NA

TIO

NA

LA

STRONOMY AND IONOSPHE

RE

CE

NT

ER

ARECIBO OBSERVATORY

Photo: David Parker, 1997/Science Photo Library

The NAIC is operated by Cornell University under a Cooperative Agreement with the National Science Foundation.

March 2001, Number 32

INDEX

Resetting the Primary Reflector ..... 1

Radio Astronomy Highlights ........... 4

Observing with the Upgraded Arecibo

Telescope ....................................... 5

Space and Atmospheric Sciences .. 12

Planetary Science ........................... 16

Computer Department News ........ 16

Employee of the Year 2000 ............ 17

A School Science Project ................ 17

Visit from Congressional Staffers . 19

Colloquia since the last Newsletter 19

Comings and Goings ...................... 19

Observing Proposal Reminder: The next proposal deadline is June 1, 2001. Please make a note to get your proposalsfor observations using the Arecibo Observatory facilities submitted by that date. Details can be found at our web sitehttp://www.naic.edu/vscience/proposal/proposal.htm.

Resetting the Arecibo PrimaryReflector Surface

Paul Goldsmith

Although not strictly considered partof the Arecibo Upgrade project, the

surface of the 305 m telescope plays acritical role in the overall system per-formance, particularly at the higher fre-

Figure 1: An image of the errors in the main reflector surface processed from Lynn Baker’s photogram-metry data by Germán Cortés. Blue indicates positive devation from an ideal surface and red/yellow

means negative deviation. The unweighted rms is about 15 mm. (Courtesy Germán Cortés).

NAIC/AO NewsletterMarch 2001, Number 32 2

quencies that is one goal of the Upgrade.The surface was last surveyed and ad-justed about 15 years ago, and a lot hashappened since. The main cables sup-porting the surface are held to the groundby about 2000 cables which connect toconcrete blocks sitting on the ground be-neath the reflector. Soil motions thus candirectly impact the shape of the reflec-tor surface.

Based on monitoring by surveying,José Maldonado (NAIC) had indicationsthat these motions had been significant,particularly in the southeast quadrant ofthe reflector. That part of the natural“sinkhole” in which the reflector wasbuilt had been filled in extensively withdirt and construction material from oth-er parts of the bowl and elsewhere. Thisarea was less stable than the rest of theground, and it was not surprising that itshould be more subject to gradual move-ment. Puerto Rico is in a fairly activeseismic zone and there are tremors thatproduce small motions—in particular, ofthis not very well compacted portion ofthe ground under the reflector.

In addition to the subsidence, theupgrade work itself was quite traumaticfor the reflector surface. There weremany panels damaged by items droppedby construction crew working on theplatform and the feed arm. Also, therewas one large cable that was dropped,and when this hit the dish surface, it de-stroyed over 100 panels, and broke someof the cables that support the dish sur-face. These panels and cables have longbeen repaired, but the process may cer-tainly have contributed to a deteriorationof the accuracy of the reflector surface.

Since one of the major goals of theupgrade was to raise the upper frequen-cy limit of operation to 8 GHz or higher(wavelengths less than 4 cm and hope-fully as short as 3 cm) it was evident thatwe would be pushing the accuracy of theprimary reflector. Some very limitedtests using the mini-Gregorian carriedout by Phil Perillat (NAIC) in April 1991showed that things were not terrific inthis frequency range, with a sensitivity

of 0.25 K/Jy at 10.67 GHz. And this wasbefore the contractors started redoing thesurface for us! The mini-Gregorian il-luminated only a small fraction of thetotal surface, but the derived surface rmsfrom those measurements was 3.3 mm—not too terrible, but higher than onewould like for efficient operation.

