Image Process Flow

The process of finding and recovering an FMO is time critical. To aid the reviewer in understanding the process along with its challenges we have created a rather lengthy discussion of our system design and the flow of our data reduction process.

Using our interface, an FMO reviewer watches for new images to appear, reviews the images for FMOs, submits candidates for auto-feedback (which should be received within 10 minutes of submission) and then resubmits candidates that pass the feedback layer to the observer. This process is unfortunately time-critical in that the observer may not see the final submission for hours if it is received later than when they shut down the telescope.

Ideally, the resubmissions for observer review should occur during the observing night - before the observer closes the telescope and leaves to sleep. To facilitate this 'real time review' system we provide as much timing information as feasible. A description of the timing information is offered in the second section of this page, but first we discuss the goals of the project and the challenges of recovering FMOs.


The Spacewatch Mosaic telescope's purpose is to maximize sky coverage while detecting moving objects. To detect a moving object we need three visits to the same sky location - in three time separated images, a moving object supplies three different positions that define a straight line. The time between visits needs to be of a length that allows objects to move a large enough distance to be discernible without moving so far that the software has difficulty grouping the three positions together. Another time element of FMO work is the exposure time, set by the need to collect enough light to see faint objects while avoiding saturation of bright stars. A longer exposure time means less sky covered but a shorter exposure time means we cannot catch smaller or more distant asteroids. Our detection limit with the Mosaic telescope reaches approximately 21st magnitude in the equivalent of the Visual (V) bandpass using a 120 second exposure for each of the three visits (passes). Longer exposure times are not as ideal for FMOs since the longer the FMO trail 1) the lower the likelihood of getting three full positions from the discovery images (the object will move out of range) and 2) the more uncertain the trail tips due to lightcurve effects. With the current system, the three visits are made at approximately 40 minute intervals producing an eighty minute observational arc for any moving object found in all three passes. This interval is determined by asteroid detection goals, not to maximize FMO detections.

To meet the 40 minute separation criteria, the Mosaic's survey plan is organized into 'Regions'. Each region typically has 7 'Centers' where a center is a location for a single exposure. Each center is visited 3 times - hence 21 exposures (of 120 seconds each + read/move time) completes a typical region. As the mosaic images are quite large (1.8 degrees by 1.8 degrees), the images must be split to make a suitable size for the FMO Project review tool. The logical split is by amplifiers - the mosaic images are produced by eight amplifiers and these amplifiers are cut in half - producing 16 FMO Project jpegs per single exposure. The name of the jpeg as seen in the FMO Project review tool reflects each image's origin (see picture below). Note that FMOs found in 8.1 and 3.1 are at a disadvantage due to the missing corners. FMOs found in 7.1, 2.1, 1.1 and 4.1 have the best chance of producing three positions in the same center but FMOs found on the edges may appear in other region centers (FMO Project discovery SW409R is an example - center 1 of a region produced 2 positions and the third position for the same FMO was supplied by center 3 of the same region, an exposure of the neighboring sky.)

2003. 12. 14. 63. 08. 5. 0.jpg is of the form 
Year. Month. Day. Region. Center. Amplifier. Half.jpg 
This is the first half of the fifth amplifier of 
the eighth center of region 63 as taken on 2003 Dec 14th.

Below is a single exposure of a region center broken down to amplifiers. The numbers on the image are the last two digits of image name.

Example of full mosaic image 

To find objects, the three visits of a particular center are processed. It is during this processing step that the FMO Project's jpegs are produced and shipped to the FMO Project's server. The timing of this shipment is entirely set by the observer - it is possible the observer may wish to 'retake' one of the three images so the observer must have complete control over the processing step. One observer may 'process' after each center of the third pass over the region, causing the images to download to the FMO Project in 7 short bursts of 16 jpegs. Another observer (or the same observer on a later region) may be too busy for individual processing requests and may wait until the entire region is complete. This would send down all the jpegs in one long stream (typically 112 images). The time interval between new jpegs is unpredictable except for the limitation that, if the telescope is still open and taking images of suitable quality, the start of one region should be at most 2 hours after the start of the previous region. If the time of the 'Last Download' is older than 2 hours, look to the weather pages or sky clock for evidence of poor observing conditions.

On the longest winter night, on observer can produce 6 regions. Using the above information, we calculate that 7 regions = 49 centers = 147 exposures = 2352 jpegs. But as seen from the Statistics page, we produce at most 800 jpegs per night. Why? The reason is efficiency. A FMO observation is nearly useless unless we have three positions, thus finding an FMO on the edge of center 10 in the third pass is going to cause excitement for nothing. We therefore only ship the jpegs associated with pass 2 - if the FMO was not moving off the CCD in pass 1 or moving into the CCD on pass 3 (single measurement producing events of no value) then the object should appear in pass 2 - giving us the best chance to find the object in either pass 1 or pass 3, preferably both.

