Unsually large GINI segmap with statmorph

[bijusaha-astro]

Dear Sir,
I was following the statmorph tutorial available on the webpage for calculation smoothness. The simulated images of galaxies are obtained from GALMER, for a given galaxy (sb type), but each corresponding to inclination angles i=20, 30, 40, 50, 60, 70 degree. I convolved the galaxy FITS file obtained from GALMER with Gaussian2D kernel of std. dev= 1.7 (corresponding to FWHM = 4 pixels, for SDSS). The convolved images are attached. These images are given as input in the for loop under which stat morph is applied. I further did a few other steps to incorporate for gain and s/n as per the instructions given on the webpage. I have attached the code. But the GINI segmap appears unusually larger in comparison to the images and the smoothness value looks absurd.

Can you please have a look at this code and suggest any improvements.Google drive link for codes and images

[vrodgom]

Hi,

The example from the tutorial consists of a 2D Sersic model with an amplitude of 1.0 at the effective radius, which is where sky_sigma = 1.0 / snp comes from (assuming snp = 100.0, i.e. a very high S/N per pixel at the effective radius). In general, however, every astronomical image will have its own values for the gain and sky_sigma, so you cannot just use the values from the tutorial.

In this particular case, since you are dealing with simulated images and are not trying to match a specific set of observations, I would suggest leaving a high value for the gain (the current value of 10^5 is fine). However, the current value for sky_sigma is clearly too large for your dataset, resulting in low S/N values. A simple solution is to assume a fixed value for the average S/N per pixel across the entire image, say snp = 25.0, and then calculate the necessary amount of background noise as:

sky_sigma = np.mean(image) / snp

I hope this helps.

[bijusaha-astro]

Thank you for the reply. While the above instruction work for the above dataset, but fails for another. Even the original image is not fitted properly. Would you please help me understand the reason? I kept decreasing snp and at snp=10, it is fine. But at snp=5, it again fails. I fail to understand this.