Dear Observers, We all too frequently receive error-filled submissions of astrometry here at the MPC, and I've decided to write a little "how-to" describing what common pitfalls can be avoided, and how to improve, in general, what folks submit to the MPC. First, read this: http://cfa-www.harvard.edu/iau/info/Astrometry.html These two documents overlap significantly. Please read both carefully. MOST COMMON PROBLEMS: 1) False detections due to hot pixels, bad pixels, poor flat-fielding, and internal reflections. This/these are the most common mistakes when submitting to the MPC. Professional observers still make this mistake. The key is to avoid lining up these objects and "creating" a real object. By far the easiest thing to do is simply DITHER THE TELESCOPE BETWEEN SUCCESSIVE IMAGES. This will eliminate 99% of all false detections immediately. At CSS, we had this problem, and we eventually settled on simply moving the telescope after each image. So... image 1 was at the expected coordinates. Image two was 30 arcsec NORTH of this location. Image 3 was 30 arcsec SOUTH of the expected coordinates, while image 4 was 30 arcsec SOUTH AND WEST of the expected coordinates. This simple procedure makes is nearly impossible for anything fake to line up with linear motion within a few arcsec tolerance! The second problem with false detections is using a bare minimum of images. I would never use 2 images for discovery purposes unless I had very small pixels (0.5" /pixel or so) and I had a good point-spread function for both detections. Three images is in my opinion the bare minimum for consideration, but I do not like this. USE MORE THAN 3 IMAGES, AND DITHER THE TELESCOPE. Another very common mistake is absolute bare-minimum time intervals between images. It is best to have at least 30 minutes worth of coverage on each and every object. Note that CSS, LINEAR, and Spacewatch all have t > 30 min. LINEAR averages around 70 minutes interval on each object. This gives a more robust Vaisala orbit, better linking probabilities on the subsequent nights, and lastly, prevents bad links by the observer and the MPC. These short-interval links are often misidentified, or worse, spurious objects! The MPC now requires t > 20 minutes for designations except in extreme circumstances. Keep in mind this is t > 20 min on EACH night. In addition, almost all false detections won't show linear motion over a 20 or 30 minute interval... most of them come from very short intervals. 2) Obviously bad astrometry It is surprising that this is the second most common problem, and perhaps the most annoying. We often get observations of objects that are so clearly wrong, a simple cursory examination would show this. For example, we DAILY receive reports of objects that are reversing direction of motion between 3 measurements, or worse, two good positions and one that is off by 5 arcmin or something. In this category is also bad links by the observer. This also happens on an almost daily basis. For example, an observer will go to the expected coordinates for an NEOCP object, find a bright object, measure it, and send it in as the NEOCP, only to find later this was a routine numbered asteroid. This is precisely why the NEOCP, and the MPC Ephemeris service gives you the SPEED AND DIRECTION of the object in question. I'm sure you'd be surprised how many routine MBAs moving 30" / hour are turned in where the observer thought this was an NEOCP object that was supposed to be moving 300" / hour. We also receive bad links where observers simply went to the MPChecker, found the object closest to their object, and pasted this designation in the observation string. This causes me no end of grief, because in most cases, it is simply easier to paste a new observer-assigned TEMPORARY designation on each object submitted. In this fashion all "new" objects are identified by our automatic software, and designations are e-mailed automatically. If the observer doesn't receive a designation, it usually means that the object is a new one. This also reduces e-mail traffic back-and-forth between parties, one thing that definitely helps me, since I get a few hundred e-mails per day :) 3) TIME PROBLEMS Of all things, this should *NEVER* be a problem, but is routinely, and perhaps a day-to-day error in observations sent to the MPC. IT IS AN IMPERATIVE TO CHECK YOUR CLOCK EACH AND EVERY TIME YOU OBSERVE. Badly timed images are not only an amateur problem. EVERY SINGLE professional survey has had some sort of timing problem, from bad local - UT time corrections, bad computer clock time, mistiming the exposure start and end, and messing up the midpoint of the exposure. Yes, these were all done in one form or another by professional surveys. Amateur observers, however, really excel in messing this up. We're constantly fixing or deleting poorly timed images. Please be careful here. There's no excuse for this one. And trust me, I'm talking from experience. CSS had a -12 second error on all images due to improper coding of the start time in the FITS header. And this is more or less my fault for not checking it... 4) JUNK ASTROMETRIC SOLUTIONS Another frequent problem is astrometry that is clearly the right object that also clearly does not fit. This arises in some cases for horrible or non-converged fits on astrometric solutions. If your RMS on your solution is extremely small and uses only 3 or 4 stars, it is probably wrong. Likewise, if it is over about 0.7", you've also probably got a problem. Keep a close eye on those solutions! 5) OVEROBSERVING BRIGHT OBJECTS (Not an astrometric problem, but something of note) A good chunk of objects need no astrometry whatsoever, and yet we still will receive literally THOUSANDS of observations each month of these objects. Bright numbered asteroids, unless they are occultation targets, radar targets, or mission targets, really don't need astrometry. So don't go out of your way to target them. You may, and should, measure them if they just happen to appear in your frames. But targeted astrometric observations of routine numbered MBAs by amateur facilities is almost certainly a waste of time. Likewise, many non-numbered NEOs are absolutely hammered by amateurs for no apparent reason. A good rule of thumb is to only observe objects that you and your system can actually help. So if the current ephemeris uncertainty is only 0.3", there is no possible way you can dramatically improve the orbit if your astrometry is only good to 0.5". Also, given the sky coverage and sheer number of professional surveys in action, it is very likely that your hard work will simply be obviated the next night by a survey. Notes regarding professional programs. At this point, it is probably important also for amateurs to know a thing or two about the professional programs. LINEAR takes FIVE (5) images of each field, total spacing about 1 hour. CSS, E12, and LONEOS all take FOUR (4) images of each field, with intervals varying from 20 - 60 minutes from first to last image. SPACEWATCH and NEAT take only 3 images, spaced by 20-60 minutes, but they both have SMALL ( ~ 1 arsec) PIXELS. Each and every of the aforementioned programs submits *all* objects as new objects. Meaning, each object observed on a given night has its own, observer-assigned unique temporary designation. These observations pass flawlessly through our automatic processing code, and the remaining one-night objects that might be NEOs are left for further examination. Given that a good deal of MPC effort has been put forth to process the bulk of the data this way, other observers should consider operating in a like manner. Tim Spahr Near Earth Object Technical Specialist at the MPC Former observer with the Catalina Sky Survey Addendum 1 (Dec 20, 2004): Please cease and desist from sending single positions of any object on a given night, unless this object is **CRITICALLY** important, and then only do so with gratuitous comments regarding the accuracy of the measure, and why no other measures were obtained. For example, this might be OK for NEOs and comets from skilled observers. But single, isolated positions for MBAs are very likely to be deleted.