Previous campaigns to set the prima-ry reflector were based on optical sur-veys with theodolites. In this procedure,the location of targets located above themain (North-South) cables was deter-mined by triangulation based on mea-surements made from several pointsaround the reflector rim. The locationsof these selected points could be mea-sured and adjusted to an accuracy of ap-proximately 1 mm rms. However, thepanels are only 1/4 the size of the spac-ing of the main cables, and each of themcan be adjusted. In the approach usedearlier, the positions of panels betweenthe main cables were “interpolated” be-tween the measurements of the widely-spaced targets on the main cables. It wasthought that the overall rms was on theorder of 2.5 mm, only slightly less thanimplied by the X-band measurementsmentioned above.

To perform really well, one needs theoverall rms surface error to be less than1/20 wavelength, which translates to 3mm rms at 10 GHz. The panels them-selves are thought to have an error ofapproximately 1 mm rms, and the sec-ondary and tertiary reflectors contributesmaller errors. So it would be desirableto get the primary surface adjustmenterror below 2 mm rms. It was judgedimpractical to reach this level using thetechnique employed previously. In as-sessing options, we decided to adoptoptical photogrammetry.

For this approach, reflective targetsare put on the panels; these targets are 3inch diameter disks of retroreflectivematerial. Using a special camera, pho-tographs of the dish are taken from thetop of each of the towers. You can imag-ine each photograph as yielding the an-gular coordinates of the target. If you

combine the angles to a given target fromthree or more viewing positions, you cansolve for the three dimensional locationof the target. This technique has beenrefined and turned into a commerciallyavailable combination of hardware andsoftware by a company called GeodeticServices Inc. NAIC has been workingwith the president of GSI, Mr. JohnBrown, since 1994, and last year we fi-nally were able to get an order in for thespecial equipment needed. A somewhatdifferent version of this same approachwas used to measure the secondary andtertiary reflectors—the main differenceis for those relatively small reflectors, aCCD camera was used.

For the measurement of the primary,we have to use a large-format film cam-era. Part of the reason why is evident ifyou compare the number of pixels in a 6inch by 8 inch piece of film, versus eventhe biggest “megapixel” CCD. My crudeestimate is that we get at least a “gigapix-el” format with the film camera. This isnecessary if you want to measure a tar-get 500 m away to an accuracy of 1 mm.

What happens in practice is that thecamera is taken up to the top of one ofthe towers. It is accompanied by severalintrepid NAIC staff members, typicallyLynn Baker, Felipe Soberal, and some-times others. From the tower top, theytake a number of photographs of the dishsurface—several photographs are neces-sary to cover the entire area, and theyalso take photographs with the camerarotated by 90 degrees to be able to iso-late any distortion in the camera’s imag-ing system, and take views from twodifferent positions on each tower top aswell. The illumination is provided by apowerful strobe lamp, which togetherwith the retroreflective properties of thetarget, guarantees that the targets standout with good contrast relative to thegeneral dish surface. It also means thatthe effective exposure time is very short,minimizing any mechanical vibrations,etc. Getting all the required equipmentto the tower tops is no mean feat, and wehave to admire those who carried out thisdifficult and sometimes hazardous work.

NAIC/AO NewsletterMarch 2001, Number 32 3

After the photographs are taken, thefilm is developed, and then each imageis digitized using a special scanner thatis located in lab adjacent to Tony Ace-vedo’s (NAIC) office. This scanner is aclose relative to plate measuring ma-chines used by astronomers; it measuresthe centroid of each of the “spots” pro-duced by the targets to an error of nomore than a few microns. These cen-troid positions are entered into a data fileon the PC controlling the scanner.

After all the photographs arescanned, the data files are combined us-ing a special program developed by GSI,which outputs the location of each tar-get in three coordinates. Next, Lynn fitsa sphere to the data set, and derives theerrors for each target relative to the best-fit sphere. The software also gives theuncertainty in each position; this dependson where on the dish the target is locat-ed, and in how many photographs thetarget appears. We have been impressedthat the formal uncertainty in positionwhen we have a full set of photographsis about 0.6 mm rms. Thus, the systemappears to really have the capability tomeasure the whole surface to the re-quired accuracy, but then it will be up tous to adjust the panels to achieve ourgoal.

