Photcal transformations
Fenris wrote on Jul 12, 2007
I have been exploring the photcal package and attempting to make a good fit for a set of five standard stars. I've applied mkimsets, mkapfile, mkconfig, mknobsfile, mkcatalog, fitparams and invertfit in roughly that order. However, the output of invertfit does not seem right:
The corresponding catalog file is as follows:
Particularly the second-last star appears to have been fitted well off. As far as I understand invertfit should yield back magnitudes close or equal to those of the catalog. I am not sure why it does not, but my first guess would be that the transformation equations I have used are not in order. The config file:
My images were taken in the V, R and I filters and I arrived at these equations by modifying the new Landolt UBVRI equations slightly, having also read some advice on these forums. If that isn't what's wrong, I'm really open to any advice. The parameter file, in case it may be helpful:
Thanks in advance for any help.
cl> type ztd.calib
# Mon 22:53:33 09-Jul-2007
# List of observations files:
# std.obs
# Config: ztd.config
# Parameters: ztd.params
#
# Computed indices for program and standard objects
#
# Columns:
# 1 object id
# 2 V
# 3 error(V)
# 4 VR
# 5 error(VR)
# 6 VI
# 7 error(VI)
BD+28-4211 8.654 0.211 0.189 0.040 0.375 0.080
HD186408 6.163 0.103 0.393 0.019 0.762 0.039
HD195919 13.785 0.142 -0.860 0.027 -1.754 0.054
HD217014 -5.186 0.105 2.347 0.020 4.731 0.040
HD5612 13.851 0.180 -0.992 0.034 -2.083 0.069 The corresponding catalog file is as follows:
cl> type std.catalog
# CATALOG: std.catalog
# NCOLS: 5
# HDRLENGTH: 63
#
# ID V BV VR RI
# 15 10 10 10 10
HD5612 6.32 0.90 0.462 0.421
HD217014 5.455 0.67 0.373 0.326
HD195919 8.981 0.05 0.031 0.038
HD186408 5.980 0.648 0.357 0.341
BD+28-4211 10.53 -0.34 -0.147 -0.17 Particularly the second-last star appears to have been fitted well off. As far as I understand invertfit should yield back magnitudes close or equal to those of the catalog. I am not sure why it does not, but my first guess would be that the transformation equations I have used are not in order. The config file:
cl> type ztd.config
catalog
V 2
BV 3
VR 4
RI 5
# Declare the observations file variables
observations
TV 3 # time of observation in filter V
XV 4 # airmass in filter V
xV 5 # x coordinate in filter V
yV 6 # y coordinate in filter V
mV 7 # instrumental magnitude in filter V
error(mV) 8 # magnitude error in filter V
TR 10 # time of observation in filter R
XR 11 # airmass in filter R
xR 12 # x coordinate in filter R
yR 13 # y coordinate in filter R
mR 14 # instrumental magnitude in filter R
error(mR) 15 # magnitude error in filter R
TI 17 # time of observation in filter I
XI 18 # airmass in filter I
xI 19 # x coordinate in filter I
yI 20 # y coordinate in filter I
mI 21 # instrumental magnitude in filter I
error(mI) 22 # magnitude error in filter I
# Sample transformation section for the new Landolt UBVRI system
transformation
fit v1=0.0, v2=0.17, v3=0.000
const v4=0.0
VFIT : mV = V + v1 + v2 * XV + v3 * VR + v4 * VR * XV
fit r1=0.0, r2=0.08, r3=0.000
const r4=0.0
RFIT : mR = (V - VR) + r1 + r2 * XR + r3 * VR + r4 * VR * XR
set VI = VR + RI
fit i1=0.0, i2=0.03, i3=0.000
const i4=0.0
IFIT : mI = (V - VI) + i1 + i2 * XI + i3 * VI + i4 * VI * XI
My images were taken in the V, R and I filters and I arrived at these equations by modifying the new Landolt UBVRI equations slightly, having also read some advice on these forums. If that isn't what's wrong, I'm really open to any advice. The parameter file, in case it may be helpful:
cl> type ztd.params
# Mon 22:52:46 09-Jul-2007
begin VFIT
status 0 (Solution converged)
variance 0.1189731
stdeviation 0.3449247
avsqerror 1.
averror 1.
avsqscatter 0.
avscatter 0.
chisqr 0.1189731
msq 0.04758924
rms 0.2181496
reference mV
fitting V+v1+v2*XV+v3*VR+v4*VR*XV
weights uniform
parameters 4
v1 (fit)
v2 (fit)
v3 (fit)
v4 (constant)
derivatives 4
0.1
0.1
0.1
0.1
values 4
-1.269951
1.433493
5.268209
0.
errors 4
1.74234
0.7701116
1.011417
0.
