71 replies to this topic

[SeymoreStars]

I've seen the recommendation but I've never seen the result that demonstrates why. Now I wonder what steps in wbpp need to be changed to get the same results. I imagine rejection and stacking method are the big things, but there are lots of settings to play with.

Before using Tycho Tracker I only knew how to calibrate images with PixInsight (manually and with WBPP). Taking pvdv's direction I used Tycho Trackers calibration and found it to be excellent and will not be using pixInsight.

 

See my next post below for clarification.

[SeymoreStars]

Note: these images were taken (before using Tycho Tracker) to test the viability of the telecscope system.

 

All steps are performed within Tycho Tracker
Observatory XXX is  "active"

 

calibrate images straight from the camera - success
plates solve images - success
align images - success
load known objects - success
select object 2002 TH41  (mag 18.7)
add to track navigator - success
verify track  - success
note: object is viewable in all 3 frames
object designation (populated)  - success
add observations  - success
View with Published Observations  - success (see screenshot #1)
computer orbit - success (see screenshot #2)
generate report (MPC1992 format)  - success  (see screenshot #3)

 

note: this report was NOT submitted to MPC

 

Thank you pvdv for your guidance, much appreciated!

 

Steve

Attached Thumbnails

• 
• 
• 

Edited by SeymoreStars, 30 December 2024 - 12:23 PM.

[pvdv]

Essentially you have in this case (but also for exo-planets or anything that involves variability), to make sure the biases/darks/and flats are as plain/untouched as possible. Stacking (of the images) by pixinsight doesn't play a role here (and if it did, it would get even worse). I used here WBPP because it produces "calibrated" files along the way and that's how most people would use Pixinsight.

Edited by pvdv, 30 December 2024 - 12:35 PM.

[twoc]

Pixinsight, in its default configuration, is shockingly bad at preserving the interesting signal.

Same test field, same data set.

On the left, the result after a default wbpp calibration by pixinsight (correct bias, correct flat, correct dark). The shot_name_c is used. Tycho calibration is disabled.

 

On the right, uncalibrated frames with hot pixels removed and pseudo Tycho flat.

You have the names of the objects present in the field (by catalogue/orbit) and the confidence level of the Tycho detection.
 

People who discourage the use of Pixinsight do have a point.

Huh interesting, I'm surprised at the difference. What does a stack on one of the marginal mag 19.x asteroids looks like with PI vs Tycho pseudo flats? I usually do a standard calibration in Tycho with PI used only to generate the master flat / bias / dark. Perhaps the difference is related to the interpolation used or that the image normalization helps with detection (Tycho has a check box for this, wbpp output won't be normalized). Just guessing.

[555aaa]

I do all my calibration and alignment in ASTAP because that’s my normal workflow.

I was gonna ask about the orbit computation. The residual of interest is the residual of your observation set so I’m not sure what the screenshots above are showing. I’ve only used findOrb externally to check my residuals.


Lastly it’s probably worth having a discussion on how one verifies the time accuracy of their measurements.

[pvdv]

The report looks OK - it is a validation vs previous observations

JPL Horizons seems to agree

2024-Dec-27 02:00     03 19 06.58 +03 40 33.0   18.686    n.a.  1.35116687752456  12.7512670  131.3113 /T   20.2645   0.1325010   337.69521   80.367708         n.a.     n.a.
2024-Dec-27 03:00     03 19 06.37 +03 40 40.4   18.687    n.a.  1.35147386410605  12.7616144  131.2713 /T   20.2778   0.1325373   338.08926   80.370638         n.a.     n.a.

You can also invoke findorb from within tycho itself

 

quick test from dirty field previously posted
 

Edited by pvdv, 31 December 2024 - 09:53 PM.

[SeymoreStars]

pvdv Happy New Year!! Is your post above in reference to my observation?

 

If so I am not following you, my apologies.

 

Steve

[pvdv]

Hi Steve,

I wish you a happy and productive 2025
 

pvdv Happy New Year!! Is your post above in reference to my observation?

 

If so I am not following you, my apologies.

 

Steve

Partly - and double checking it with the JPL Horizons data (which would be the absolute reference for known bodies).

But also using find_orb, the external program 555aaa uses, from within Tycho on my (not so good) test data.

There are several way to check the data quality. Re-computing an orbit from all the previous observations, for example for objects that don't have good data yet, downloading orbit elements for known objects with good data, on-line with data download, off-line. Etc...

And, as 555aaa points out, that's not the only thing one has to worry about. Imprecise observatory position and imprecise timing may have a significant impact on one's measurements. This is especially important if you use a largish scope (because imprecise location will prevent it from reaching its accuracy potential) or image fast objects (because even wtih precise location, imprecise timing will increase uncertainty dramatically).

For time sync, the Dimension 4 software can be useful on Windows. Or you can use a GPS syncing dongle.

