SOI Logo The MDI Resident Data Archive

Introduction

The Michelson Doppler Imager (MDI) was designed to make spatially resolved measurements from space of the Doppler and first-order Zeeman shifts of the solar photospheric line Fe I 6768 Å. It was one of the suite of instruments comprising the Solar and Heliospheric Observatory (SoHO), a joint ESA-NASA deep space mission launched into a halo orbit about the Sun-Earth L1 Lagrangian point on 2 Dec. 1995. Scientific observations nominally began around 1 May 1996, although various forms of test and calibration data were acquired beginning as early as 16 Dec. 1995 (first light was on the 19th). Apart from two extended interruptions from 24 June – 20 Oct. 1998 and 21 Dec. 1998 – 1 Feb. 1999 due to loss of spacecraft control, the instrument operated essentially continuously, until being powered off on 11 Apr. 2011. In all, approximately 33 million observations were acquired, processed, and archived during that period (see Table 1). The instrument remained capable of resumed operation at the end of the mission, and it has indeed been briefly and successfully powered on to acquire data acquired since that time.

Types of Data

In order to understand the variety of data types acquired by MDI, it is important to have an idea of the way the instrument operates and the way the data are acquired. This is a very rough summary. For further details, refer to Scherrer et al., 1995.

The MDI operational principle is to combine multiple filtergrams taken in several very narrow passbands selected by a tunable filter across and slightly beyond the wavelength width of the absorption line in order to estimate the Doppler and/or Zeeman shift of the line as well as the line depth and the continuum intensity in the immediate vicinity of the line. For each observation, filtergrams from four tuning positions across the line are used. A continuum intensity proxy is estimated in one of two ways: either as the base of the overall line profile fit for the Doppler shift, or simply as the intensity of a filtergram in a fifth tuning position further from line center than the others. There are thus a total of five “observables”: Doppler shift (V), Zeeman shift (M), Line depth (Ld), fit continuum intensity (Ic), and off-line filtergram intensity (I0).

The MDI telescope has an optical resolution of about 1.4 arc-sec. Critical sampling at this resolution (0.7 arcsec/pixel) would require a camera diameter of nearly 3000 pixels to record individual filtergrams of the entire photospheric image. MDI has a 1024×1024 pixel camera. The camera can be fed from either of two light paths, however. In one, the image is demagnified so that the entire disc of the Sun fits within the CCD, with a resolution of approximately 2 arc-sec per pixel. This is called “full-disc” (fd) mode. In the other, the image is at full resolution, about 0.7 arc-sec per pixel, but a region of only about 700 arc-sec squared, about 10% of the disc, is observed. This is called the “high-resolution” (high-res; hr) mode. The location of the high-resolution region is fixed. When the spacecraft is oriented in its nominal position with one axis aligned with the solar (Carrington) north axis, the high-resolution region is approximately centered in the solar east-west direction, but displaced to the north so that about 75% of the region is in the northern hemisphere. Of course this flips when, as occurred for half the time during the last nine years of the mission, beginning in the summer of 2003, the spacecraft is aligned with the south axis. (This was due to a failure of the spacecraft's high-gain antenna pointing mechanism.)

The MDI camera acquires one filtergram approximately every 3 sec. The telemetry bandwidth was vastly lower than needed to transmit each filtergram, so it was necessary to perform the observables calculations onboard. In a “high-rate” telemetry mode of 160 kbps, normally available for only about 8 hours per day, it was possible to download two observables for the full disc or most of the high-res field per minute. A “low-rate” telemetry mode of 5 kbps, nominally available continuously either directly or from dumps of recorded data, allowed continuous acquisition of selected observables at either greatly reduced spatial or temporal resolution, or both, via onboard binning or weighted averaging. Six so-called “Structure Program” products are available essentially continuously throughout the mission, with the obvious exceptions of the two extended spacecraft outages:

The onboard processing capability for extraction, spatial binning, and temporal summing also made it possible to design additional high-rate data products in order to enhance the number or extent of the observables acquired in a single minute. In particular, there are versions of several of the full-disc and high-res observables involving onboard 2×2 binning to effectively reduce the nominal pixel resolution by a factor of 2, to 4 or 1.4 arcsec/pixel, respectively. Additionally, there are various data products for which only a rectangular subgrid of the original observable image is extracted, the location and size possible varying from image to image. Finally, there are a few data products characterized by distinct sampling frequencies as discussed below.

