Support

TUESDAY, DECEMBER 4, 2012 AT 6:04AM

Discovering Asteroids at iTelescope.Net

<---- Part 4 | Part 6 --->

Why Can’t We Image Yet?

If you have read the previous four articles, you will be familiar with the asteroid discovery rules, the major surveys, how to select an itelescope.net scope and where to point it. At first sight there is nothing to stop us imaging the night sky, finding asteroids, measuring their positions and reporting the results to the Minor Planet Center (MPC). 

Unfortunately there is a very real problem and it concerns the “finding asteroids” part of the process. Asteroids appear as moving objects relative to a fixed background of stars but as you will see although all asteroids are moving objects not all moving objects are asteroids. 

Some of the non-asteroidal moving objects, or artifacts, are so different in appearance from the true thing that we can immediately see them for what they are. However others look like asteroids, appear to move like asteroids and indeed have motions so similar to those of true asteroids that if we report their positions to the MPC, the MPC will believe us and add the artifact to their database. 

If that is not bad enough some artifacts can mimic the motion of Near Earth Asteroids (NEO’s) and if we report one and the MPC believe us, they will add it to their NEO Confirmation Page. 

This lists all the suspect NEO’s that have been reported in the last few days and is used by the major surveys when selecting objects for follow-up measurements. We are now in the situation where we are responsible for professional observatories wasting time looking for an object which does not actually exist.

Clearly this is something we really do not want to do and this is why I will show you how to avoid it happening.

Before I go into details, let me define some terms and introduce you to some software we will be using:- 

Pointing

When you use one of the itelescope.net scopes, you tell it the centre point of the field of view that you want it to image and you do this in terms of RA and Dec coordinates. The telescope then slews to that general area, takes a location image, compares it with the star map stored in its memory, calculates exactly where it is pointing and how it needs to move to point to the position you have requested. 

Tracking

Once the telescope has been pointed correctly it has to be moved continuously and precisely in a way that compensates for the rotation of the Earth. This is the job of the telescope mount. In the case of some mounts, for example the one used on T11, the motors and gearing are so well constructed and programmed that once the scope has been pointed it will remain on target for up to 10 minutes without any further input being required. This process of correcting for the Earth’s rotation is termed tracking. 

Other mounts are not so accurate and require some additional information input in order to remain on target. This information input and position correction process is termed guiding. itelescope.net provides information for each scope regarding the maximum time that you can image without guiding. 

Guiding

The guiding process involves the use of software which selects a suitable star as close as convenient to the target centre then measures its position at the start of the exposure and every few seconds during the exposure period. If the tracking is inaccurate and the star appears to drift out of position, the computer will calculate the distance and the direction that the scope will have to move in order to correct for this drift. This information is then fed to the mount which carries out the necessary correction. 

Reducing

This process involves the use of software that can select a number of stars in an image and then obtain their RA-Dec coordinates using a suitable stellar library. Using the positions of these reference stars it is possible to calculate the RA Dec coordinates of any point within the image. 

Aligning

This software operation is a process where a group of reduced images with the same field of view are shifted laterally and rotated until each star in each image is lined up with the corresponding star in all the other images. For example if we had a set of such images on clear acetate sheets we could hold the entire set up to the light and manipulate each sheet until we got a perfect alignment.

The alignment process is a necessary first step prior to both stacking and blinking. 

Stacking

Discovering asteroids inevitable requires us to detect faint moving objects. In order to do this we need to collect as many photons as possible. For non-moving objects like stars the longer the exposure time the more photons we collect. The problem with moving objects is that if we use too long an exposure period the image start to trail, in other words it will change from an essentially circular spot to a more elliptical shape and then into a line or trail which marks the motion of the object across the sky. 

One way around this problem is to take numerous shorter-exposure images in each of which the moving object appears as a circular spot in a slightly different position. We then use suitable software to superimpose these images one on top of the other. If we superimpose them exactly we would end up with circular stars and a trailed image of the moving object. However by some cunning electronic trickery we can displace each image slightly and end up with trailed stars and a circular image of the moving object. The point of doing is that effectively we produce a circular image of the moving object over a longer exposure period. In this way we concentrate the maximum number of photons on the minimum number of pixels and get a brighter image which is easier to detect. 

Blinking

I normally take 15 images each with an exposure time of five minutes and then stack them as 3 sets of 5 images. Each set therefore has an exposure time of 25 minutes. Using suitable software I then display each set one after the other repeatedly and look for objects that appear to move. This process is termed blinking and is equally applicable to individual unstacked images. 