During the Fall of 2000, about 2,000targets were placed on the primary sur-face of the antenna. Most of these werelocated above the points where the “tie-back cables” (which connect the surfaceto the concrete anchors on the groundbelow, mentioned above) are located.Some extra targets were put in densepatches to fully sample the panel-to-pan-el setting errors. It was a real struggle toget the necessary data, as that was oneof the rainiest Fall periods in recentmemory, but this was finally accom-plished. The usual learning curve for de-veloping, scanning, and reducing datawas ascended, and we obtained the firstset of post-upgrade surface measurementdata.

An image of the errors (processedfrom Lynn Baker’s data by Germán

Cortés—NAIC) is shown in Figure 1.Blue means high and red/yellow meanslow. The big surprise is that the un-weighted rms is about 15 mm! This isworse than had been determined in 1991.The obvious conclusion is that all theupgrade work (plus the passage of theintervening 10 years!) severely degrad-ed the surface accuracy. It is difficult tocompare this photogrammetric rms di-rectly with that derived radiometrically,because the Gregorian system does notilluminate the entire surface, and largescale errors of the illuminated region(linear gradients and quadratic errors) aretaken out by calibration runs, appearingas pointing and focus offsets, respective-ly. However, there is no doubt that wehave adequate explanation for the rela-tively poor performance we have seenat 5 GHz, and also for the variability ofgain as a function of source declinationand hour angle. Note in particular thelarge errors seen in the “fill area”. Thelarge errors seen in the panels right atthe center of the dish are not surprisingas those are the “new” panels recently

installed by José Maldonado’s team, andthey have not yet been adjusted. Thelargest errors outside the center are onthe order of 100 mm! This is even big-ger than José Maldonado had expected,and shows how much that part of the dishsurface had sunk.

While the first round of photogram-metry was going on, José Maldonadoand his crew were undertaking to refur-bish a lot of the panel support hardwarethat had corroded since installation in1974. Several thousand panel supportsneeded to be replaced, and many moreto be cleaned up and greased so that ad-justment of the individual panels wouldbe possible. This work is still ongoing,and should be completed in April 2001.

That work was interrupted by the ar-rival of the results of the first round ofphotogrammetry indicating the presenceof very severe large-scale errors. Weimmediately started a project to adjustthe 2000 or so tieback cables to get thesurface closer to the desired sphericalshape. This work was completed in a

Figure 2: This image is from the second set of photogrammetry tests completed in January by Lynn

Baker and Felipe Soberal. The improvement is immediately evident when compared with Figure 1,and is quantified by the reduction in the rms surface error from ~15 mm to just over 5 mm.

NAIC/AO NewsletterMarch 2001, Number 32 4

remarkably short time, and in JanuaryLynn went again to Arecibo, and withFelipe obtained a second set of data,which is shown in Figure 2. The im-provement is immediately evident (thecolor tables are the same), and is quanti-fied by the reduction in the rms surfaceerror from ~15 mm to just over 5 mm.The situation is really somewhat better.If one allows for fact that some of theadjustments were done in the wrong di-rection (something that always happensin a campaign like this, despite the bestefforts of the field crews), and one ex-cludes them, the rms is down to about3.5 mm. Again, this does not allow forany low-order terms, and includes theentire antenna surface (except the cen-tral panels which still were not adjust-ed). Part of the remaining error may bedue to the fact that some of the adjust-ments were so large that there were in-teractions between adjustment points andpossibly even nonlinearities in the rela-tionship between tieback cables and sur-face position.

