Space Surveillance Sensors: GEODSS (Ground-based Electro-Optical Deep Space Surveillance) System (August 20, 2012)

The GEODSS (Ground-based Electro-Optical Deep Space Surveillance) System is the United States’ primary deep space tracking system.  The Deep Stare upgrade of GEODSS, carried out in about 2003-2005, greatly increased the capabilities of the system.  GEODSS uses a total of nine large telescopes at three different locations to track space objects by using reflected sunlight, (and thus can only operate at night and when not cloudy).  It can likely detect objects with sizes under 0.5 meters in geosynchronous orbits.  GEODSS provides about 60% of all the SSN’s deep space (orbits with periods greater than 225 minutes) observations and nearly 80% of all geosynchronous observations.[1]   There are over two hundred GEODSS-tracked objects that are not tracked by any other sensor.[2]

A GEODSS Telescope.[1]

 Locations and Coverage

There are three GEODSS sites:  Socorro, New Mexico (32.82° N, 106.66° W), Diego Garcia atoll in the Indian Ocean (7.41° S, 72.45° E), and Maui, Hawaii (20.71° N, 156.26° W).  A fourth GEODSS site in South Korea was closed in 1993, apparently due to both poor observing conditions and financial considerations.  The similar but smaller, single-telescope MOSS (Moron Optical Space Surveillance) System (37.10° N, 5.37° W) in Spain also acts in coordination with the GEODSS network.  The first GEODSS site (Socorro) went into operation in 1982. The GEODSS sites are part of the 21st Operations Group of the 21st Space Wing (Socorro is Detachment 1, Diego Garcia is D2, Maui is D3 and Moron is D4).

 The Maui site is at an elevation of just over 3 km (3.06 km) and Diego Garcia is at sea level.  The Socorro site is at White Sands Missile Range, which has a minimum elevation of about 4,100 feet.

Together these sites provide complete or nearly complete coverage of the geosynchronous orbital belt.  Figure 1 shows the elevation of the geosynchronous belt as a function of the relative longitude difference between the GEODSS (or MOSS) site and a point on the geosynchronous belt.  It shows that, for example, with a 20º minimum elevation angle, each site can see in excess of 100 degrees of the geosynchronous belt.  Figure 2 then shows the longitude coverage of the geosynchronous belt by the GEODSS sites plus the MOSS system.  This shows that the system has nearly complete coverage at a 20˚ minimum elevation angle (complete coverage is obtained with a slightly lower minimum elevation angle).

Figure 1: Visible longitude extent of the geostationary belt (± degrees) as a function of minimum viewing elevation for each GEODSS site.


Figure 2: GEODSS coverage of the geosynchronous belt.  The longitudes of the observing sites relative to several major cities (as viewed from above the south pole) are shown on the inner circles.  The outer curves show the longitudinal coverage of each site, assuming a minimum elevation angle of 20˚.  For a 20º minimum elevation angle, there are two small gaps in longitude coverage.  These gaps disappear if the minimum elevation angle is reduced to 15º.


GEODSS Up To the Deep Stare Upgrade

Each of the three GEODSS sites has three primary telescopes.  The sites used to have only two primary telescopes and one smaller, wide-field auxiliary telescope, however, all three sites now have three primary telescope.  These primary telescopes are f/2.15 (that is, the focal length is 2.15 times the aperture diameter) Ritchey-Chretien telescopes with a diameter of about 1.01 meters.[1]  For light gathering purposes, due to losses within the telescope, their effective apertures are about 0.46 m2.[2]   The telescope has an 80 mm circular focal plane.[3] 

As originally constructed, the detector of each primary telescope was a low-light-level television camera system using an electron-bombarded silicon (Ebsicon) vacuum tube.  The system had a circular field of view of about 2° in diameter.  Each of the 832 pixels of resolution across the diameter thus had a field of view of 8.8 arc-seconds.[4]  However, the system was usually operated using an electronic zoom that gave a field of view of about 1° in diameter (4.4 arc-second per pixel).

 The Ebsicon-equipped GEODSS was often described as being able to detect 16th magnitude targets, but 15th magnitude appears to be more realistic.  One source describes it as being capable of detecting 15.3 mv targets against an mv = 19.5 background.[5]  Another describes it as capable of detecting mv = 15.8.[6]  A brightness of mv = 15.5 corresponds to about a 1.1 m object (diffuse reflecting sphere, reflectivity = 0.1, phase angle = 0) in geostationary orbit.

