22 degrees altitude in Sagittarius. Only a 0.10 magnitude drop, the max duration is 76 seconds. Bright target permits short integration to get high accuracy. It should be do-able even in our 8" scopes. The target star is magnitude 12.8, which by itself should be doable with good accuracy at a high dark site at 2x or 4x. Probably have to reduce gain quite a bit to avoid saturation. Saturation of pixels will be the killer, a saturated image won't show any dip at all, of course. Also, the target is rather close to a brighter 9.6 magnitude star.
We'll be getting this from MIRA Observatory, as a late development.
This is the view in the Q70 eyepiece ifusing our 8SE telescopes |
This is the same 8SE scale but unflipped, so that it would correspond to the MIRA telescope w/o diagonal. |
This is the 8SE (larger) and MIRA with 0.5x reducer (insert) views. Likely will be best to not use the 0.5x reducer for such a bright target; spread the light over more pixels to avoid saturated pixels. |
I recorded on the 14" Planewave Cassegrain telescope on the 36" MIRA telescope mount at f/7.2. I was worried about saturation, and set my parameters to reduce light a little more than necessary. The 9.6 magnitude star above was used for tracking, and showed only ~1 saturated pixel typically. The target showed no saturated pixels. I 'reset' the Watec, then made the following settings, by seat-of-the-pants looking at the image on the monitor and trying to make sure the target was still scintillating and hoping the reference star wasn't saturated.
Watec settings were:
integration = 2x
gain = 11
gamma = 0.2
sharpness =19
In hindsight, I should not have used gamma=0.2, but instead left it at 1.0, and then taken up any excess brightness by defocusing and lowered gain. However, Kirk set his gain much higher, yet still had no saturation trouble with the reference 9.6 magnitude star. His much longer focal length spread the light out well enough for a very good recording. I will say that this occultation is the shallowest event we've ever tried, so how to maximize S/N was going to have to be an educated guess.
For me, the target had too much of its signal subtracted, and the reference star had a saturated pixel which no doubt dampened its variance during the long event. Given how shallow the event was, this was a problem. Kirk was able to avoid saturation through the 36" scope at larger plate scale and more pixels with signal.
I tried analysis with 3 different choices in PyMovie 3.7.3. Experienced PyMovie users will understand the wonky terms...
First choice: let the target be a static aperture, and the reference star be a 'snap to blob' aperture.
 |
 |
 |
 |
 |
 |
 |
The variations in the target star, especially near the end of the recording, show that the reference star was significantly saturated; it's co-variance was much less, so that using it as a reference star did not help much with quieting the target variance and isolating the occultation event. That wasn't going to be fixed by changing reduction parameters, alas. The problem came from the Watec settings at the event time. Still, it was likely going to be a poor choice to have photometry methods differ between target and reference star. Anyway, here's the timings...
D: 10:51:40.46
R : 10:53:08.05
Duration = 87.59s
This interval is a full 5 seconds longer than Kirk's, at the identical geographic position. Obviously they should agree, and don't.
Second choice: Let both target and reference star be 'snap to blob' apertures. This will insure consistent photometry methods between the two stars. I used the last bright point at the D to define the beginning interval containing the D, and I used the first full brightness point as the last point in the R interval, and it yielded a solution much closer to Kirk's better data. Kirk's S/N=2.19. Mine on the 14" scope was only 0.75 due to the low gamma decision, less signal to work with.
 |
 |
 |
Trial #2: I found this was the best solution of all the trials I made (some not included here in this webpage).
PyOTE log file
D: 10:51:43.027
R: 10:53:08.05
Duration = 85.02s This is just 2 seconds longer than Kirk's, and is well within the 2-sigma error bars.
Event center = 10:55.54 UT which is 49 seconds early by my data - very surprising for such a high rank event with a well known and long-observed early discovered main belt asteroid as Eunomia. And, which had an expected event time 1-sigma error of only 11 seconds. This will get some scrutiny from the data auditors, I expect.
I did try one more...
Third choice: Let all apertures be static apertures, and let target be on a range of apertures.
The range of apertures were from 2.5 to 5 pixels, but it in fact made no perceptible difference which ap size choice was best. I thought perhaps the 5px choice was marginally better, and also that was what Jerry Bardecker found for this event, and so that is the one used below.
 |
 |
 |
|
|
While all histograms say there is 0 chance it was a false positive, the choice of where to place the D and R intervals showed that a variety of choices all resulted in nearly identical 0 false positive results, so the D and R times here should not be over-interpretted as the right ones. I chose by eye where I thought the highest probability of the D and R points lay. This solution did not improve on the 2nd Choice trial above. That could possibly change with a later version of PyMovie and PyOTE, although Kirk did not find that to be the case here, in his trials.
I tried also to give PyOTE a range of event intervals from 2270 to 2700 data points in order to find the highest probability event, but that means checking over 4 million possible choices, and PyOTE is being extremely slow about solving that. It gave up, after a couple of hours of trying. I wanted to see if it gave a better fit for a shorter D-R interval, as my 85 seconds is longer than Kirk's 83 seconds on the same scope mount, and also longer than Jerry's 77 sec on the nominal centerline further north. There are comparable dips of much shorter time length in the target asteroid/star, clearly. Given the partial saturation of my reference star, the wiggles late in the recording period should be viewed with suspicion and are almost certainly very light haze layers or otherwise noise. I am choosing the 2nd choice as the one to report, but my feeling is that Kirk's data should be more reliable. All 3 observers (Jerry B in Carson City, NV) agree the event was quite early. Kirk's signal on the 36" is especially strong and it's unambiguous the event was very early.
RN's Report sent to IOTA at 1am on May 25, 2023.
Kirk used the 36" MIRA main telescope, at f/10. His plate scale spread the stars over more pixels and his settings seemed optimum for extracting a clean reference-calibrated event. The reference star shows the same wiggles as the target towards the end of the recording period. No doubt very thin cloud came over, which is masked on my light curves because the reference star was too saturated. Kirk's reference star also showed saturation of 1 or 2 pixels, but he had far more pixels showing signal than I did, so my 1 saturated pixel no doubt counted for a lot more brightness change dampening than his 1 or 2 pixels did. That's my thinking on that issue.
His settings for the Eunomia recording were:
integration: 2x
gain 34
gamma 1
sharpness 4
 |
Raw light curve showed noise or cloud wiggles in addition to the occultation, which looked to be a full diameter early (however, some of this might be due to being at MIRA instead of Santa Cruz? Not likely nearly enough to account for the time difference. Kirk's vertical bar corresponds to the predicted center time for Santa Cruz. |
Passes the zero false positive test very well. Even though the dip is shallow, having thousands of data points in the occultation interval is what gives such a solid histogram. Even my own noisier light curve passes almost as well. |
The most striking difference bewteen my light curves and Kirk's, is that his reference star showed the same dips as the target, so that calibration was able to remove those cloud-caused wiggles well. His larger aperture was 10x the light gathering, and spead over many more pixels, gave a good signal. |
Zoomed in. |