Lots of potential constraints as you can see. That is, by the way, the reason I operate as "247" and never submit measurements: I operate on a roving basis mostly from the garden of my country house, and I am not always sure I am set up properly for all the constraints. 
 

[Xilman]

And, as 555aaa points out, that's not the only thing one has to worry about. Imprecise observatory position and imprecise timing may have a significant impact on one's measurements. This is especially important if you use a largish scope (because imprecise location will prevent it from reaching its accuracy potential) or image fast objects (because even wtih precise location, imprecise timing will increase uncertainty dramatically).

Likely teaching many people here how to suck eggs but for the benefit of those who do not yet realise it (including future readers) Google Maps will give a position which is almost always accurate 20 metres and often good to 10m. This is easily accurate enough for the MPC. On my land there are two observatories which are about 15m apart. MPC told us to use the same code for both, J22 in this case.

[pvdv]

Yes, Google Maps seems to be the recommended standard these days. A GPS dongle is also useful as a timing source for occulations. 

But I am fundamentally a paranoid pessimist: I can't ever get rid of the feeling that, surely, I must have forgotten or done something wrong

[SeymoreStars]

When the skies clear next the goals are ...

 

Capture a NEO object Miwablock (mag 14.96) using Bin 2X2 and Bin 4X4 without over exposing.
Bin 2x2 resolution = 0.53" per pixel
Bin 4x4 resolution = 1.06" per pixel

 

Capture 2-3 image sessions as conditions allow. Evaluate the FOV for other asteroids with an eye towards satisfying conditions to apply for an MPC observatory code.

 

Comments and criticism welcome, as always your experience is appreciated!

 

Steve

[Xilman]

Good luck, especially with the weather!

[SeymoreStars]

Still testing with December 27th results, the next brightest object 1999 RP103 at mag 19.1, processing with Tycho Tracker was successful.

 

Reprt (MPC1992) screenshot attached.

 

And another from December 27th results, object 2002 RR185 at mag 19.8, processing with Tycho Tracker was successful.

 

I not certain that Tycho Tracker is "seeing" the objects in my images because they are very dim. I will try the dimmest object mag 23 which is beyond the capability of my system. I expect it to fail.

Attached Thumbnails

• 
• 

Edited by SeymoreStars, 01 January 2025 - 09:07 PM.

[SeymoreStars]

The second screenshot (above) for object 201269 is not successful, because the object is not tracked by the 3 stacked images. I need to learn a definitive method for validating my data, beyond  the "residuals below 1.50" mentioned in TychUserGuide.pdf Example#1.

 

Thanks

Steve

[555aaa]

Steve - I think your timespan is too short. 201269 was only moving about 0.16 arc seconds per minute when you imaged it so a seven minute arc is about an arc second which isn't enough. I would probably not report anything that didn't have at least a ten arc second track which for an object at that speed would be about an hour. So that does mean leaving and then revisiting the target if it's slow moving. If it's moving that slow you could use a five to ten minute single exposure and frankly this is where Astrometrica will do better IMHO because you can just take one ten minute exposure (or substack, stacked conventionally), go off to another target, come back a half hour later and take the second, go to another target, and then get the third and you will have an hour long arc and a nice set of points (you would just blink the set of three and the object should be pretty obvious).

 

As an example you can go to the NEOCP recent observations and it shows a yellow dot for each NEO Confirmation Page reported obs, live. This is an important graph showing you who is working what objects and which have no coverage (or were bogus). You usually see groups of dots that cover about an hour or so. When you mouse over each dot it shows the actual observation data in the 1992 MPC report format. 

 

Tycho works great with faster objects where within your stack the object has moved many pixels, many times the diameter of a star image. And most everything on the NEO confirmation list is fast moving. The slow moving objects are slow so there's no rush to cover them. The fast moving ones are the ones at risk of being lost or are misidentifications of other objects.

 

Some of the surveys do a revisit schedule where they do say 100 five second exposures on a given point of sky, then go off and do an adjacent sky patch, but then an hour later they do another 100 five second exposures on the same sky location, and then an hour later do a third revisit. This covers both slow and fast moving objects and you get three synthetic tracks per object per night, which gives high confidence that the synthetic track result is real.

 

https://minorplanetc...r.net/neocp_obs

 

 

[pvdv]

Hi Steve,

One thing you could do to revisit fields is set up a 3x3 or 4x4 mosaic and cycle through the panels. 

Imaging an area where you can find many objects and that is regularly covered by the major surveys such as Atlas/MrLemmon such as the currently well placed Taurus will give you plenty of material to hone your skills on. 
(and eventually identify what systemic error remains in your configuration). If possible confirm your data with T05, T08, G96 etc...

As far the NEOCP page objects are concerned, most of them are either faint or fast or both. Make sure to check the ephemerides of the objects before wasting time.

While they are obviously interesting and contributing to confirmations might be your ultimate goal, a lot of them are just beyond our apertures and skill levels.