Now that we have defined terms we need to consider the software required to carry out each task. The good news is that we don’t have to worry about pointing, tracking, and guiding as these are all carried out by itelescope.net as part of the service. The remaining tasks are our responsibility and for these I recommend Astrometrica. 

You can get detailsof this software here: http://www.astrometrica.at/ There is also a user group. 

As you will see this is an extremely versatile tool and in addition to reduction, alignment, stacking and blinking it also enables us to measure the position of asteroids and to email the results to the MPC in their required format. 

It’s the Real Thing

Before we start looking at artifacts pretending to be asteroids here are an images of a real one:- 

This is asteroid 34997 at magnitude 17.9 imaged from Nerpio using T17. The animation is produced by blinking three stacked images in sequence repeatedly. On of the many useful features of Astrometrica is the option to display a labelled location box around all asteroids for which the MPC has calculated an orbit. 

How about This? 

At first sight it looks believable: a moving object with magnitude 18.5, travelling in more or less a straight line and at a speed typical of asteroids in this region. The absence of a location box means that the MPC has not calculated an orbit. If we searched using the MPC Minor Planet Checker for objects that MPC knows about but has not yet calculated an orbit we would draw a blank. At this point we might be thinking in terms of a potential new discovery but before we report it to MPC let’s just stop and check. 

One good initial check we can carry out involves the use of Find_Orb.

Find_Orb is a piece of software which can process the position measurements that we make using Astrometrica and find the best orbit that fits the data. Like Astrometrica there is a user group. 

As well as generating an orbit, Find_Orb will also give us an indication of how accurate our measurements are. If we take asteroid 34997 as an example we have used three images to produce the animation. If we measure the RA and Dec position of the asteroid in each image and process this data using Find_Orb it will calculate the residuals for each of the six data points. It is rather like drawing the best straight line through a set of points on a graph and then measuring how far away each point is from the line. The residuals are expressed in arcseconds and as a general guide we should be able to produce measurements with residuals less than 1.00 arcseconds and ideally less than 0.50. If any of the residuals are greater than or equal to 1.5 arcseconds the MPC will not regard them as good enough for use when they try to link observations made on different nights. 

Processing the coordinates from 34997 gave the following results: 

These small residuals show that our measurements were pretty accurate although to be fair it is a bright object which makes it easier to get good results. 

Now let’s check the residuals from nfxx2: 

Admittedly the residuals are not as good as those from 34997 but I am reasonably certain that if we reported them they would be accepted by the MPC. 

In fact nfxx2 is not an asteroid; it is a small cluster of defective pixels (hot pixels) which act as if they are detecting photons even when they are not illuminated. The result is that each image taken with this camera will show this small spot of light in the same position. 

At first sight it is not easy to see why a defect at a fixed location on the camera chip should give rise to an apparently moving image. The answer lies in the image alignment process. 

The Alignment Illusion

This illusion can be created using either stacked or unstacked images and is a result of tiny errors in pointing between successive images. I have deliberately used unstacked images in the following animations to prove to you that the illusion is created by alignment prior to blinking. Had I used stacked images they would have also shown the illusion but it would have been created by the alignment prior to blinking process and not simply by stacking. 

The image is an animated screen print of what you see when you blink three images without first aligning them. In this mode the CCD frame appears virtually motionless. During the 100-minute imaging session the scope has drifted slightly off-target both downwards and to the left and as a consequence we see stars appearing to move upwards and to the right In the second image only we see a line of illuminated pixels appear close to the left-hand edge of the image. This artifact is a cosmic ray strike. 

The object labelled nfxx3 is a hot pixel which appears in all three images. It occupies a fixed position on the CCD chip and like the CCD frame appears virtually motionless. 

It is possible, using Astrometrica, to align these three images so that the stars are superimposed on each other. This animation shows the result. 

The alignment process involves moving the images to compensate for the off-target drift. This results in the apparent motion of the CCD frame and of course the nfxx3 hot pixel. In this animation the illusion is revealed because we can see the hot pixel moving with the frame but if the frame was not in view then all we would see is nfxx3 appearing the move relative to the stars just like an asteroid. 

Aren’t Hot Pixels Easily Recognised?