Naturally, although this already rep-resents a huge improvement, we are notsatisfied, and a second set of requiredadjustments has been generated. As thisis written, over half of the tieback cableturnbuckles have already been adjustedfor the second time. The central panelswill also be included in this round. Sowe can really hope to get the large-scalerms down to a few mm. Unfortunately,this has all happened so quickly that wehave not had time to schedule the re-quired telescope time to see what hashappened to the antenna performance. Insome limited time creatively obtained byJohn Harmon (and thanks to those whogave up their scheduled observations!)Phil Perillat and Mike Nolan (NAIC) didcarry out some measurements. There areindications (but these must be consideredpreliminary) that the L-band gain maybe up by about 10%, that the S-band gainis up from 5.5 to 7.0 K/Jy, and that at C-band (5 GHz) we have a single beamwith 4 - 5 K/Jy consistently. I am goingout on a limb to even put these results inprint, but I know that they are what ev-eryone wants to hear about. I caution

again that these are very preliminary.However, I am confident that the photo-grammetry is giving us the right answers,and that we can do better yet.

So —what happens next? After wecomplete the ongoing second round oftieback cable adjustments, Lynn andFelipe will do the photogrammetry for athird time. We also will be schedulingadditional telescope time to define theantenna performance more completely.This is all a prelude for Phase II, in whichwe adjust the position of each panel in-dividually. The first step here is to getapproximately 39,000 targets out on theantenna surface. The targets themselvesare currently on order and should arrivewithin a month or so. By that time, allof the panel support hardware should berefurbished, and the nontrivial task ofputting those targets on the antenna sur-face will be accomplished. Then, thereally demanding job of doing the pho-togrammetry, but measuring 39,000 rath-er than 2,000 targets will begin. This isconceptually not different, but in prac-tice the amount of time and effort to scanthe photographs will increase greatly,simply due to the increased number oftargets.

Adjustment of the individual panelscan then begin, and this too may requireseveral iterations. Thus, this project islikely to go on for another year. In addi-tion to the surface adjustment itself, wewill be installing the tertiary actuatorsand computer control system, which willbe necessary to make the small focus andpointing corrections necessary for oper-ation at the shortest wavelengths. BillSisk (NAIC) has been working with thissystem extensively and it is almost readyto go, but installation needs to be syn-chronized with a couple of other nastytasks including shimming the elevationrails. It does seem that efficient opera-tion at 5 GHz is now within our grasp,and 10 GHz is not too far off. I hopethat in the next newsletter we can giveyou some detailed results of antennameasurements at the higher frequencies,and before long, some scientific resultsas well.

Radio Astronomy Highlights

Chris Salter

Pulsar Scintillations

The Oberlin/Cornell collaborationlead by Dan Stinebring (Oberlin)

continues to investigate the high-Q “par-abolic arcs” that they have been seen inpulsar secondary spectra (power spectraof the dynamic spectra). These arcs,which are the transform domain equiva-lent of the criss-cross patterns that haveoften been noted in pulsar dynamic spec-tra since the early 1970s, will be famil-iar to faithful readers of these pages. Infact, these arcs made their debut as“wisps” in the Spring 1999 NAIC-AONewsletter (No. 27) after the group madeintensive observations during January,1999. It will interest some readers thatthat article caught the attention of noneother than Ronald Bracewell, who hadsome interesting suggestions to makeconcerning further analysis of the pat-terns.

In addition to roughly biweekly ob-servations — mostly performed remote-ly — that the group makes to monitortime variability of the phenomenon to-ward half a dozen strong, nearby pulsars,they are continuing to explore the effectin archival data, much of it taken atArecibo by Jim Cordes (Cornell) duringthe 1980s. The most remarkable resultto come out of the analysis of this earlierdata is how little the arc pattern changes

As indicated above, many peoplehave been working very hard on the sur-face adjustment project. Lynn Baker,Don Campbell (NAIC), José Maldona-do, Mike Nolan, Phil Perillat, and FelipeSoberal have been extensively involved,and they have been supported by manyothers at Arecibo and also in the NAICMaple Avenue laboratory. Mr. JohnBrown of GSI has been extremely help-ful in getting us up to speed with the pho-togrammetry system at Arecibo. Thesepeople are the ones who deserve creditfor getting Arecibo working through theentire cm wavelength range.