Next: Donald R. Farris: 1937 - 1997
Previous: First Observations with Fast 2K x 2K CCD Camera
Table of Contents - Search this issue - NOAO Newsletter Home Page

NOAO Newsletter - Global Oscillation Network Group - March 1997 - Number 49


Global Oscillation Network Group

The Global Oscillation Network Group (GONG) Project is a community-based activity to operate a six-site helioseismic observing network, to do the basic data reduction, provide the data and software tools to the community, and to coordinate analysis of the rich data set that is resulting. GONG data is available to any qualified investigator whose proposal has been accepted; however active membership in a GONG Scientific Team encourages early access to the data and the collaborative scientific analysis that the Teams are undertaking. Information on the status of the Project, the scientific investigations, as well as access to the data, is available on our WWW server whose URL is www.gong.noao.edu.

Data Analysis

With the year-long spectra in hand, we are entering entirely new realms of sensitivity and spectral resolution. The high quality of the GONG spectra, free of temporal sidelobes, has revealed an unexpectedly rich structure of spatial sidelobes. These "spatial leaks" arise from the decomposition of the velocity field into spherical harmonics on only a portion of the sphere. In this case, the spherical harmonics do not represent a complete set, and the cross products of harmonics with different degree and order do not vanish. We will be changing the shape and apodization of the spatial area over which the images are decomposed into spherical harmonics to address this. Preliminary tests indicate that this will significantly reduce the number of spatial leaks present in the spectrum. Even so, if the remaining leaks are not correctly incorporated into the model fitting, systematic bias can be introduced in the estimated frequencies. As we are attempting to measure the frequencies to a part in 106, small systematic differences of tens of nanoHz just cannot be ignored, nor can subtle differences between various "peak-bagging" methods.

We have assembled a "tiger team" to attack these interrelated "opportunities." Yeming Gu and Rudi Komm have been developing methods to smooth the spectrum using multi-taper spectral estimates and wavelet denoising techniques. Frederic Baudin and David Feirry-Fraillon are applying methods developed in the context of non-imaging helioseismology - e.g. homomorphic deconvolution. Philip Stark is providing us with the latest statistical techniques, and Rachel Howe has developed a spatial leakage matrix for the GONG observations, Stuart Jefferies is contributing his deconvolution and global fit method, and his line asymmetry model. Finally, Ed Anderson has been kept very busy implementing all of these suggestions! We hope that within the next six months, a new (most likely hybrid) peak bagging scheme will become available to produce substantially improved frequency estimates with lower systematic and random uncertainties.

Other changes are also in the works for the analysis of GONG data. Cliff Toner's optimization method to determine the relative angular offsets of the cameras has been implemented. The site-to-site variations in the velocity calibration scale factor can now be removed, and a temporally varying sidereal-synodic correction is being installed. We are also investigating the effects of uncertainties in the assumed position of the solar rotation axis on the frequencies, and investigating the effect of the horizontal component of the oscillatory velocity field on the spherical harmonic decomposition. The details of the apodization and the padding of the time series are also being reevaluated.

Data Management

During the past quarter, "month-long" (36-day) time series and power spectra were produced for GONG months 13 and 14 (ending 96/09/21) with fill factors of 0.73 and 0.76. The fill factors for these two months were significantly lower than for the previous two months (0.94 and 0.82). The difference is attributed to changes in global weather patterns (including monsoons at Udaipur and typhoons at Learmonth), preventive maintenance visits to several sites, and some instrument problems. Ten-month (the concatenation of GONG months 4 through 13, the first full GONG year) time series and power spectra for spherical harmonic degrees less than or equal to 45 were produced; the fill factor is 0.87. The next GONG year products will be produced from months 9 through 18.

The delay between data acquisition and completion of the month-long power spectra, which had been about 17 weeks, increased during the past quarter (primarily, for month 15 which is still being processed) because of holidays and vacation. We received and processed 82 cartridges containing 578 site-days from the seven instruments. 383 site-days were processed through the site-dependent data reduction stages. The difference of 195 site-days is attributed to bad weather at the network sites and 81 site-days from the engineering test unit in Tucson.

During the past quarter, we serviced 20 data distribution requests for 61,427 files totaling 13.2 Gigabytes of data. The average delay between receipt of a request and shipping the media containing the data products was about 1.5 days. In addition, there were 2,378 data cartridge transactions (library check-ins and check-outs) in response to requests from the data reduction pipeline and other internal operations. The copying of new data product cartridges, the delivery of these copies to the off-site storage facility, and the testing of old media in both the local library and in the off-site storage facility continued as expected.