Over the years, the system underwent several upgrades.  The GEODSS Modification Program (GMP) upgrade, which went into operation in early August 1999, significantly increased the number of tracks GEODSS could make, primarily by improving the scheduling of the system’s telescopes.[7]


The Deep Stare Upgrade

The most recent and significant upgrade, known as Deep Stare, included replacing the Ebsicon vacuum tube detector in the primary telescopes with a charge-coupled device (CCD) array detector.   The upgrade, which was apparently completed by about 2005, also involved new telescope mount hardware and software.  While the upgrade was initiated primarily because the Ebsicon tubes were no longer being produced and could not be maintained, it also brought about significant increases in capability.  The upgrade was expected to bring about a 2.5 magnitude gain in detection sensitivity, an improvement of more than a factor of two in position measurement accuracy, and a 40% increase in search rate.  A service life extension program (SLEP) is currently planned for GEODSS that would replace aging sensor controller, data processing, and data communications equipment. 


The Deep Stare CCD Detector

The new CCD detector for the Deep Stare upgrade to GEODSS was built by the Sarnoff Corporation based on the Lincoln Laboratory’s CCID-16 CCD design.[8]  It is a 2560×1960 pixel CCD, and, paired with the GEODSS telescope gives a field of view of 1.23° x 1.61°.  Each CCD pixel is 24 μm square and has a field of view of 2.27 x 2.27 arc-seconds.  (Thus at geosynchronous distances, the total field of view corresponds to about 860 x 1125 km, and a single pixel to about 0.44 x 0.44 km.)

The CCD has a split-frame-transfer architecture with eight output ports and a 2.0 MHz clock rate that gives a maximum frame rate of 2.7 per second (a frame time of 0.37 seconds).   The CCD has a peak quantum efficiency (QE) of about 0.86, and its QE is greater than 0.3 across the entire 0.4 to 0.9 μm band.[9]  The solar-weighted QE averaged over this band is about 0.65, eight times greater than the 0.08 QE of the pre-upgrade vidicon detector.[10]

 The CCD’s well capacity is greater than 140,000 and it has 13 bits of dynamic range (indicating that its gain must be at least 17 electrons/count).  Its average dark current is less than 6 electrons/pixel/second and its readout noise is less than 12 electrons. 

The CCD has a separate 32 x 32 pixel split-frame array for photometry and Space Object Identification (SOI).  Its characteristics are similar to that of the larger array, except that it is capable of a frame rate of 1,000 frames/second at a 1.25 MHz clock rate.[11]


Limiting magnitude

GEODSS can operate in either a sidereal mode or a target-track (rate-track) mode.  In the sidereal mode, the telescope tracks the stars, which thus appear as points.  Orbiting space objects are then readily identified as streaks against the fixed star background, and observations are formed by measuring the beginning and end locations of the streaks relative to the stars.  This is GEODSS’ standard operating mode for tracking space objects.  In target track mode, the telescope tracks the space object, which thus appears as a point while the stars appear as streaks.  Somewhat fainter objects can be detected in this mode, because the signal from the object is not distributed in many pixels along the length of a streak.  This mode is also used for Space Object Identification (SOI) measurements.

 Analysis before the upgrade indicated that a Deep Stare GEODSS telescope should be able to detect and track geosynchronous objects as dim as mv = 17.7 in sidereal mode and mv = 18.9 in target-track mode (the background and S/N were not stated).[12]  This was said to be about two magnitudes better than before the upgrade.

A later pre-upgrade analysis similarly concluded that a Deep Stare upgraded GEODSS telescope would be capable of detecting a 17.9 mv object against a 19.5 mv background.[13]    This was described as an improvement of 2.5 magnitudes over the pre-upgrade system, which had a limiting magnitude mv =15.3 against the same background. 

 A limiting magnitude of mv =17.9 corresponds to a sphere in geostationary orbit (diffuse reflecting, reflectivity = 0.1, phase angle = 0) with a diameter of about 36 cm and a magnitude mv = 18.9 corresponds to a sphere diameter of about 22 cm. 

 The limiting magnitudes are in part determined by the short integration time of 0.37 seconds used by GEODSS.  The telescopes used in the Lincoln Near-Earth Asteroid Program Research (LINEAR) program, which are identical to GEODSS telescopes, are capable of achieving a magnitude of mv = 22 (S/N = 4) with less than 100 seconds of integration times.[14]  This would correspond to a sphere diameter of about 5.6 cm in geosynchronous orbit.  However, there is no public indication that the GEODSS telescopes are operated with integration times longer than 0.37 seconds.