Mission Phases

It is convenient to divide the MDI mission into seven phases defined as follows Phase 0 represents the period prior to the beginning of the Structure Program and the acquisition of data for the first of its observables (loi_V). It was only during this phase that observables incompatible with the Structure Program, notably those requiring a higher cadence than 30 sec, were acquired. Phase 1 is the 32-day period during which various other of the Structure Program observables began to be acquired. Phases 2–4 constitute the regular mission, with the two lengthy interruptions between phases caused by loss of contact with or regular telemetry from the spacecraft (the “SoHO Vacations”). During Phase 5 data from the Structure Program continued to be acquired, but virtually all regular high-rate observations were abandoned. (There was an additional spacecraft-caused substantial interruption in the Structure Program data during this phase in 2011 from 22:52 on Jan. 11 through 21:11 on Jan. 28.) Post-mission (Phase 6) observations have been made and data acquired during one interval, 6–15 Mar. 2012. None of these data have as yet been processed through to Observables.

Observing Cadence, Sampling, and Integration Times

The normal observing cadence was designed to produce one image of any given observable per minute. There are a few exceptions: three of the Structure Program observables rwbin_Ic, rwbin_Ld and limb_figure) are sampled once every 12 min, and one of the two types of magnetograms requires 5 minutes integration. In addition, there were occasions in which.observables could be sampled more frequently by interrupting the regular observing program; most of these were obtained during the checkout and calibration phase of the mission (Phase 0).

Most of the observables can be produced from appropriate combinations of individual filtergrams in different wavelengths and polarizations acquired over an interval of either 30 or 60 sec. (The individual filtergram exposure times are less than a second.) During the regular mission (Phase 1 onward), when the Structure Program was running, the “upper half” of each minute, i.e. from 15 sec before the minute to 15 sec after, was always devoted to the observations required to produce the relevant observables (V, Ic, and Ld), that is, filtergrams in linear polarization at the standard wavelengths and in full-disc mode. The lower half of each minute could be devoted to either the same observations, or to those required for magnetograms (using the circular polarizer), or to any observation in high-res mode, or to calibration measurements. When the circular polarizer was in use during the lower half of the minute, Doppler and intensity measurements, even if made in full-disc mode, were not used for the Structure Program observables, because of optical displacements caused by the circular polarizer, degrading the per pixel onboard integrations.

Sampling requirements for various types of data are quite different, depending on whether their primary purpose is helioseismology, focused campaign studies, or regular monitoring. Structure Program data were sampled as nearly continuously as possible at their nominal cadence. High-rate data were sampled either continuously in dedicated blocks of time, regularly at rates of one to fifteen per day, or only sporadically. For data expected to be continuous, there exist data records (metadata) for each time step, regardless of whether the associated image data for that time step exist or not. This allows for determination of actual data coverage (duty cycle) during any given interval and for the extraction of continuous metadata parameters not dependent on the presence of actual image data for the target time, such as spacecraft position and velocity. (The exception is that there are no metadata records during the SoHO vacations.) For data that were sampled continuously only in dedicated blocks of time, the metadata are filled continuously only within blocks of a full hour or full day (depending on the series) for which some image data are present. For sporadic observations, metadata records exist only for times at which real data are present. These three cases are denoted by C (continuous), H or D (depending on the block length, hourly or daily), and S (sporadic), respectively,

Observable Data Series Naming Scheme

Combining all the images of the same observable, nominal resolution, and target cadence into a data series, there are 29 such series, as summarized in Table 1. The naming scheme is generally straightforward, combining a code for the nominal resolution with the observable code and any special features. The primary exception is the limb_figure series of Ic measurements at full-disc resolution, but cropped to an annulus a few pixels wide.

Table 1: MDI Level 1(.5) Observations
Coverage during mission phases (For Phase 0, image counts are shown rather than fractional coverage)

series	  metadata    images	coverage during mission phase			Data Volume
	      fill		  0	1	2	3	4	5	    (GB)

