[gnuastro-commits] master d96ecc33: radial-profile: setting spatial resolution to zero

[Mohammad Akhlaghi]

branch: master
commit d96ecc3310f1861302ee9b712235d1f111f21918
Author: Raul Infante-Sainz <infantesainz@gmail.com>
Commit: Mohammad Akhlaghi <mohammad@akhlaghi.org>

    radial-profile: setting spatial resolution to zero
    
    Until now, the spatial resolution used in the radial profile script was not
    specified.  Consequently, the default value was used in the call to
    MakeCatalog for measuring the radial profile
    ('--spatialresolution=2'). This parameter is related with the uncertainty
    in the measurement of the area. But since for the radial profile script we
    are considering fixed apertures, the errors of this apertures are zero.
    
    I found this issue when obtaining the surface brightness error
    ('--sberror') of simulated profiles. For each radius of a bunch of
    simulated sersic profiles, I compared the scatter (STD) of the surface
    brightness measurements ('--surfacebrightness') with respect to the
    measured surface brightness error ('--sberror'). With this experiment, I
    realized that the measurements are compatible with the expected results
    when the '--spatialresolution' is 0.0.
    
    With this commit, the parameter '--spatialresolution' has been set to zero
    for the radial profile script. We are still investigating about how this
    parameter has to be set in other/general situations.
    
    This bug was fixed with the help of Mohammad Akhlaghi.
    
    This fixes bug #62694.
---
 NEWS                         |  2 ++
 bin/script/radial-profile.in | 17 ++++++++++++++++-
 2 files changed, 18 insertions(+), 1 deletion(-)

diff --git a/NEWS b/NEWS
index 991ec33c..4f921ec3 100644
--- a/NEWS
+++ b/NEWS
@@ -203,6 +203,8 @@ See the end of the file for license conditions.
               standard deviation of the values image, while it is actually
               the root mean square of the Sky standard deviation
               pixels. Found by Raul Infante-Sainz.
+  bug #62694: Radial profile script overestimates the surface brightness
+              error. Found and fixed with the help of Raul Infante-Sainz.
 
 
 
diff --git a/bin/script/radial-profile.in b/bin/script/radial-profile.in
index c537cfbc..0f783775 100644
--- a/bin/script/radial-profile.in
+++ b/bin/script/radial-profile.in
@@ -695,10 +695,25 @@ fi
 # a catalogue using the segmentation image previously generated (the
 # elliptical apertures) and the original input image for measuring the
 # values.
+#
+# --spatialresolution=0.0: in a radial profile, the areas used for each
+#   measurement are pre-determined analytically, so the error in area
+#   (relevant for the estimation of the surface brightness error for
+#   example) should be considered as zero. The default value is '2.0' which
+#   will give over-estimated error bars in the center.
+#
+#   We tested this by making a mock profile and adding different noise
+#   realizations (just changing the random number generator seed). The
+#   radial profiles of each noise realization were then measured and their
+#   scatter was compared with the '--sberror' column of MakeCatalog when
+#   given to this radial profile script. We saw that without
+#   '--spatialresolution=0.0', the measured errors were overestimated in
+#   the center, but after adding this, the '--sberror' column was nicely
+#   consistent with the scatter in the mock profiles at central radii.
 cat=$tmpdir/catalog.fits
 astmkcatalog $apertures --hdu=1 --valuesfile=$values --valueshdu=1 --ids \
              $finalinstd $finalmeasure $finalsigmaclip $finalzp \
-            --output=$cat $quiet
+            --spatialresolution=0.0 --output=$cat $quiet