Field of View

The Deep Stare upgraded telescope/detector also has a larger field of view.  Prior to the Deep Stare upgrade, GEODSS telescopes were typically operated in an electronic zoom mode that resulted in a circular field of view with a diameter of 1.05˚, for a total field of view of about 0.87 square degrees.[15]  Following the upgrade, the system had a rectangular field of view of 1.23˚ x 1.61˚ = 1.98˚ square degrees, more than twice as large.


Track and Search Rates

Key measures for a system such as GEODSS are the number of tracks it can take or the amount of sky it can search in a given time interval.  In general, for space surveillance systems, a track consists of a number of separate observations (measurements), with the number of observations per track depending on the system.

 The basic method of producing a track changed substantially with the Deep Stare upgrade, allowing much more efficient operation.  Before the upgrade, largely due to unpredictable non-linearities in the pre-Deep Stare Ebsicon detector, after detecting a target it was necessary to then center it in the telescope’s field of view before making a metric measurement.  The angular position of the centered object was then determined based on the pointing angle of the telescope’s mount.  

Following the Deep Stare upgrade, it is no longer necessary to center an object in the field of view and the positions of multiple objects in a single field of view can be measured simultaneously.  The metric measurements on detected objects are now made by comparing their positions to the known positions of stars in the same field of view.

Prior to the GMP and Deep Stare upgrades (~ 1998 and earlier), a GEODSS telescope had ideal operational capabilities of about 931 tracks per eight hours = 116 observations/hour and a search rate of 583 deg2/hour.[16]   With five observations per track, this translates to a rate of about 23 tracks/hour (per telescope).[17]   Another source gives an ideal operational rate of 20 tracks per hour for pre-GMP GEODSS.[18]

 However, in actual practice, the track rate appears to be much lower.  Based on data from August through December 1998, prior to the GMP upgrade, GEODSS produced 40,658 tracks, corresponding to less than 5 tracks/hour per telescope.[19]  Other data (covering the period from April 13 to December 31, 1998, 263 days), gives about 5.3 GEODSS deep space tracks per day[20]  Thus the actual track rate of the original GEODSS system was about 5 tracks per hour per telescope, only about one-quarter of the ideal operational capability of about 20 tracks per hour per telescope. 

There are several reasons why the actual tracking rate would be expected to be less than less than the ideal operational rate.  GEODSS requires good weather, since it cannot operate through clouds.  This alone could reduce the number of tracks by a factor of two or more.   Cloud cover data shows that at the Socorro site, the sky is clear just over 50% of the time, for Maui just under 50% of the time, and for Diego Garcia, less than 40% of the time.  The system is also affected by adverse wind and humidity conditions and by the Moon.  One early estimate was that availability of a GEODSS site would generally be less than 30%.[21]  In addition, the system also performed tasks other than tracking.  For example, it collected SOI data and also conducted searches for objects that are not in their expected positions. 

 After the GMP upgrade, the ideal operational track rate increased to about 1800 observations per telescope per eight hours and the search rate to about 600 deg2 per hour.[22]  The number of observations per streak at this time was about 3.3, which would give an ideal operational track rate of about 68 tracks/hour for each telescope.[23]  In actual practice, just after the GMP upgrade the GEODSS system produced a total of 116,052 tracks from August through December of 1999.[24]  At that time there were seven GEODSS main telescopes operational, so this corresponds to an actual track rate of about 13.5 tracks per hour per telescope.  

 The Deep Stare upgrade significantly improved GEODSS’ tracking and search capabilities.  It is reported that a Deep Stare upgraded GEODSS telescope would be able be able to search 840 deg2/hour, or in tracking mode make 4,600 observations/8 hours.[25] 

 In its streak detection mode, GEODSS reportedly takes a series of exposures for a total of 6 seconds.[26]  It then waits 6 seconds and takes a second set of exposures for another six seconds.  Since the time required to step and settle to the next search position is about one second, the search of one field of view thus takes just under 20 seconds.  It generates two streaks, and since an observation (a position measurement) is made for the beginning and end of each streak, there are thus four observations every 20 seconds.

Operating in this way, a GEODSS telescope could make 4 x 3 x 60 = 720 observations/hour or 5,760 observations/8 hours.  Allowing for some additional overhead time, this seems reasonably consistent with the reported values.

 The field of view for GEODSS is about 1.23° x 1.61° = 2.0 deg2.  Operating in the above mode, GEODSS can thus search a maximum of 6 deg2 per minute or 360 deg2/hour, well below the claimed value of 840 deg2/hour.  If, however, in search mode it simply generates one streak per search position, the time per search position could be as low as 7 seconds, giving a search capability of 1,030 deg2/hour, reasonably consistent with the reported value allowing for some additional overhead time. 

