After capturing a few shots of the recent flyby of asteroid 2015 TB145, I thought I would look into software tools for astrometry to determine the location of the object and confirm it was in fact TB145 in my images. This analysis is used to discover, confirm and observe new asteroids and comets and to produce observation reports.
The image processing and planetarium packages I own do support astrometry, but only at higher license levels than I purchased. So I looked up a program I had used quite some years ago called Astrometrica, and was very happy to see it was still available – and very much supported!
Astrometrica is a shareware package created specifically for analysis and reporting of solar system objects, written and maintained by Herbert Raab from Austria. You can download and try out the package for free for a generous 100 days, and it is well worth the purchase price of 25 Euro.
The images from the session taken early in the morning on Halloween were acquired using Maxim DL and saved as FITS files which can be read right into Astrometrica. First, though, it is vital to enter the image parameters as Program Settings in order to correctly interpret and fit the images. The first settings are for Location which are key for reporting observations (my actual values changed somewhat):
The next set on the CCD tab is critical as it defines the scale and orientation of the image. These are used to convert the location of detected stars and objects to the coordinates of the star catalog used:
The focal length is that of the lens or telescope used in the imaging. I have entered the nominal value of 750mm +/- 5% for my 6″ Netwonian. After calibrating an image this could be set to a more accurate value and a lower variation for a faster fit, but this should work for a first pass. I’ve also set the Pointing error quite high, as I know the object is somewhere in the frame but don’t know the image center location very accurately. So we’re asking it to do a pretty wide search.
The CCD chip is 5.2 x 5.2 microns for the Canon DSLR, with no binning. Another key factor is the Position Angle that specifies the orientation of the camera on the telescope. Note that this can depend on whether the mount has flipped or not. Also, the image may need Horizonal or Vertical flipping depending on the type of scope used and attachments. I spent quite a long time trying to work these out but could not get a fit at all with the settings I was expecting.
Then I remembered I had taken some shots of the Moon earlier in the session. Looking back at those, Tycho is on the left hand side of the images, and rotating 90 degrees counter-clockwise puts that and the other features in the “right” place with North at the top. As is, the image has East at the top, so the Position Angle is set to 90. All of the images could be rotated to have North at the top, but it’s probably best to leave them as is and have the program work out the orientation of reference stars. I was expecting that the image should also be flipped but it looks like that is not needed, so the Flip check marks are off. Perhaps the camera takes care of mirror flipping the image, since it would need to do so for the camera lens as well.
There are other settings to get the correct time from the file, and the remaining tabs have other parameters to adjust for fitting, but these seem to be OK as is. So next I loaded in the set of 3 images from the second location used in the session.
The program has a great blinking tool, and that can be run without all of the setup details above. Simply select the Blinking Tool icon or menu item and this will extract objects and then align and display an animation of all of the select images in a new window. You can set the blinking rate, and stop and step through to take a closer look at anything moving. The images in this first set show a clear streak moving across a fair portion of the display towards the right hand side of the section below:
But to determine the actual coordinates of the streaker, we need to run the Data Reduction tool to fit the star field. From the Astrometry menu, we select Data Reduction and are prompted for a location. We can enter the known or estimated center of the image, or have the tool look up a given object and estimate the position for us. So we enter 2015 TB145 (after loading the MPCOrb database) like so:
The location could be off a ways for our quickly moving neighbor, and it is not at the center of the screen, but we’ve asked for a pretty wide search in the settings. Now we select OK and Astrometrica will run the fit. To do so, it needs a detailed star catalog. These can be downloaded and installed locally, but the tool can now fetch stars for a given region using an Internet service from a configured source – provided we are on line!
After that works its magic for a while we see:
This result indicates a poor fit, but we do have over 50 reference stars aligned. If we are way off in the settings, we would only see one match or a handful which would indicate we are probably lost! In this data set, the images have a lot of distortion and are not calibrated for brightness, so the fitting is pretty rough. If we accept the automatic fit we get a summary table:
Not sure how to interpret all of this, but it looks like we can determine locations in the images to about 1/3 of a minute in RA and Dec which seems pretty good to me! So now we can select any point and see its estimated position. If we select a star for example:
This object was identified as a potential star found in 2 or more images but was not included in the fit, presumably because it did not meet the entered fit parameters. But this is not the asteroid – we want to find the streak on each image and mark it’s position at the center of the streak like so:
Next we look up known objects in the area by selecting the button next to Object Designation and see:
We select the one known object in the vicinity, enter an optional note and then Accept to generate the observation for the selected object. (Some of the sets also indicated a nearby bright asteroid but that appeared to be just off of the frame). After selecting the Known Object K15TE5B which seems to be the official designation of TB145, it is labelled on the screen:
After doing this for the 3 images in the set, we can view the results, formatted for submission to the Minor Planet Center. To ACTUALLY submit the observations, I would have to be vetted and assigned an Observatory Code, but we can see just what the submission would look like:
COM Long. 72 07 24.2 W, Lat. 41 07 24.2 N, Alt. 30m
ACK MPCReport file updated 2015.11.16 21:41:42
K15TE5B C2015 10 31.39037 05 19 38.34 +18 56 12.5 12.1 V XXX
K15TE5B C2015 10 31.39209 05 19 57.11 +19 02 17.4 11.8 V XXX
K15TE5B C2015 10 31.39337 05 20 10.47 +19 06 39.7 12.1 V XXX
—– end —–
This shows 3 observations for K15TE5B or 2015 TB145 on Oct 31 with the time expressed as a fraction of a Julian Day, followed by the RA and Dec values and the estimated brightness or magnitude around 12. The default Observatory Code XXX is used for this example.
The other sets were fitted in a similar way, resulting in 16 observations of the asteroid over the session.
The positions determined from the fit were a ways off from the estimates provided by the program, so how do we confirm this was in fact TB145? The Minor Planet Center site has a number of tools to check results.
The MP Checker takes a time and location and reports any known objects in the area. We can enter the details on the form including a limiting magnitude and get
This shows our expected friend TB145. Note that if we increase the limiting mag to 20 the service returns a couple of dozen minor planets in this limited area – so there’s always a lot of them up there in the sky!
So it looks like we did find it! But how close are the positions we determined from the fit? One way to check is to compare the observed position with the result from the MPC Ephemeris service page. We can enter a list of objects and desired time and get calculated positions based on the current orbital parameters. The MPC presumably received lots of accurate observations from the flyby, so we can assume the position should be pretty well known. After getting a list of positions at each minute over the session, here is a comparison of the estimated positions vs my observed position in the first set of 3 images:
The Observed positions (Obs above) are compared to calculated positions on the surrounding minutes. The coordinates are all between the bracketing positions, so this looks fairly good! At a glance, there is considerable variation between the observed results and the interpolated positions, so there is definitely a fair amount of error in my results. But this looks quite good for something moving so fast and close by!
So next I think I will try this out on some of the brighter and slower Main Belt minor planets, and further explore use of Astrometrica.