# Mon 22:52:50 09-Jul-2007
begin RFIT
status 0 (Solution converged)
variance 0.107224
stdeviation 0.3274508
avsqerror 1.
averror 1.
avsqscatter 0.
avscatter 0.
chisqr 0.107224
msq 0.0428896
rms 0.2070981
reference mR
fitting (V-VR)+r1+r2*XR+r3*VR+r4*VR*XR
weights uniform
parameters 4
r1 (fit)
r2 (fit)
r3 (fit)
r4 (constant)
derivatives 4
0.1
0.1
0.1
0.1
values 4
-2.333274
1.578989
6.340056
0.
errors 4
1.651672
0.7272941
0.9640136
0.
# Mon 22:52:54 09-Jul-2007
begin IFIT
status 0 (Solution converged)
variance 0.09798328
stdeviation 0.3130228
avsqerror 1.
averror 1.
avsqscatter 0.
avscatter 0.
chisqr 0.09798328
msq 0.03919331
rms 0.197973
reference mI
fitting (V-VI)+i1+i2*XI+i3*VI+i4*VI*XI
weights uniform
parameters 4
i1 (fit)
i2 (fit)
i3 (fit)
i4 (constant)
derivatives 4
0.1
0.1
0.1
0.1
values 4
-0.3178893
1.084767
3.626348
0.
errors 4
1.428565
0.6251633
0.4498163
0.
Thanks in advance for any help.
Francisco Valdes wrote on Jul 12, 2007
Hi,
I realize this is something you asked about quite some time ago but if you still want someone to take a look at it please respond. You sent good detailed info but did not send the actual observation file (std.obs). If you could do that I would be glad to give you comments.
Yours,
Frank Valdes
I realize this is something you asked about quite some time ago but if you still want someone to take a look at it please respond. You sent good detailed info but did not send the actual observation file (std.obs). If you could do that I would be glad to give you comments.
Yours,
Frank Valdes
Fenris wrote on Jul 12, 2007
Hello,
indeed it has been some time but the issue remains. I was actually just about to post again and add some things, so nice coincidence. About the obs file, had a feeling I had forgotten something...
I have attempted to retrace the steps of this procedure with slightly simpler transformation equations (without the last term, the one including both colour and airmass) but the results are effectively the same.
***
It has occurred to me that it might not be the actual photcal procedure that has me wrong but rather that the photometry itself is somehow in error. Looking for instance at the HD5612 observations, my photometry (through daophot/allstar) yields for magnitudes:
But looking at the catalog values for that star...
...it would seem that, looking only at the differences in magnitudes between colours, that something is a bit odd. As I read the catalog HD5612 is faintest in V, then somewhat brighter in R and yet somewhat brighter in I. But my photometry has V and I roughly equal, with R about one magnitude brighter than both? I can see how that would confuse the photcal procedures later on! Of course airmass is not accounted for, but the images were taken after one another on very short exposures so that couldn't make up such a difference.
Having briefly reviewed my photometry scripts, I couldn't see anything wrong so somewhat lost at the moment. But I've probably added enough questions for the time being anyhow, and will conclude this post for the time being.
Thanks for your help, it would be a very good thing if I could get this sorted one way or the other quite soon.
indeed it has been some time but the issue remains. I was actually just about to post again and add some things, so nice coincidence. About the obs file, had a feeling I had forgotten something...