The claimed ideal operational track rate of 4600 observations/8-hours, with four observations per track, gives an ideal track rate of about 144 tracks per telescope per hour.  However, as with the pre-Deep Stare GEODSS, the actual search and track rates will be almost certainly be considerably less than the ideal rates.

 One aspect of the Deep Stare upgrade that likely enhanced the ratio of actual to operational rates was the addition of an Infra-Red Cloud Imager capability.  This part of the upgrade, begun in 2002 and completed in 2004, allows the system to operate efficiently in partly cloudy skies by determining which parts of the sky are not cloud covered.  According to one site manager (detachment 1), this addition has increased the amount of time the system can operate “upwards of approximately 15 to 20 percent per month.”[27]


Metric Accuracy

The Deep stare upgrade was expected to increase the system’s metric accuracy (in sidereal track mode) from about 4 to 6 arc-seconds to less than 2 arc-seconds (about 1 x10-5 radians).[28]


Space Object Identification.

In Space Object Identification mode, GEODSS measures the time variation of the brightness of an object, producing an intensity versus time signature of a specific space object.  In SOI mode, the telescope is operated in a target-track mode.  Since important brightness variations can occur on time scales much shorter than the integration times used for search or tracking (0.37 seconds for the Deep Stare system) a much shorter integration time, typically 0.01 seconds, is used.  Accordingly, in SOI mode, the minimum target brightness is several magnitudes less than in tracking or search mode.  In SOI mode, all the target energy is collected (that is, the signal is collected from multiple pixels centered on the target) and the sky background is also measured, so that an absolute measurement of the target brightness can be made.

Prior to the Deep Stare Upgrade, SOI was accomplished by splitting the incoming signal.  Most of the incoming signal was directed to a photo-multiplier tube for the SOI measurement, while the rest was directed to the main Ebsicon detector to maintain closed-loop tracking on the target.  This limited SOI to objects with mv = 12.8 or brighter.

Following the Deep Stare upgrade, SOI is accomplished using a separate 32×32 array (with otherwise the same characteristics as the main array) on the focal plane.  This array can operate at rates up to 1,000 Hz.  The system can make SOI measurements on objects as dim as mv = 16.3 (presumably this is at about 100 Hz, not 1,000), or about 25 times dimmer than before the upgrade.

[1] A Ritchey-Chetrien is a type of reflecting telescope that uses hyperbolic mirrors to achieve a relatively-wide distortion-free field of view. The Hubble Space Telescope is a Ritchey-Chetrien design, although with a much larger f-number of  f/24.

[2] Grant H. Stokes, Donald K. Yeomans, William F. Bottke, Jr., Steven R. Chesley, Jenifer B. Evans, Robert E. Gold, Alan W. Harris, David Jewitt, T.S. Kelso, Robert S. McMillan, Timothy B. Spahr, and S. Peter Worden, “Study to Determine the Feasibility of Extending the Search for Near-Earth Objects to Smaller Limiting Diameters,” Report of the Near-Earth Object Science Definition Team, NASA, August 22, 2003, Table 4-4, p. 40.    This number  is for a LINEAR (Lincoln Near Earth Asteroid Research) program telescope, which is identical to a GEODSS primary telescope.

[3] Walter J. Faccenda, David Ferris, C. Max Williams, and Dave Brisnehan, “Deep Stare Technical Advancements and Status,” MITRE Technical Paper, October 2003.  Available at:

[4] Faccenda, et. al., “Deep Stare Technical Advancements.”

[5] Faccenda, et. al., “Deep Stare Technical Advancements.”

[6] E.C. Pearce, “The Transportable Optical System,” Proceedings of the 1997 Space Control Conference, Vol. 1, Lincoln Laboratory, March 25-27, 1997, p. 27.

[7] J.G. Miller and W.G. Schick, “Contributions of the GEODSS System to Catalog Maintenance,” Proceedings of the 2000 Space Control Conference, Lincoln Laboratory, April 11-13, 2000, pp. 13-27.

[8] The numbers in this and the next three paragraphs are from: John R. Tower, Pradyumna K. Swain, Fu-Lung Hsueh, Robin M. Dawson, Peter A. Levine, Grayzna M. Meray, James T. Andrews, Verne L. Frantz, Mark S. Grygon, Michael A. Reale, and Thomas M. Sudol, “Large Format Backside Illuminated CCD Imager for Space Surveillance,,” IEEE Transactions on Electron Devices, Vol. 50., No. 1 (January 2003), pp. 218-224, and Faccenda, et. al., “Deep Stare Technical Advancements and Status.”

