MDI-SOI Observations and Observables

Observables

Spectral Intensity

There is one set of fundamental observables of the MDI: intensities at each pixel of the 1024*1024 CCD camera during an exposure. Each exposure can be made independently with any selection of the following parameters:

Wavelength Band

The center of the wavelength bandpass is tunable in steps of about 8 mÅ over a range of 377 mÅ centered on the Ni I photospheric absorption line at 6767.8 Å. The width of the bandpass is fixed at 94 mÅ.

Polarization

The incident light can be in any of four polarization states: s-wave, p-wave, right-hand circular, and left-hand circular.

Optical Imaging and Pointing

(see Fields of View below)

Camera readout time limits the rate at which successive filtergrams can be taken to about 1 every 3 seconds.

Individual filtergrams are not generally of much interest; because the telemetry bandwidth severely constrains the amount of data which can be recovered, virtually all the actual observables are derived quantities based on multiple sets of filtergrams taken under specific sets of parameter sequences. They all involve sets of filtergrams at five fixed wavelengths separated by 75 mÅ, denoted by I0, I1, I2, I3, and I4. The filtergrams are combined onboard by an image processor to produce the secondary observables, which can then be sampled, binned, and/or filtered depending on the observing program. Nevertheless, it should be borne in mind that recovery of individual filtergrams at nearly arbitrary wavelengths within the bandpass is possible.

Line Depth

A proxy for the depth of the Ni-I absorption line is computed as:

I_depth = sqrt (2*((I1 - I3)^2 + (I2 - I4)^2))

The per-pixel measurement uncertainty due to shot noise for a one-minute measurement involving 20 filtergrams is 0.7% (1 standard deviation).

Continuum Intensity

The proxy for the continuum intensity near the Ni-I absorption line of the measurements is computed from the standard five filtergrams according to the equation:

I_cont = 2*I0 + I_depth/2 + (I1 + I2 + I3 + I4)/2

The per-pixel measurement uncertainty for a one-minute measurement is 0.3%.

Doppler Shift (Velocity)

The Doppler shift is calculated as a tabulated nearly linear function of the filtergram difference ratios:

(I1 + I2 - I3 - I4) / (I1 - I3)

(I1 + I2 - I3 - I4) / (I4 - I2)

The per-pixel measurement uncertainty for a one-minute measurement is equivalent to 20 m/s.

Zeeman Splitting

The Zeeman splitting is determined by the difference between the Doppler shifts calculated from the filtergram components taken separately in right-hand and left-hand circularly polarized light. The per-pixel measurement uncertainty for a two-minute measurement is equivalent to 20 Gauss.

Fields of View

Full-Disc

A focused image of the Sun with a plate scale of 2 arcsec per pixel and a resolution of 4 arcsec. The center of the field is usually nearly coincident with the center of the solar image, with deliberate off-pointing of up to 9 arc-min in any direction and up to 13 arc-min in the north-south and east-west directions possible. The orientation is usually parallel to the solar rotation axis, with arbitrary deliberate roll possible (requiring spacecraft roll). Both off-pointing and roll disrupt the continuity of the Structure Program (see below) and require special scheduling.

High-Resolution Field

A focused image of an 11 arc-min square field of the Sun with a plate scale of 0.625 arc-sec per pixel and a (diffraction-limited) resolution of 1.25 arcsec. The center of the field is displaced 160 arc-sec to the north of the center of the full-disc field in the usual orientation. The same remarks as to off-pointing and roll apply as for the full-disc field.

Unfocused

An image of the telescope objective illuminated by sunlight formed at the CCD. This mode of observing integrated sunlight is intended for instrument calibration.

Dark

The shutter can be used to take dark frames for purposes of instrument calibration. There is also a front door which can be closed.

Binned, Smoothed, and Masked Fields

The onboard image processor can be used to combine observables from multiple pixels by binning into arbitrarily shaped regions, to select only certain subsets of pixels for telemetry, or to smooth the data by forming weighted averages. Data from multiple observations can also be combined with appropriate weighting functions to form smoothed temporal averages. The programming of these functions, other than masking and binning, is a complex procedure and generally reserved for the elements of the predefined Structure Program. It is possible in principle to use this capability for carefully designed and strongly justified campaigns, however.