The Project is preparing for a reprocessing campaign that will probably begin before April to regenerate p­mode power spectra from calibrated velocity images. This will apply the software updates that occurred during the past eighteen months to all previously acquired network data, resulting in a more consistent and homogeneous data set. Additional changes to the routine data reduction are also being evaluated. These include correcting the scale of the velocity images by fitting the spatially averaged velocity to the observer motion, compensating the registration of the velocity images for the difference between synoptic and sidereal heliographic coordinates, changing the spatial window function from an aperture with constant range of heliographic latitude and longitude to an aperture with constant area on the observed solar disk, apodization of the images as they are registered onto the heliographic grid, increasing the sampling of the heliographic grid to eliminate some spatial aliasing that resulted from undersampling the images in camera coordinates, and changes to the temporal gap-filling, apodization, and length of the FFT.

Network Operations

Calendar Year 1996 did not pass into history for GONG network operations without event. The first sign of trouble for the operations group in Tucson came during the early evening on 30 December. We got a call from Learmonth, where their 31 December observing day was just beginning. Their rotating waveplate (the optical modulator) was suffering electronic resyncs every three minutes. John Kennewell noted that the problem began at about zero hours UT, but we thought nothing of it for the moment. It became apparent, however, that every station in the network had begun the same behavior at precisely the same time: 0:00 UTC on 31 December. In general (though not in detail) every site would take about two images, then resynchronize to the time base.

Upon further investigation, the Tucson staff noted that the GPS receiver in our Tucson laboratory (identical to the ones in service in the field) was toggling between day number 365 and day number 366 about every three minutes. We were so proud of the way that we slid through the leap day last 29 February without incident (although we were holding our breath), but "pride goeth before destruction, and a haughty spirit before the fall."

A phone call to the GPS receiver vendor yielded a busy line. This continued for quite some time. When we eventually got through to the service people, we learned that they were having a very, very bad day as well. It seems their firmware had a bug, still not understood at that time, but clearly related to 1996 being a leap year. They were hearing from most of their customers owning that model series as time went on, but could suggest no effective solution until their analysis was complete. An attempt on our part to disable the GPS time and date-reading function was not successful resulting in more frustration.

By this time we were within a few hours of a new day in Greenwich, and could do little but ride out the crisis. Most of the sites on the sunward side of the planet were cloudy anyway (as luck would have it) by that hour. At 5:00 P.M. Tucson time (New Year in Greenwich) we began checking on the sites and shaking them back to life. Each had to be roundly slapped (reboots, generally) to return it to its duty. It took a couple more hours (between Internet problems and overseas phone connections) before we completed our rounds and everything was returned to normal.

It remains to be seen what can be made of the images that were obtained on 31 December. The tapes containing these data are just beginning to show up. This sort of very gappy coverage (missing one out of three images at short intervals) is difficult to imagine merging effectively. Time will tell.

By the way, the GONG operations group is looking for a few volunteers from our community to serve as duty responders on New Year's day in the year 2000. We're planning to be on vacation, or stalking the GPS receiver vendor.

The rest of the last three-month (and two week) period has gone very well. In spite of the one "very dark day" in December our down-time statistics improved. We lost about 5600 images due to various instrumental problems, presuming that all of the clear-weather images from 31 December have to be thrown away. This is about 1.4% of the total number of images that could be obtained, bringing the overall down-time average since full-network operations began back down below 2%.

As for the weather, some of the northern-hemisphere sites have a tough go of it from time to time during this season. El Teide lost another anemometer propeller in an ice storm, and our Big Bear maintenance trip was held up for four days as we waited for a series of snow storms to move out of the mountains of southern California. But the weather in India and Hawaii has been quite good, and Learmonth and CTIO can form a pretty good network all by themselves during the long dry days of the Austral summer.

To Be Continued...

The Project is continuing with its plans to continue the observation phase for a full eleven-year activity cycle. Tracking the evolution of subsurface changes through the cycle has been identified as a key element of NSO's long-range plan and has been warmly received by NOAO. Detailed planning is also continuing for an upgrade of the camera system to a higher-resolution format in the 5122 - 10242 realm. Seeing studies using the current GONG instrument indicate that the present GONG sites have adequate seeing to support such observations, while studies using the GONG prototype show that the instrument optical system is also equal to the task. A recent test of the GONG data management system, using 1024 × 1024 images from the NSO High Degree Helioseismometer, demonstrated that keeping cadence with the larger data flow from such an instrument will indeed be achievable with the workstations conservatively projected to be available on the market in the year 2000. A formal plan for both the continuation and the new camera system is currently being developed.

John Leibacher


Next: Donald R. Farris: 1937 - 1997
Previous: First Observations with Fast 2K x 2K CCD Camera
Table of Contents - Search this issue - NOAO Newsletter Home Page

NOAO is operated by the Association of Universities for Research in Astronomy (AURA), Inc. under cooperative agreement with the National Science Foundation