[9] Tower, et. al., “Large Format Backside Illuminated CCD.”

[10] Evans, et. al., “Detection and Discovery of Near-Earth Asteroids.” This paper states that the detector used in the LINEAR telescope, which is the same as the GEODSS detector, has a “peak quantum efficiency exceeding 95% and a solar-weighted quantum efficiency of 65%.”

[11] Tower, et. al., “Large Format Backside Illuminated CCD.”

[12] C. Max Williams and Sam D. Redford, “GEODSS Upgrade Prototype System (GUPS) Program Status,” Proceedings of  the 1996 Space Surveillance Workshop, Lincoln Laboratory, 1996, pp.99-108.

[13] Faccenda, et. al., “Deep Stare Technical Advancements and Status.”

[14] Grant H. Stokes, Frank Shelly, Herbert E.M. Viggh, Matthew S. Blythe and Joseph S. Stuart, “The Lincoln Near-Earth Asteroid Research (LINEAR) Program,” Lincoln Laboratory Journal, Vol. 11, No. 1 (1998), pp. 27-40.

[15] Faccenda, et. al., “Deep Stare Technical Advancements.”

[16] “Ground-Based Electro-Optical Deep Space Surveillance (GEODSS) System,” MITRE Poster.  Available at:

[17] Five observation tracks were the type most commonly assigned to ground-based optical sensors at this time. From August to December 1998, the average number of observations per track was 5.0. J.G. Miller and W.G. Schick, “Contributions of the Space-Based Visible Sensor to Catalog Maintenance,” in L.B. Spence,  Proceedings of the 1999 Space Control Conference, Lincoln Laboratory, April 13-15, 1999, pp. 163 -174.  In sidereal tracking mode, GEODSS would produce metric measurements based on two sets (separated in both space and time) of five streaks, with at least three observations being required from each set of five streaks to generate a track.

[18] Pearce, “The Transportable Optical System (TOS).”

[19] J.G. Miller and W.G. Schick, “Contributions of the GEODSS System.”  Based on seven operational primary telescopes, 153 days, and eight hours per day, this corresponds to an hourly rate of about 4.7 tracks per hour.  However, this hourly rate is only approximate because the GEODSS system had some downtime for GMP testing and because tracks from auxiliary telescopes

[20] Based on figures 4 in J.G. Miller and W.G. Schick, “Contributions of the Space-Based Visible Sensor to Catalog Maintenance,” in L.B. Spence, Proceedings of the 1999 Space Control Conference, Lincoln Laboratory, April 13-15, 1999, pp. 163 -174.

[21] Robert Weber, “The Ground-Based Electro-Optical Detection of Deep Space Satellites,” Proceedings of the Society of Photo-Optical Instrumentation Engineers, Vol. 143: Applications of Electronic Imaging Systems, March 30-31, 1978, pp. 59-69.

[22] Faccenda, et. al., “Deep Stare Technical Advancements.”

[23] During the period from August to December 1999, the average number of observations per track was 3.3.  Miller and Schick, “Contributions of the GEODSS System,” p. 16.

[24] Miller and Schick, “Contributions of the GEODSS System,” p. 14.

[25] Faccenda, et. al., “Deep Stare Technical Advancements.”

[26] Faccenda, et. al., “Deep Stare Technical Advancements.”

[27] Martha Petersante, “New Imager Allows GEODSS to Find Holes in Clouds,” Hansconian, December 5, 2003 (

[28] Faccenda, et. al., “Deep Stare Technical Advancements.”

[1] This is actually a photograph of a former GEODSS telescope now being used by the Lincoln Near-Earth Asteroid Research (LINEAR) Program at the Lincoln Laboratory’s Experimental Test Site adjacent to the Socorro, New Mexico GEODSS site.  Picture source: Jenifer B. Evans, Frank C. Shelly, and Grant H. Stokes, “Detection and Discovery of Near-Earth Asteroids by the LINEAR Program,” Lincoln Laboratory Journal, Vol. 14, No. 2 (2003), pp. 199-220.

[1] Eugene Burgio and Ken Grant, “Unique Search and Track Procedures Utilizing the Ground-based Electro-Optical Space Surveillance (GEODSS) Worldwide Sites,” Advanced Maui Optical and Space Surveillance Technologies Conference, September 13-16, 2011.  Available at:

[2] Richard F. Colarco, “Space Surveillance Network Sensor Development, Modification and Sustainment, Programs,”  Advanced Maui Optical and Space Surveillance Technologies Conference, September 1-4, 2009.  Available at:


Leave a comment

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: