The Global Oscillation Network Group GONG Experience

1
The Global Oscillation Network Group (GONG)
Experience

• SONG Workshop
• March 2006

2
The GONG Network
3
Brief History Of GONG 1

• Site Survey
• Model developed in 1985
• Simple pyrheliometer to measure cloud cover
• 15 instruments built for total cost of 75k
• Deployed in 1985, up to eight years of data per
  site eventually obtained
• Six sites selected in 1991
• Funding
• Proposal submitted to NSF in 1985
• Funded at low level 1986-1991
• Significant increase in 1991 allowed completion
  of development and deployment

4
Brief History Of GONG 2

• Operations
• Deployment started in January 1995, finished
  October 4, 1995
• Operations still continuing, one solar cycle
  nearly complete
• GONG is now an NSO Flagship Program
• Monthly (36 days) duty cycle 0.58 to 0.96,
  average around 0.87
• Instrument upgrades
• Original 256X256 rectangular-pixel CCD replaced
  by 1024X1024 square-pixel CCD in 2001
• Original polarization modulator being upgraded now

5
Example of site survey analysis
6
GONG Measurement Principle

• Fourier Tachometer concept developed by Tim
  Brown, Jacques Beckers Jack Evans
• Create an interference pattern via a polarizing
  Michelson interferometer
• Sweep the fringes in wavelength across a solar
  spectral line
• Measure phase of first Fourier component of the
  line, proportional to Doppler shift

7
GONG Instrumentation Details

• All custom built
• Entrance turret with 100-A wide filter for window
• Filter wheel with calibration targets,
  polarization modulator
• Calibration optics
• 5-A interference filter
• 1-A Lyot Filter
• Polarizing Michelson interferometer
• Spinning 1/2 wave plate
• Camera rotator CCD
• Custom data and instrument systems
• Shipping container shelter

8
More details
9
Major maintenance issues

• Electronics (Most common)
• Camera failures, particularly thermal control
• Power supply failures
• Mechanical
• Water in the turret
• Tape drive failures
• Optical (Rare)
• Moisture destroyed a Lyot
• Entrance windows degrade

10
Data Processing

• Data delivered via DLT tapes
• Processed in Tucson
• Complex reduction pipeline of calibration, image
  selection, spherical harmonic decomposition
• Approximately 2 FTE to operate global reduction
  steps
• 1 FTE for archiving, 2 FTE for local
  magnetogram processing, 2 FTE for development
• Hardware moving from Sun to Linux
• Approximately 60 TB of disk space

11
Data Processing 2

• Network requires data merging
• Probably simpler for SONG
• Spectral fitting is a central issue
• Stochastically excited oscillations create a
  challenge
• Multi-tapering helps considerably

12
GONG spectrum fit
13
How much does it cost to run?

• 600k/year for maintenance of six sites
• Salary support at sites
• Salary support of GONG operations team
• Travel
• Hardware
• General guide line is that annual operation cost
  is about 10-15 of initial construction.
• GONG also supports a substantial data processing
  center, and a scientific staff.

14
Operational challenges for SONG

• Choice of object
• Selection of observing run length
• Coordination
• Staff at the sites

15
Some Advice

• Select the sites to form two interleaved networks
• Dont scrimp on the resources to maintain the
  network
• Prepare SONG to last for a long time
• All software must be fully documented
• Invest up front in maintainable hardware
• Minimize the impact on the host sites
• Automate as much as feasible (but it is
  impossible to fully accomplish)