I sense that the more experienced among you will feel that nfxx3 looks nothing like an asteroid and that you would never mistake it for one. That is a valid point and I chose it for demonstration purposes simply because it happened to be conveniently located at the corner of the image where you can see both the object and the frame. During the past decade I’ve seen my fair share of hot pixels and although the majority of them can be recognised by their small size, abnormal brightness and generally blocky appearance there are some that are quite good asteroid mimics. 

Object nfxx4 is more difficult to distinguish from an asteroid. It is in fact a larger than average cluster of hot pixels and in the bottom right-hand corner of the image you can see a more recognisable hot pixel moving in step with it. 

One way to avoid confusing artifacts with real objects is a process known as dithering. 

Start Dithering

I use the term dithering to describe multiple re-targeting during an imaging session in order to reduce the possibility that any drift off-target with time can mimic the motion of an asteroid. 

Those of you who have used ACP Observatory Control Software may know that one of the directives is #DITHER. This enables you to shift the target at the beginning of each exposure by a random number of pixels in both RA and Dec. I have tried this but found that there is a disadvantage when using it to make relatively large shifts when a telescope is also being guided. Effectively what you are doing is asking the scope to lock on to a guide star but then at the beginning of each exposure giving the mount a crafty kick. Another disadvantage is the random nature of each shift which brings with it the possibility that the shifted positions may result in a linear drift. 

I find that a better way is to actually re-target at about one-third and two-thirds of the way into the imaging session and to displace the RA by 10 seconds of time and the Declination by 30 seconds of arc. I do this in such a way the image centres cannot form a straight line and consequently any defect on the chip will appear to move in a non-linear fashion. 

This is the pattern I use when taking 15 individual exposures 

This image shows the positions of the example coordinates plotted out using planetarium software. As you can see the 1 arcsecond diameter circles do not lie in a straight line and the apparent motion of any hot pixels when blinking the three images will be correspondingly non-linear. 

Just Checking

If you do discover what you believe to be a Near Earth Object and if you do report it to MPC, you may get an email from them asking if you are absolutely certain that your measurements have been made on a real object. I speak from experience when I say that then is not the time to start checking. 

In such a situation before I report anything, I first locate some hot pixels and check that their motion is non-linear and/or distinctly different from that of my NEO suspect. If I am still in doubt I then re-measure the position using un-stacked single images. I have found that although three stacked images can (as we saw with nfxx2 and nfxx4) give rise to apparent linear motion, a check involving 15 images is very likely to reveal signs of irregular motion with an artifact. 

If after all these checks I cannot be completely certain that I am looking at a real object then I do not report it. I either call it a day or I use the measures I have obtained to predict its position the next night and then try and image it again. If when I check where it should be I find myself looking at a patch of empty sky then I know it was an artifact. However if I find it and the motion and the magnitude match my prediction then I know it is real. I can then report two nights and wait, fully prepared, for any MPC reality check. 

Other Artifacts

1. Random Noise 

If I detect what appears to be a very faint moving object i.e. one that is only just within the limit of detection, I have to consider the possibility that it could be a chance alignment of random noise. This animation shows an example of such a case. 

As you can see the pixels representing the star-free sky vary in brightness from image to image. This variation is due partly to variations in the very low levels at light pollution at Mayhill and partly due to electronic effects which cause individual pixels to produce photoelectrons even when they are not impacted by photons. 

These random effects can give rise to small clusters of pixels that are brighter than those that surround them. If for example you blink three images, you can get the situation where each image has such a cluster and where the clusters lie more or less on a straight line. 

In this particular example when I checked the residuals using Find_Orb as described above they were not particularly good and when I re-stacked the images using a different speed and angle the “object” was no longer detectable. 

This re-stacking test is a good way of distinguishing between real faint moving objects and random noise clusters. In the case of single images I find that although you may see chance alignment of random noise when you blink three images, the illusion is very unlikely to persist once you increase the number of images that you blink. 

2. High Energy Radiation

The cosmic ray strike that I pointed out earlier in this article is but one of many examples of how CCDs can detect things other than visible light photons. 

This publication http://snap.lbl.gov/ccdweb/ccdrad_talk_spie02.pdf gives further details. In my experience you see examples of these in most images but the number increases significantly at time of high solar activity. The only ones that resemble asteroids take the form of circular spots that appear in individual images. Fortunately these spots do not occur frequently enough to give any realistic possibility of a chance alignment being mistaken for a real object. 

What Next?

In my next article I plan to detail the process of astrometry using unstacked images.