As I said above, I am extremely cautious but imho, there is no shame in spending significant training and tuning time on easy objects. 

[SeymoreStars]

Thank you 555aaa and pvdv. I am very new to MPC and asteroid hunting. Your comments and suggestions are very welcome, even if I don't fully understand them at the moment. I have much to learn and expect struggle. Thanks for the guidance!

 

Steve

[SeymoreStars]

I am not close to compiling the data to submit for an MPC observatory code, but the following line from the requirements strikes me as vague.

 

"Three to five observations of each object from each night should be included. Do not report single positions per night; single positions will cause your entire submission to be rejected. Reporting more than five measurements in a single night is generally not useful for constraining an orbit."

 

I conclude 3 measurements per object would be good but do they want these measurement spaced far enough apart that they show significant movement?

 

If so what would be "significant movement" 1 arc minute?

 

I would like to approach this in an organized way and clearly understand the requirements going in.

 

 

Your help and experience is appreciated!

Thanks!!

Steve

Steve

[Random2310]

I conclude 3 measurements per object would be good but do they want these measurement spaced far enough apart that they show significant movement?

 

If so what would be "significant movement" 1 arc minute?

I am hoping someone with experience will chime in and give their take; The Minor Planet Center's Guide to Minor Body Astrometry has this to say:

 

How many observations should I make of each object?
As a general rule, when pursuing high precision astrometry, it is preferable to obtain small quantities of deep, high SNR data.

• The MPC typically recommends taking a few observations over a period of an hour or so per object, per night. Additional astrometric positions are typically not helpful for the determination of the orbit.
• Observations of specific objects are best made on pairs of nearby nights as the accuracy of isolated single-night observations can be difficult to judge. By observing on pairs of nights any ambiguity is removed.
• Please try to not make only one observations of each object per night. Without specific appropriate reasons, if a batch contains any single positions, the entire batch will not be accepted and it will be returned to the sender.
• Observations of a potentially new object in groups many hours apart on a single night can be useful in particular in the case of a newly-discovered object that may be close to the earth.However, we recognize that there might be cases in which more observations are needed, e.g. for photometric purposes, or for an object during a close encounter with the Earth. Even though the MPC always encourages the acquisition of high quality astrometry, we emphasize that it is not our place to discard large numbers of observations when they get sent to us. There will be cases in which objects are going to have hundreds of published astrometric and photometric measurements. While the MPC already de-weights these measurements for our orbit fits, it is up to the end user to decide what they want to do with them.

 

Edited by Random2310, 03 January 2025 - 11:36 PM.

[555aaa]

What they mean is that they want three to five astrometric measurements of the same asteroid. Forget about the synthetic tracking thing for a minute and just consider single images. You measure the asteroid position on the image. Then you wait an hour. Then you take another image. The asteroid has moved many arc seconds. You measure it. Then you wait another hour and it's moved again and you take another image. You're trying to show that you can provide an accurate track across the sky and your data is both low error and self-consistent. Tycho can't do this. It has to have a bunch of images because it only does synthetic tracking AFAIK.

 

When you use Tycho's synthetic tracking feature and you take say 20 images, it's producing ONE measurement. One position.  So when you are actually going to report astrometry you do what I said above and you come back to the same field after an hour or so and so another set of 20 images, do the synthetic stack of those, get another position, and then come back after an hour and do another 20. So your 60 images are gonna turn into three measured positions. It's going to use the time stamp in each image and knowledge of whether the timestamp is start or mid exposure to determine an equivalent time as if you took one image at that specific time.

 

The tool (Tycho) lets you artificially divide your stack down so that you get three "measurements" but for slow moving objects you need a LOT of exposures or really long because the average main belt asteroid moves so slowly.  This part of the tool I don't use.

[pvdv]

What they mean is that they want three to five astrometric measurements of the same asteroid. Forget about the synthetic tracking thing for a minute and just consider single images. You measure the asteroid position on the image. Then you wait an hour. Then you take another image. The asteroid has moved many arc seconds. You measure it. Then you wait another hour and it's moved again and you take another image. You're trying to show that you can provide an accurate track across the sky and your data is both low error and self-consistent. Tycho can't do this. It has to have a bunch of images because it only does synthetic tracking AFAIK.

Everything you said about the process is of course correct. But Tycho doesn't only do synthetic tracking. In fact, the Tycho manual starts with "conventional" examples with 4 images for known and unknown asteroids (at page 20).

I think Tycho's optimized implementation of synthetic tracking has been so spectacular that it eclipses all its other features.

 

Edited by pvdv, 04 January 2025 - 06:50 AM.

[SeymoreStars]

80 Minutes of movement of the asteroid Asteroid Minos (6239) through the background stars. Tracking the stars makes them appear to be stationary, which makes the asteroids movement stand out.

 

https://youtu.be/EqaX8AfzYeE

Edited by SeymoreStars, 06 January 2025 - 12:19 AM.