signal through.

So, I bought Baader narrowband filters that are significantly wider (Ha 7nm, SII 8nm, OIII 8.5nm).

In order to compare them, I shot images of the jellyfish nebula. It has a very strong signal but also dark areas nearby. In order to minimize the influence of changing seeing conditions, I shot one hour with one filter, then one hour with the other and then again with the first and so forth.

I'm still struggling with my unreliable lodestar guider, so I had to throw away a number of images. But ended up with enough data to compare them. I decided to compare them by contrast and signal to noise ratio.

I calculated contrast by first measuring average signal in a bright area and in a dark area

and then calculating it Contrast = (Bright - Dark) / (Bright + Dark)

For Signal to Noise ration I measured the images with Pixinsight's subselector script. It doesn't report absolute values but relative values.

Here are the results:

	Bright	Background	Contrast	SNR
Ha 4.5nm	3.50E-04	7.83E-05	6.34E-01	2.099
Ha 8nm	4.84E-04	9.95E-05	6.59E-01	4.114
OIII 3nm	1.54E-04	9.62E-05	2.30E-01	1.543
OIII2 8.5nm	1.67E-04	1.07E-04	2.18E-01	2.481

The contrast is very similar between both filters. But the Signal-to-Noise ratio is significantly better with the wider filters.

The other question is how bad the background gradients are that get created with both filters. I used the OIII images to compare as the Jellyfish nebula is very weak in OIII:

3nm	8.5nm

Using AutomaticBackgroundExtraction, I get the following background images:

3nm	8.5nm

Clearly, the wider filter has larger gradients. But after the first background removal the residual background is:

3nm	8.5nm

Looks like we can remove the stronger gradient fairly easily.

So, this does seem to support my suspicion, that the smaller narrowband filter extend the integration time. Although I am not sure what the reason is - the Ha/SII/OIII signal should be the same with the smaller and the wider filters.

Anyway: I'll probably stick with the wider narrowband filters.