On rare occasions, WISE data exhibit an interesting phenomenon whose nature is unknown. A large annulus of W2 pixels, centered roughly on the pixel at coordinates [912,568] (0-indexed, as in IDL) and hereafter referred to as the "resonant pixel," appears then subsequently decays in later frames. This halo is usually preceded by a dramatic elevation of signal, colloquially referred to as a "TV test pattern" (also of unknown origin), that appears in both W1 and W2. The phenomenon is often accompanied by a W1 latent at approximately the same row as the W2 resonant pixel halo, but on the left side of the array. This W1 latent is generally present in strong instances of the W2 phenomenon.
The following figures illustrate, in chronological order, the forty-two recorded instances of this W2 resonant pixel phenomenon. In all figures, each row represents a simultaneously taken set of frames that, from left to right, represent W1, W2, W3, W4, and a three-color image formed using the W1+W2(+W3+W4) frames. The frame number in the scan is shown to the far left. Icons with open white squares or question marks indicate missing bands of data. For example, data taken during the NEOWISE Reactivation period lack W3 and W4 data.
The first instance of the resonant pixel phenomenon occurred during the WISE 4-band cryogenic period (Figure 1), followed by one instance in the WISE 3-band cryogenic period (Figure 2), three instances during the WISE post-cryogenic period (Figures 3-5), and thirty-seven instances so far during NEOWISE reactivation (Figures 6-18x).
Figure 7 - Scan 67842b, framesets 097-098. Frameset 097 exhibits a "TV test pattern". Frameset 098 shows a weak W2 resonant pixel halo. |
Figure 8 - Scan 70877b, framesets 177-178. Frameset 177 exhibits a "TV test pattern," and frameset 178 shows a weak W2 resonant pixel halo. |
Figure 9 - Scan 73648a, framesets 135-136. Like Figures 7 and 8 above, the first frameset shows a "TV test pattern" followed in the second frameset by a weak W2 resonant pixel halo. |
Figure 10 - Scan 77064a, framesets 142-146. The first frameset shows a "TV test pattern" followed in subsequent framesets by a W2 resonant pixel halo. |
Figure 11 - Scan 77978a, framesets 254-256. The first frameset shows a "TV test pattern" followed in subsequent framesets by a W2 resonant pixel halo. |
Figure 12 - Scan 78569a, framesets 148-150. The first frameset shows a "TV test pattern" followed in subsequent framesets by a W2 resonant pixel halo. |
Figure 13 - Scan 79980a, framesets 134-135. The first frameset shows a "TV test pattern" followed in subsequent framesets by a W2 resonant pixel halo. |
Figure 14 - Scan 80992a, framesets 152-159. Frameset 153, where a "TV test pattern" would be seen, failed processing. It is followed in subsequent framesets by a W2 resonant pixel halo. |
Figure 16 - Scan 82157b, framesets 151-153. The first frameset shows a "TV test pattern" followed in subsequent framesets by a W2 resonant pixel halo. |
Figure 17 - Scan 84826a, framesets 146-148. The first frameset shows a "TV test pattern" followed in subsequent framesets by a W2 resonant pixel halo. |
Figure 18 - Scan 87990a, framesets 023-026. The first frameset shows a "TV test pattern" followed in subsequent framesets by a W2 resonant pixel halo. |
Figure 18a - Scan 91539a, framesets 187-188. The first frameset shows a "TV test pattern" followed in the subsequent frameset by a W2 resonant pixel halo. |
Figure 18f - Scan 05932r, framesets 045-047. The first frameset shows a "TV test pattern" followed in the second frameset by a (very) weak W2 resonant pixel halo. |
Figure 18h - Scan 11960r, framesets 036-038. The first frameset shows a "TV test pattern" followed in the next two framesets by a decaying W2 resonant pixel halo. |
Figure 18k - Scan 28226r, framesets 152-154. The first frameset shows a "TV test pattern" followed in the next two framesets by a decaying W2 resonant pixel halo. |
Figure 18m - Scan 33862r, framesets 139-142. The first frameset shows a "TV test pattern" followed in the next two framesets by a decaying W2 resonant pixel halo. |
Figure 18s - Scan 50265r, framesets 146-147. The first frameset shows a "TV test pattern" followed in the subsequent frameset by a W2 resonant pixel halo. |
Figure 18v - Scan 52920r, framesets 151-152. The first frameset shows a "TV test pattern" followed in the subsequent frameset by a W2 resonant pixel halo. |
Figure 18w - Scan 53461r, framesets 214-215. The first frameset shows a "TV test pattern" followed in the subsequent frameset by a W2 resonant pixel halo. |
Figures 19 and 20 show 31×31 pixel cut-outs centered at the nominal resonant pixel, for each of the first framesets in Figures 1-10 that show the feature. The resonant pixel itself seems to have very low responsivity, but is also transient. Pixels in its vicinity are also transient, are excessively noisy, or have very low responsivity. Note that dead pixels have constant signal at the digital saturation level. The figures show that the nominal resonant pixel itself is not unique but more likely is a member of a tight cluster of masked pixels. Color coding on the figures is explained in the legend below:
Table 1: Color Coding of Masked Pixels
|
Excessively noisy, or known broken hardware. Static mask |
|
Dead, or low responsivity or low dark current. Static mask |
|
High dark current or responsivity, or saturated at any sample, or uncertain non-linearity. Static mask |
|
Broken pixel, or negative slope fit |
|
Saturated at any sample |
|
Transient, or unreliable non-linearity, or spike outlier |
Figure 19 - Zoom-ins centered at the resonant pixel feature from Figures 1-6. See Table 1 for an explanation of the color coding of masked pixels. |
Figure 20 - Zoom-ins centered at the resonant pixel feature from Figures 7-10. See Table 1 for an explanation of the color coding of masked pixels. |
It has been observed that bright objects falling near the right edge of the W2 array tend to produce longer-lasting latents than bright objects falling elsewhere, as discussed in IV.4.g.iv.1.e. Figure 21 shows an example of this effect. The bright satellite trail in this figure leaves a W2 latent but only in the vicinity of the resonant pixel. Figure 21a shows a similar instance where a brighter satellite trail leaves W2 latents in the subsequent four framesets, but markedly brighter at the resonant pixel position. That the resonant pixel falls within this region, where bright stars and satellite trails leave more prominent latents than elsewhere in the array, suggests that the two effects may be related. This point is further illustrated in Figures 22 and 23, which are zoomed versions of satellite trails and of the resonant pixel, with 100-pixel-radius circles centered at the latter.