Observing Programs

Structure Program

The Structure Program is a suite of observations designed to make use of the narrow telemetry channel, that is, observations involving comparatively small amounts of data, but for which long-period continuity is essential to the scientific analysis requiring them. All of the observations in this program are based on the Doppler Shift, Continuum, and Line Depth observables made from the five standard filtergrams of the full-disk image, repeated either twice (the 30-second program) or four times (the 60-second program) per minute. The 60-second Structure Program is run at all times when when no wide-channel telemetry is available, nominally 16 hours per day for 10 months a year. At other times the choice between the 60-second program and the 30-second program depends on which observations are being made to support the other observing programs. There is no important distinction between the near-real time and recorded telemetry intervals.

The Structure Program has the following components:

LOI Mask

Velocity and Continuum observables binned into a set of 156 fixed regions of roughly similar projected areas on the Sun, but selected so that they exactly subdivide the projections of the 20 pixels of the Luminosity Oscillations Imager (LOI) of VIRGO, one of the other SOHO helioseismology instruments. The cadence is one per minute, and these observations are recorded for redundant retransmission to assure absolute continuity. The Continuum is flat-fielded on board prior to binning.

Limb Figure

The Continuum observable in a mask of pixels approximating an annulus of thickness 10 arc-sec (five pixels) covering the solar limb. The individual pixel values are flat-fielded on board, then temporally smoothed under a 23-minute long Gaussian filter and sampled once every 12 minutes. The mask is modified from time to time as necessary so as to keep the solar limb approximately centered in the annulus. (For details of the temporal filter, see the Limb Figure Team Science Specification.)

Continuum and Line Depth Averages

The Continuum and Line Depth observables are averaged into 8*8 pixel bins. The averaged quantities are temporally smoothed under the same filter as used for the Limb Figure. Both observables are flat-fielded on board prior to binning. (For details of the temporal filter, see the GOLF Intercomparison Team Science Specification.)

Smoothed Velocity Averages

Most of the data associated with the Structure Program consists of the spatially smoothed Doppler data sampled once per minute. Initially, the data will be spatially averaged over the inner 90% of the solar disk into square bins of 12 arc-sec width with a 24 arc-sec (FWHM) two-dimensional Gaussian smoothing filter. We expect to modify the spatial sampling scheme after several months to a non-uniform binning in order to provide higher spatial frequency response in the equatorial and meridional directions. Different binning or averaging schemes may be attempted.

Dynamics Program

For two months a year nominally, sufficient telemetry is allocated to permit downlinking of two suitably compressed full images per minute continuously. During this time (60 days contiguous), one full-disc Dopplergram will be produced per minute. The second quantity will be a full-disc continuum map per minute during one month, and a fraction of a high-resolution Dopplergram per minute during the other month. During the month without high-resolution data the 60-second Structure Program is run concurrently, providing virtually redundant data. During the other month the 30-second Structure Program is run concurrently.

Magnetogram Program

The 5-minute bursts of real-time high-rate telemetry usually available three times per day (except during the Dynamics Program) are used to downlink full-disk magnetograms made and recorded once every 90 minutes (16 per day), and sets of single full-disk, high-resolution, and calibration mode observations of Doppler shift and continuum intensity, one set per burst.

Campaign Programs

During the approximately 10 months per year when continuous high-rate telemetry is unavailable, the daily 8-hour passes of high-rate telemetry are allocated to a variety of observing campaigns, mostly to be determined. Virtually any observations of which the instrument is capable are possible at these times, subject only to the constraints of the 160 kbps bandwidth, the onboard computational resources limiting compexity of the observables, and the necessity of accomplishing the standard 10 filtergrams in 30 seconds of each minute for the Structure Program (these filtergrams can also be used for the campaign observations). Campaign observations can be as short as a single frame and as long as eight hours.