VI. Analysis of Release Products
4. Photometric Accuracy
b. Photometric Accuracy
The WISE photometric accuracy requirement concerns bright, S/N > 100, stars that not
confused by nearby sources, or any way resolved, and are free of saturation.
The stated requirement is as follows:
The root mean square error in relative photometric accuracy in the WISE Source Catalog shall be better than 7% in each band for unsaturated point sources with SNR>100, where the noise flux [includes] errors due to zodiacal foreground emission, instrumental effects, source photon statistics, and neighboring sources.
This requirement shall not apply to sources that [are] superimposed on an identified artifact.
We will be assessing the photometric accuracy of the WPRO (profile-fitting) photometry
of bright stars. However, in order to draw the sample to study, we will be using the S/N
as given by the standard aperture photometry. The reason is that the WPRO S/N includes a contribution
due to the uncertainty in the Point Spread Function (PSF), essentially flooring all
S/N values to be less than S/N=60 or so; whereas, the aperture photometry provides a reasonable
estimate of source's S/N in a given band.
Highlighted Sample
For detailed study (Figures 1-4, presented below), we will first draw a clean sample of bright stars. And to test the
photometric accuracy in specific regions of the sky, including more complex ones,
we will also draw an additional seven samples. The results are presented below (see Figure 5).
- Aperture photometry magnitude uncertainties: w?sigm < 0.01086 mag (S/N > 100)
- Avoid Galactic Plane and confused regions (|glat| > 40 deg)
- Avoid saturation, artifacts, blending, etc
- Avoid upper limits
- Apply w?rchi2 threshold to avoid extended sources and poor WPRO photometry
- WISE Level 3 Atlas Source Table
For example, the SQL command for W1 (band-1) extractions is as follows:
WHERE (abs(glat) > 40 and (wx >= 36.4 and wx <= 4058.6 and wy >= 36.4 and wy <= 4058.6)
and use_src = 1 and nb = 1 and w1frl = 0
and w1rchi2 <= 3 and w1sat = 0 and cc_flags[1]
not matches '[DHOP]' and w1sigmpro is not null and w1sigm < 0.01086) '
The database draw for each band extracts ~1/2 million W1 sources drawn, one-quarter million W2 sources,
25,000 W3 sources and 2000 W4 sources. Examples of what some of these sources look like are
shown in Figure 1.
 |
 |
| Figure 1a - Unsaturated and unconfused bright, S/N > 100, sources in the field RA,DEC = 213.980, +10.070 deg, as viewed by the W1 (blue), W2 (green) and W3 (red) bands.
|
Figure 1b - Zoomed-in view of bright sources in field RA,DEC = 213.980, +10.070 deg.
|
The extracted photometry for the unsaturated, bright S/N > 100 sources is presented in
Fig. 2., comparing the WPRO (profile-fitting) photometry
(parameter w?mpro & w?sigmpro) with the standard aperture photometry
(w?mag & w?sigm).
The estimated W1 magnitude uncertainties are shown in Fig 3 for both sets of photometry.
Note the floored uncertainties for WPRO, w?sigmpro > 0.02 mag, while the standard aperture
mag uncertainties fall below 1% (S/N > 100) for W1 < 13th mag.
 |
 |
| Figure 2 - WPRO vs. WAPPco photometry for bright S/N > 100 sources (see Fig. 3).
|
Figure 3 - Estimate magnitude uncertainties for W1 photometry.
Note the floored uncertainties for WPRO, w?magsig > 0.02 mag, while the standard aperture
mag uncertainties fall below 1% (S/N > 100) for W1 < 13th mag.
|
Internal Repeatability
The WISE photometric accuracy is assessed using the
internal repeatability of WPRO measurements.
For a given source found on M individual frames (typically ~12), WPRO estimates the flux for each one and
the distribution statistics are part of the extracted information (per band) for the source.
The reported statistics include weighted mean, the unbiased weighted sample variance
(re: population variance) and the standard error of the mean.
Internal repeatability is a proxy for photometric accuracy, although it is not sensitive
to confusion noise. For our purposes here, confusion is not an important contributor to the flux uncertainty.
The W1 repeatability results for the bright S/N > 100 sample are shown
in Fig 4. The top panel shows the standard deviation in the population (parameter
w?sigP1),
representing the RMS of WPRO measurements for one frame. The bottom panel shows the standard deviation
in the mean
w?sigP2),
which basically represents the accuracy of the WPRO measurement from combining M (w?m)
individual frame measurements. The results show that the photometric accuracy for W1, as indicated by the
repeatability RMS (w?sigP1), is 2 to 3%, which is well within the requirement of 7% for bright S/N > 100 sources.
 |
| Figure 4 - WPRO internal repeatability RMS of W1. The top panel shows the standard deviation in the population,
representing the RMS of WPRO measurements for *one* frame. The bottom panel shows the standard deviation
in the mean, which basically represents the accuracy of the WPRO measurement from combining M individual frame measurements.
The red dashed line demarks the Photometric Accuracy requirement.
|
The following tables show the W1, W2, W3 and W4 repeatability results for seven other regions of the sky, including
high S/N>100 draws from the South Equatorial Pole, Ecliptic Plane, SDSS, CDF-S, open cluster
M67 and the Galactic Plane. Consistently in every case, the standard deviation in the repeatability population
distribution (w?sigP1) ranges between 2 and 3%, surpassing the requirement.
 |
 |
 |
 |
 |
 |
 |
Fig 5a - South Equatorial Pole,
internal repeatability RMS (w?sigP1). |
Fig 5b - Taurus Molecular Cloud. |
Fig 5c - SDSS. |
Fig 5d - CDF-S |
Fig 5e - Cluster M67 |
Fig 5f - Galactic Plane |
Fig 5g - |glat| > 20 deg |
Summary
WISE photometric accuracy, estimated using the internal repeatability of WPRO multi-frame measurements,
surpasses the survey requirement by more than a factor of 3, in all 4 bands, in the 8 test fields presented
in this document.
Last update: 2011 April 11
Previous page Next page
Return to Explanatory Supplement TOC