Figure 21a - Scan 57227r, framesets 058-062. The bright satellite trail in frameset 058 leaves W2 latents in framesets 059-062. These latents are pronouncedly brighter at the position of the W2 resonant pixel, thus further suggesting that the latter is related to the phenomenon of longer-duration short-term latents in this region of the array. |
Figure 22 - Zoomed illustration of four of the resonant pixels from Figures 1-10. The 100-pixel radius circles are centered on each instance of the resonant pixel. |
We looked for a correlation between the intensity of the resonant pixel feature, and the signal in the illuminated frameset preceding it. To this end, we carried out aperture photometry with the ATV tool in IDL, centered at the nominal resonant pixel position, and using the radii for aperture and sky annuli illustrated in Figure 24. We separately calculated the centroid of the feature, via its first moment of inertia, and found out that it is reasonably constant, except for the faintest examples, due to noise.
Figure 24 - GUI for ATV photometry centered at the resonant pixel position, illustrated for frameset 148 in scan 11984a. |
Table 2 lists our measurements for the first 22 instances of the resonant pixel. We obtained the mean signal of the illuminated frameset (the one that often shows a "TV test pattern"). For the framesets in Figures 9 and 16, we could not obtain reliable aperture photometry of the resonant pixel feature, because of its faintness. For the framesets in Figures 1, 2, 4, 6, 14, and 18c, there were no level-1b data available for the illuminated frameset, so we used level-1a data instead. In cases where a TV test pattern was discernible, we also measured the W2 signal jump between the elevated region and the rest of the array. Figure 25 illustrates an example estimate of signal jump in a TV test pattern.
Table 2: Measured Signal in the Resonant Pixel Halo and in the Preceding Frame
Figure |
Aperture Photometry of Resonant Pixel (DN) |
Mean Signal Preceding Illuminated W2 Frame (DN) |
Signal Jump In TV Test pattern (if visible) (DN) |
Resonant "Pixel" Feature Centroid (pix) |
1 |
430,132 |
26,752 |
N/A |
[912.7, 567.3] |
2 |
80,263 |
8,423 |
600 |
[912.7, 567.3] |
3 |
647,687 |
224.6 |
6,300 |
[911.9, 567.5] |
4 |
936,058 |
9,417 |
~18,000 |
[912.2, 567.7] |
5 |
241,604 |
10,982 |
N/A |
[912.3, 567.5] |
6 |
634,134 |
14204 |
5,000 |
[912.5, 567.4] |
7 |
64,177 |
401.9 |
900 |
[912.4, 567.2] |
8 |
99,804 |
96.88 |
230 |
[912.3, 567.1] |
9 |
< 0 |
1,324 |
400 |
[912.0, 567.0] |
10 |
267,156 |
11,050 |
1,300 |
[911.6, 563.7] |
11 |
258,044 |
2,527 |
700 |
[914.0, 556.7] |
12 |
48,953 |
532.6 |
1,200 |
[908.6, 564.5] |
13 |
200,573 |
5,128 |
300 |
[912.3, 566.0] |
14 |
698,415 |
46,168 |
N/A |
[912.9, 568.8] |
15 |
41,645 |
50.6 |
10 |
[912.9, 567.2] |
16 |
< 0 |
144.6 |
460 |
[911.9, 567.1] |
17 |
150,387 |
900 |
1,200 |
[911.2, 567.2] |
18 |
148,969 |
448 |
2,050 |
[912.6, 567.0] |
18a |
111,000 |
739 |
900 |
[912.2, 566.2] |
18b |
49,065 |
730 |
815 |
[908.4, 564.0] |
18c |
852,538 |
30,011 |
N/A |
[913.2, 566.2] |
18d |
57,570 |
1,682 |
2,200 |
[913.1, 566.2] |
Figure 26 shows both the mean signal of the preceding illuminated frameset, and the typical signal jump in a TV test pattern, as a function of aperture photometry of the resonant pixel halo. Figure 26 shows that there may be a correlation between the signal jump in the preceding frame and the signal in the resonant pixel halo. We do not see a solid correlation with the mean signal of this preceding frameset. Our measurements of the latter are compromised by the lack of level-1b data for them (due, for instance, to failure to complete processing) or by our inability to measure the peak signal if this preceding frameset already represents the signal in decay.
The resonant pixel halo is related not to a single pixel, but to a cluster of them in W2, and it is located in a region of the array where long-term latents are observed. Instances are rare and are preceded by a frameset with a dramatically elevated signal, often with a "TV test pattern" effect.
Last update: 23 February 2024