NEO surveys

A comparative study of current and planned NEO surveys
Alain Maury - January 2006 version

There are currently several operating Near Earth Objects (NEOs) surveys. The goal is to perform a complete, or as complete as possible, inventory of Earth Crossing objects (asteroids and comets), in order to detect potentially hazardous objects a long time in advance. Such surveys will very likely find no objects which threatens us in the coming century, but in the very unlikely case where this would happen, would provide substantial time to undertake mitigation actions.

Some history :
The first Near Earth Object was discovered visually more than 100 years ago. All Near Earth Objects have been discovered by hotography first, then using CCD detectors.
Many groups are currently engaged in asteroids surveys, most of the efficient ones being funded by NASA. It is expected that in the future, these groups are going to improve their detections systems in order to reach smaller objects.

Survey efficiency :
It is interesting to study the efficiency of each system being used or planned in order to see where improvments can be made.
It seems relatively obvious that a survey system will be efficient if it uses a wide field of view, a very efficient detector, is able to observe many nights per month, while loosing as little time as possible not collecting photon (reading out the CCD detector for example). A commonly accepted figure of merit for a survey quoted many times in the litterature, is : FOM = W . D². DQE . Eff (1) Where :
W = Solid angle seen by the detector on the sky (in square degrees)
D² = Square of the diameter of the telescope (in meters)
DQE = Detective Quantum Efficiency of the detector
Eff. = Observationnal efficiency of the detection (ratio of the time spent collecting photons versus total observing time)

Short preliminary discussion :

For a given telescope, W is a measure of the number of square millimeters of detector in the focal plane. If a is the area of the detector, one can write

a = F^{2 .}W (if the angle can be considered as small). It is therefore possible to rewrite formula 1 as : FOM = a . D². DQE . Eff / F² = DQE . Eff . a / (F/D)^{2
(2)} This clearly shows the importance of the focal ratio F/D of the telescope being used.

D is usually fixed, and increasing its value usually is very costly.
DQE depends of the detector. Since detection is usually performed using unfiltered detector, it can be taken from 2 to 4% for photographic plates, to around 35% for thick front illuminated CCDs, to 85% for thinned back illuminated CCDs. While photographic surveys are basically obsolete, the choice is with either relatively inexpensive but less efficient CCDs to a very efficient but much more expensive devices.
Eff. could be taken in a wide variety of ways. It could be taken as the number of seconds spent effectively collecting photons in a year, divided by 31536000 (365x24x60x60). It would include the fact that we have to close our domes during the day, during cloudy nights, during nights where the full moon is up in the sky, that there are some nights which are or not allocated to NEO survey on some telescopes, and finally that using some observing modes, one must readout the CCD, and during this time, one clearly is not collecting photons. In practice, some of these parameters are very difficult to avoid. One can optimize by placing the telescope in a very good site. Very often one is left with an already existing telescope, or whatever telescope time is given to him/her by a telescope allocation committee. In order to compare existing and future system, we took Eff. as the ration of the time collecting photon versus total exposure time. I.e. a measure of the CCD controller efficiency, ignoring the other parameters.

3 operating modes are possible using today's CCDs :

Stare mode, where an exposure is taken, the shutter closed, and the CCD readout during a given time. On "normal" telescopes, great care is taken to have a very low readout noise. In survey telescopes, which are usually very fast telescopes, used without filters, the sky background is the dominant source of noise. It is therefore much more efficient to have a fast readout CCD controller.
Frame Transfert mode : In order to avoid loosing time reading out the CCD, some CCDs, called transfert frame CCDs can be used. They contain a storage zone to which the image is very quickly transferred (in a few milliseconds). The next exposure can be started, while the storage area is readout. No time reading out is lost.
Scan mode : In this mode, the shutter is left open, and the CCD readout is performed while the integration is being done. In the sidereal tracking mode, the telescope is simply idle (no tracking), and the stars are drifting across the CCD. If the CCD is rotated so that its line are parallel to the motion of the stars, and if the electric charges are moved synchronously with the star image motion across the CCD, a band of the sky is recorded, at a rate of 15 degrees per hour times the height of the field of view. While with Spacewatch, scan mode was the first operating mode, now, it is seldomly used in asteroid surveys.

While formula 1 does not take exposure time and pixel size into account, it is interesting to compare the raw values of current existing systems. However, it worth noting that a large scale will usually produce relatively poor astrometry, and small limiting magnitudes. It will allow a very wide field to be surveyed. There is a clear compromise between the number of square degrees being covered and the quality of the measurements and the depth of the survey. While the completude of the current inventory is in the order of magnitude 17 reaching 18, it is possible to trade limiting magnitude for square degrees. In the future, this will not be possible anymore, in fact there will come a time when asteroid survey telescope will have to be large telescopes.
I have not pushed further yet the adequacy of the survey system to the current distribution of unknown asteroids, nor have I taken care to remove trailing losses effects while detecting fast moving objects.

Performances of currently existing NEO survey telescopes using a typical factor of merit :

Name	Diameter (mm)	Focal length (mm)	Focal ratio	CCD size	Pixel size (um)	detector size (mm2)	Scale ("/pix)	Angle (square°)	DQE (%)	mode	Eff.	Exp. time (sec.)	hourly coverage (square °)	Limit. Mag.	FOM
CSS	680	1234	1 .8	4096x4096	15	62x62=3775	2.5	8.14	0.8	stare	0.75	60	366.2	20	2.25
Mt Lemmon	1500	3000	2	4096x4096	15	62x62=3775	1.03	1.38	0.8	stare	0.75	60	66.24	22	1.86
SSS	500	1750	3.5	4096x4096	15	62x62=3775	1.8	4.06	0.35	stare	0.75	60	183	20	0.26
LONEOS	590	1110	1.91	4096 x4096	13.5	55x55=3057	2.5	8.14	0.8	stare	0.6	45	390	19.3	1.36
Spacewatch	946	3000	3.17	4x2048x4608	13.5	4x27x62=6878	1	2.9	0.8	stare	0.5	120	43.5	21.7	0.22
NEAT	1200	3150	2.5	112x600x2400	13	27256	0.85	9	0.8	stare	0.57	60	308	21	6.41
LINEAR	1000	2200	2.2	2560x1960	24	47x61=2890	2.25	1.96	0.8	stare	1.00	6	1200	19	1.32

The 3 first surveys are in fact only one using 3 different telescopes.
LINEAR is using 2 telescopes, so the factor of merit should be doubled. Here I put the data for only one telescope.
NEAT is currently surveying asteroids during only 40% of the available time. Same remark as for LINEAR, in this case the factor of merit should be multiplied by 0.4
SSS should get an upgrade of its CCD to a thinned CCD.
I have not taken central obstruction into account, not having the correct data for all the telescopes.
I didn't include other surveys than NASA's Spaceguard survey, or other telescopes (NEAT is using a telescope in Hawaii too, Spacewatch at times is using its Spacewatch II telescope).

For any comments or modifications, please send me an email.