PSF Kernels for Convolution

[JosephHall]

Hello,

I am currently trying to convolve emission models with the eROSITA PSF. I saw this post from the EDR forum about working with PSFs, and when I checked that folder I saw that there were 2D PSF images included as well as the "slet" PSF files pointed to in that thread. I have a few questions about these files:

* Are these 2D images appropriate to be used as image convolution kernels?

* I was also wondering what the difference is between the different versions available for each of the telescope modules?

* Finally, for a "TM0" PSF image, am I right in thinking it is just the average of these images over all of the telescope modules?

Any help you can give me would be greatly appreciated, thank you.

Joseph Hall

[JeremySanders]

Hi Joseph

* Are these 2D images appropriate to be used as image convolution kernels?

Yes, that is how srctool does its PSF correction, and what ermldet uses in image mode source detection. They are derived from PANTER ground-based measurements. Each HDU is a different energy. You may want to make them symmetric about the centre, as the asymmetric structure will be washed out by the sky coordinate transformation and different TM module rotation, though srctool/ermldet doesn't do this yet. Also check the pixel scale in the image HDU.

* I was also wondering what the difference is between the different versions available for each of the telescope modules?

There are some changes with time (mostly I remember removing >4 arcmin radius for consistency with the ARF). There are also the confusingly named survey averaged file (tmX_2dpsf_190220v03.fits) and the one which is resolved into off-axis angle (tmX_2dpsf_190219v05.fits). Check the CALDB index file to make sure you have the latest version.

* Finally, for a "TM0" PSF image, am I right in thinking it is just the average of these images over all of the telescope modules?

The PSF data for all the TMs is currently exactly the same. At some time in the future we will make a new improved PSF based on in-flight measurements (see Merloni et al. 2024 appendix).

Cheers

Jeremy

[lidawei]

Hi,

I would like to plot a PSF radial profile in the 0.2--2.3 keV range and compare it with the profile of a real source. However, I have some questions:

Firstly, there are many different types of PSF files in the caldb, and I am not sure which one to choose.

Secondly, what is the pixel scale of these PSF images in arcseconds? I am not quite certain about this.

Lastly, is it acceptable to take the radial profile of a chosen PSF image and directly compare it with my source to illustrate the extent of my source's emission?

I would greatly appreciate it if someone is willing to answer.

[JeremySanders]

Hi

I would like to plot a PSF radial profile in the 0.2--2.3 keV range and compare it with the profile of a real source. However, I have some questions:

Firstly, there are many different types of PSF files in the caldb, and I am not sure which one to choose.

For a source in the survey, tm1_2dpsf_190220v03.fits is probably the right one to use. All TMs are assumed to be the same currently. The file consists of different HDUs as a function of energy (see the EXTNAME keyword for each HDU in the file). For consistency with the ARF, they should be normalised to be 1 inside 4 arcmin radius.

Secondly, what is the pixel scale of these PSF images in arcseconds? I am not quite certain about this.

This is given in the CDELT1/CDELT2 keywords, in arcsec (currently). This changes as a function of energy, so it's different in each HDU.

Lastly, is it acceptable to take the radial profile of a chosen PSF image and directly compare it with my source to illustrate the extent of my source's emission?

Yes, though there are some (small) differences between the in-flight stacked PSF and the CALDB one, see Merloni et al 2024, Fig A.1, or https://erosita.mpe.mpg.de/dr1/eROSITA_ ... ation.html. The PANTER one is what's in the CALDB. Probably we should provide the numerical values for the current stacked profile somewhere.

Please also be aware the PSFs above only go to 4 arcmin radius. See the Churazov et al. (2023) for more information about the PSF at larger radii.

EDIT: if you compare against a real source, don't forget about pile-up (for very bright sources) and background.

[lidawei]

It works for me. Thank you!