fd_I0		 H    46 751	  0	0	.0362	.0098	.0009	.0005	  33.910
fd_Ic		 H   284 183    279	.0060	.1291	.0782	.0167	.1073	 259.928
fd_Ld		 H    95 234	  0	.1039	.0811	0	.0001	0	  83.144
fd_M_lev182	 C 2 588 296	  0	.0001	.0711	.0684	.4048	.1046	 102.830
fd_M_96m_lev182	 C    83 700     49	.7584	.9600	.8736	.9216	.7296	(102.830)
fd_V		 C 3 071 162    449	.1077	.3678	.0200	.4354	.0024	2875.291
fd_Ic_30s	 H
fd_V_30s	 H
fd_I0_extract	 H    25 761	  0	0 	.0118	.0280	.0016	0	  10.413
fd_M_extract	 H    82 523	  0	0	.0468	.1496	.0028	0	  88.377
fd_V_extract	 H    82 595	  0	0	.0469	.1493	.0028	0	  61.609
fd_Ic_bin2x2_30s H   160 004	  0	0	.1440	.0001	.0003	0	  40.721
fd_V_bin2x2_30s	 H    17 262	  0	0	.0148	0	.0001	0	   4.649
fd_V_bin2x2	 H    11 890	  0	0	.0107	.0002	0	0	   3.332
hr_I0		 H   490 279	  0	.0344	.1286	.2088	.0536	.0027	 250.576
hr_Ic		 H       985	  0	0	.0008	0	0	0	   0.522
hr_Ld		 H     2 059	  0	0	.0018	0	0	0	   1.321
hr_M		 H   754 679	  0	.0389	.1503	.2132	.0928	.0054	 794.322
hr_V		 H   699 941     25	.0308	.1672	.1926	.0814	.0011	 433.529
hr_V_12s	 H    22 725  22725	0	0	0	0	0	   6.645
hr_Ld_bin2x2	 H     1 894	  0	0	.0016	0	0	0	   0.555
hr_M_bin2x2	 H    11 727	  0	.0332	.0086	.0001	0	0	   4.609
hr_V_bin2x2	 H   196 997	  0	.0051	.0474	.0002	.0236	0	  61.789
limb_figure	 C   607 968	  1	.7440	.9792	.9084	.9564	.8616	  98.877
loi_Ic		 C 7 397 013	  0	0	.9912	.9045	.9750	.9002	   5.125
loi_V		 C 7 444 451	  0	.9692	.9911	.9044	.9750	.9002	   5.158
rwbin_Ic	 C   618 376	  1	.7500	.9864	.9156	.9744	.8688	  18.219
rwbin_Ld	 C   618 425	  1     .7632	.9864 	.9156	.9744	.8688	  18.182
vw_v		 C 7 403 494	  0	.6564	.9871	.9161	.9731	.8680	1082.303
Note that the series fd_M_96m_lev182 is merely a regular cadence subsample of the series fd_M_lev182, and since the individual magnetograms of the latter series are collected in daily directories, the total data volume counts are identical. Magnetograms during phase 0 were only included in series fd_M_96m_lev182.

An additional five series of “observables” have been defined to include various types of observations acquired for purposes of instrument calibration. They are summarized in Table 2.

Table 2: MDI Level 1.5 Calibration Data
Image counts during mission phases

series	  metadata    images	images during mission phase			Data Volume
	      fill		  0	  1	    2	    3	    4	    5	    (GB)
detune		 S    97 800	  0	 72	22833	 6507	65948	 2440	   6.103
focus		 S    15 882	  0	 91	 1701	 4796	 8814	  480	   1.873
ice		 S    30 420	  0	177	 3029	  529	25412	 1273	  18.888
lp		 S    
lp_bin		 S    
Links to Observables (Level 1) Data Series Descriptions
“Resolution /
Field of View”
Doppler
LOS Velocity
Continuum
Intensity
LOS Magnetic
Flux
Line DepthCalibration
Full-Disc fd_V,
fd_V_30s
fd_I0,
fd_Ic,
fd_Ic_30s
fd_M_lev182,
fd_M_96m_lev182
fd_Ld
Full-Disc extract fd_V_extract fd_I0_extract,
limb_figure
fd_M_extract
Full-Disc binned fd_V_bin2x2,
fd_V_bin2x2_30s,
vw_V,
loi_V
fd_Ic_bin2x2_30s,
rwbin_Ic,
loi_Ic
rwbin_Ld lp_bin
High-Res (including extract) hr_V,
hr_V_12s
hr_I0,
hr_Ic
hr_M hr_Ld
High-Res binned hr_V_bin2x2 hr_M_bin2x2 hr_Ld_bin2x2
mixed / NA detune,
focus,
ice,
lp

Level 0 Data

Level 0 data, derived from decoding and decommutation of telemetry in the science data channels assigned to MDI, represent the original data sent by the onboard image processor. Along with the instrument and spacecraft housekeeping data and the spacecraft orbital and attitude data, they are the sole and definitive source from which data representing observables can be inferred.
Level 0 Data Documentation


References


Valid HTML 4.01 Transitional this page last modified 1 Sep 2017, 16:32-0700