cl> type std.obs
# FIELD FILTER OTIME AIRMASS XCENTER YCENTER MAG MERR
BD+28-4211 V 22:56:29.0 2.162 645.460 398.210 11.477 0.002
* R 22:57:29.0 2.171 647.240 397.270 10.756 0.002
* I 22:57:58.0 2.175 648.460 397.100 11.681 0.002
HD186408 V 22:35:44.0 1.769 642.470 779.970 9.497 0.001
* R 22:36:23.0 1.772 643.910 778.730 8.724 0.001
* I 22:36:52.0 1.774 644.400 779.050 9.770 0.002
HD195919 V 22:25:49.0 2.545 282.040 492.160 11.635 0.002
* R 22:26:36.0 2.555 283.440 489.460 10.896 0.001
* I 22:30:57.0 2.612 283.460 489.920 11.695 0.002
HD217014 V 22:42:48.0 1.996 700.340 774.720 8.768 0.001
* R 22:43:25.0 2.000 701.010 774.780 8.170 0.001
* I 22:43:54.0 2.004 703.100 773.920 9.096 0.002
HD5612 V 22:49:06.0 1.690 636.750 238.530 9.775 0.001
* R 22:50:07.0 1.693 638.490 239.130 8.891 0.002
* I 22:50:31.0 1.695 639.230 238.360 9.901 0.002
I have attempted to retrace the steps of this procedure with slightly simpler transformation equations (without the last term, the one including both colour and airmass) but the results are effectively the same.
***
It has occurred to me that it might not be the actual photcal procedure that has me wrong but rather that the photometry itself is somehow in error. Looking for instance at the HD5612 observations, my photometry (through daophot/allstar) yields for magnitudes:
cl> pdump HD5612*good.phot image,mag yes
HD5612_I_1.fit 9.901
HD5612_R_1.fit 8.891
HD5612_V_1.fit 9.775But looking at the catalog values for that star...
cl> type std.catalog
# CATALOG: std.catalog
# NCOLS: 5
# HDRLENGTH: 63
#
# ID V BV VR RI
# 15 10 10 10 10
HD5612 6.32 0.90 0.462 0.421
(...)...it would seem that, looking only at the differences in magnitudes between colours, that something is a bit odd. As I read the catalog HD5612 is faintest in V, then somewhat brighter in R and yet somewhat brighter in I. But my photometry has V and I roughly equal, with R about one magnitude brighter than both? I can see how that would confuse the photcal procedures later on! Of course airmass is not accounted for, but the images were taken after one another on very short exposures so that couldn't make up such a difference.
Having briefly reviewed my photometry scripts, I couldn't see anything wrong so somewhat lost at the moment. But I've probably added enough questions for the time being anyhow, and will conclude this post for the time being.
Thanks for your help, it would be a very good thing if I could get this sorted one way or the other quite soon.
Fenris wrote on Jul 12, 2007
Allright, it seems I got it sorted out at last. Indeed the error seems to have been not with my photometry, nor with my photcal usage, but rather with my eyes. It turns out I have actually done my photometry on the wrong stars (!) in not one, but two of the five images. That'll have got it quite wrong of course and the fact that I did not realize until now has left me feeling thoroughly silly.
Using the right stars, the magnitudes I obtain with photcal are:
To compare with catalog values:
The reason there are only four stars in the .calib file is one of the standard stars I meant to do photometry on was not even in the image and so is left out. All discrepancies are now within 0.05 mag between catalog values and what was obtained backtracking with photcal, which feels reasonable to me.
Using the right stars, the magnitudes I obtain with photcal are:
cl> type std.calib
# Tue 19:28:57 14-Aug-2007
# List of observations files:
# std.obs
# Config: std.config
# Parameters: std.params
#
# Computed indices for program and standard objects
#
# Columns:
# 1 object id
# 2 V
# 3 error(V)
# 4 VR
# 5 error(VR)
# 6 VI
# 7 error(VI)
BD+28-4211 10.522 0.027 -0.150 0.034 -0.322 0.051
HD186408 6.024 0.002 0.375 0.002 0.723 0.002
HD217014 5.455 0.002 0.373 0.002 0.699 0.002
HD5612 6.283 0.002 0.447 0.003 0.862 0.002
To compare with catalog values:
cl> type std.catalog
# CATALOG: std.catalog
# NCOLS: 5
# HDRLENGTH: 63
#
# ID V BV VR RI
# 15 10 10 10 10
HD5612 6.32 0.90 0.462 0.421
HD217014 5.455 0.67 0.373 0.326
HD195919 8.981 0.05 0.031 0.038
HD186408 5.980 0.648 0.357 0.341
BD+28-4211 10.53 -0.34 -0.147 -0.17 The reason there are only four stars in the .calib file is one of the standard stars I meant to do photometry on was not even in the image and so is left out. All discrepancies are now within 0.05 mag between catalog values and what was obtained backtracking with photcal, which feels reasonable to me.
Last post on Jul 12, 2007