Petrosian Radius calculation

[igorkolesnikov13]

Hello, I'm using Sextractor to run some benchmarks on the code I'm working on.
I'm mainly focused on Petrosial Radius (PETRO_RADIUS).

I have two main questions that I could not find in the documentation:

1. Usually, Petrosian Radius is calculated on a fixed fraction of the total flux of the galaxy given by ETA function. The most common value of the fraction is 0.2. It seems that it is not possible to set this value as the input parameter. What value is used internally in Sextractor? Is it possible to influence it with some input parameters?
2. Regarding units of PETRO_RADIUS. The documentation states that "Petrosian apertures in units of A or B". Im a bit confused in a sense, how can I convert it to pixels? Do I multiply by A or B? Or sqrt(A*B)?

Thanks ahead for the help.

[ebertin]

Hi @igorkolesnikov13,

1. Yes the fraction is 0.2 and it is hard-coded.
2. In SExtractor automated aperture measurements, including Petrosian magnitudes, are done in an ellipse which is a scaled version of the ellipse derived from the 2nd order moments of the isophotal light distribution. So the area of the photometric aperture would be $\pi \times \tt{A\_IMAGE} \times \tt{B\_IMAGE} \times \tt{PETRO\_RADIUS}^2$ pixels.

[igorkolesnikov13]

Hi @ebertin, thank you so much for the answer.

I was running some tests on galaxies from SDSS DR7. I found a pretty big difference between the Petrosian radius (Rp) provided by the SDSS catalog and that given by SExtractor (using default params, except for pixel scale). I used background subtracted and cleaned (without secondary objects) stamps as input.

To get the Rp in pixels from SExtractor, I used the formula for the area you provided earlier to get the area and then got the Rp with sqrt(area/pi).

I was wondering if you could point some direction for me to try to understand why there is such a difference, especially for elliptic galaxies.

[ebertin]

Hi @igorkolesnikov13, this is quite puzzling. It works fine with spiral galaxies (except for glitches for some galaxies), so the formula seems right. Did you try to change (lower) the detection (and analysis) thresholds ? In SExtractor measurements always depend a bit on the detection threshold, as the scaling of the measurement window depends partly on the object footprint.