To reiterate a very important point, three positions for a new FMO are necessary for the Minor Planet Center (MPC) to determine if a FMO's earth relative speed is accelerating or decelerating. When measurements on a new object are mailed to the MPC, they fit possible orbits to the three positions to produce an ephemeris to facilitate recovery by any interested parties via NEO Confirmation Page (referred to as the CP). For FMOs, the timing of this recovery is critical. Ideally, Spacewatch should have all pass 2 images reviewed in near real time, allowing for the recovery by our followup telescope, the 1.8m, that same night. This produces a longer arc and reduces the uncertainty of the FMO's course through the sky the next night, greatly improving the changes of recovery. If the object has dropped below our horizon before we receive notice of the discovery from our FMO Project reviewers, we may still have a chance for telescopes on the other side of the globe to recover the object...if we mail the object to the MPC in time. If an FMO is not recovered within 24 hours, an FMO becomes too uncertain and is considered 'lost'- hence images 'expire' by 22:00 UT of the image date, giving our own observers time to evaluate potential FMO candidates, submit them to the MPC and receive the MPC's ephemeris PRIOR to starting the next night's observing. If the FMO trail is long, the timing is even more critical - these objects are moving so fast relative to earth that a large number of orbits with rapidly diverging behavior can fit the observations - if these objects are not recovered within several hours of discovery, there is no hope of recovery and therefore no hope of an MPC designation.

The ultimate goal of the FMO Project is to produce an MPC designation for a new FMO, or MPECs for recoveries of known FMOs (typically discovered a night or two prior.) Recovery of a known object theoretically requires only one position but two has a better chance of being accepted by the MPC (this was the cause of SW409M's failure to achieve an MPC designation, the first discovered/recovered FMO by FMO Project reviewers). Why? FMO measurements are error-prone due to speed and light-curve effects making the trail faint and endpoints uncertain. A measurement for an FMO typically does not consist of the two endpoints but instead consists of the estimated middle position. The use of an average minimizes the endpoint error. Thus, three trails (giving three positions) of the same object are needed for the MPC to post an FMO on the Confirmation Page (CP), the first step on the path to our goal of an MPC designation.

After that lengthy discussion, we need to remind reviewers that FMOs are relatively rare. Much patience and time is required to find a potentially reportable FMO and after finding one, conditions need to be just right for a successful recovery the following night - allowing the object to be designated by the MPC. A reviewer may review several thousand sub-images before finding an FMO, just to be disappointed due to bad conditions preventing a recovery the following night. With that warning, we hope you have the necessary patience and wish you success.

The summary below is designed to give reviewers a sense of the timing for a given night.

Summary of Observing Night:

  1. Pointing and Focus
  2. Begin pass 1 of a region (centers 1-7)
  3. Begin pass 2 of a region (centers 1-7)
  4. Begin pass 3 of a region (first center)
  5. Optional step: start processing after first center exposed - results in 16 jpegs being sent to FMO Project server
  6. Finish exposing last pass of region (all 7 centers)
  7. Complete any unfinished processing - results in remaining jpegs being sent to FMO Project server
  8. Start reviews of automatically detected objects - this same tool requests observer feedback on FMO Project reviewer resubmissions
  9. Repeat steps 2 through 9 for remainder of night
  10. Finish all reviews and go to bed
  11. Return to telescope and run any accumulated FMO Project reviews before preparing for next night's observations
  12. All unreviewed images expire at 22 UT of the date of observations - the images are too old for successful FMO recoveries

The observer begins the telescope startup process around nautical twilight, preparing the telescope then adjusting the pointing and focus. This is the 'expected open time' - each observer has different habits so the timing is not precise, it is merely an estimate. An added complication is that an observer may begin the startup sequence and discover that conditions are too poor to begin real observations - hence, if our page indicates that the telescope opened at 1:30 UT and does NOT indicate that the first pass was started within an hour of this time then the telescope didn't go into normal operating mode due to the conditions or, infrequently, a mechanical problem. Ideally, the telescope will begin acquiring the first pass of the first region around 20 minutes prior to astronomical twilight.

DOWN FOR FULL MOON - Spacewatch is an asteroid survey group that specializes in smaller and more distant potentially hazardous asteroids. As a result, our telescopes and detectors are more sensitive to the effects of moonlight and we shut down for about four days on either side of the full moon each lunation.