The Sun's photospheric line-of-sight magnetic field is observed every 96 minutes as part of a synoptic program. The MDI instrument computes the magnetic signal on board the SOHO spacecraft by differencing Dopplergrams obtained in right and left circular polarized light. The synoptic series includes both one-minute and five-minute averaged full-disk magnetograms. The per-pixel noise level in the calibrated magnetograms is ~30G for a 1-minute magnetogram and ~15 G for a 5-minute observation. The noise level is not uniform because of instrumental filter irregularities. Depending on the orientation of SOHO one quadrant generally has a higher noise. In nominal roll conditions the lower right part of the observed image is more noisy because of Doppler signal leakage.
SOHO is always aligned so that the solar rotation axis lies precisely parallel to columns of the MDI CCD camera. For the first 8 years of the mission solar north was always at the top of the image, i.e. the P_ANGLE was zero. Since 2003 SOHO periodically rolls four times per year by very close to 180 degrees to accommodate a problem with the high-gain antenna pointing mechanism. During such intervals the P_ANGLE will be close to 180 degrees and the south solar pole will be at the top of the image. Currently users must themselves correct for the P_ANGLE.
Downlinked data are reconstructed as standard FITS file images. Numerous image characteristics and observing parameters are recorded in FITS header keywords. Level 1.8 processing provides corrections for plate scale, zero offset, sensitivity, and the most severe cosmic rays. Beginning in December 2008 Level 1.8 magnetograms are being stored as FLOATs for the first time. Level 1.8 corrections are described in a paper being submitted shortly. [PREPRINT]
Other details can be found at
Lev 1.8.2 FITS files store data values as FLOATs for the first time.
The SIKETHR for cosmic ray elimination was set to 5000, probably too high to be very useful. Note that virtually every magnetogram is contaminated by at least a few cosmic rays. Sometimes the contamination is obvious, but most often the corrupted pixels are difficult to identify.
The Lev1.8.1 data have been corrected for spatially dependent sensitivity variations as determined by Tran et al. [ApJS, 2005], who made careful comparisons with data obtained in several spectral lines at the Mt. Wilson Solar Observatory. The correction depends on both solar and instrumental effects. MDI is sensitive to the spatially averaged line-of-sight magnetic field in each pixel, for convenience reported as Gauss. The sensitivity of the calibrated Level 1.8.1 MDI magnetograms is more spatially uniform. The reported values have increased by a spatially variable factor of approximately 1.8 when compared with previous uncalibrated MDI magnetic data (Level 1.5 or earlier Level 1.8.0). [This sensitivity correction will be changed in Level 1.8.2.] The FITS keyword SRC_ADJ2 indicates the sensitivity map used in the calibration. This keyword did not exist in previous versions of Lev1.8.0.
The new data have been corrected for a small instrumental zero level offset. The value BFITZORG has been subtracted from each pixel. Note that as always your FITS reader must properly use the FITS keywords BSCALE and BZERO for data stored as scaled integers. The zero offset is due to small MDI shutter variations and is determined by performing a fit to the low-field values. See Liu, Zhao, & Hoeksema, Solar Physics 219, 39-53, 2004; DOI: 10.1023/B:SOLA.0000021822.07430.d6 .
A very few obvious pixels with obvious cosmic ray contamination have been removed. The keyword SPIKECNT gives the number of pixels above the threshold indicated by SPIKETHR. SPIKETHR is currently 2500 in uncalibrated units. MDI saturates well below this level. Because a cosmic ray contaminates only one of many MDI filtergrams that contribute to the magnetogram, most cosmic ray contamination cannot be simply and automatically detected in a single magnetogram.
The known magnetic saturation in strong field regions above about 3500 G (2000G uncalibrated value) has not been corrected. In strong field regions, such as sunspot umbrae, the MDI signal is saturated and actually becomes weaker. This is primarily a consequence of limitations of the on-board processing algorithms. See Y. Liu, A. Norton, & P. Scherrer, Solar Physics 241, 185-193, 2007; DOI: 10.1007/s11207-007-0296-5 .
The Level 1.8.0 images have a time-dependent image-scale correction applied as indicated in the keywords: FD_SCALE, IM_SCALE, XSCALE, YSCALE, R_SUN. The keyword FDSCALEQ indicates the quality of the correction: 0 means no correction was done, 1 means a correction using extrapolated coefficients was done and more up-to-date coefficients may result in a better correction, and 2 means the correction was done with the final set of coefficients. See the file scale_corrections for more details.
Synoptic charts have been recomputed using the December 2008 re-calibration.
At the same time as the October 2007 recalibration, significantly improved high-resolution synoptic charts of the solar magnetic field were released. The new charts were constructed using the recalibrated MDI magnetograms. Input data selection has been optimized to provide a significantly reduced and much more uniform noise level across the map. The equivalent of 20 individual one-minute magnetograms now contributes to each synoptic chart pixel. In constructing the new charts the original magnetograms have been remapped taking into account the affect of differential rotation. Statistical tests have been used to reduce contamination from cosmic rays and other single-image anomalies. The new synoptic maps have also been corrected in an improved way for geometric projection due to the location of the SOHO spacecraft. The correction has been made assuming that the field at the photosphere is radial in the photosphere where the observations are made.
Of course the magnetic field at each pole is still observed only during the part of the year it is visible. Optional polar field corrections to the synoptic charts will be made available.
A new version of synoptic map showing the presumed radial field has also been produced.
The synoptic magnetic time series provides a magnetogram every 96 minutes, 15 times per day. Calibrated magnetograms from this series are use to compute synoptic maps of the Sun's magnetic field. The magnetogram data set is sometimes referred to as the 96m (for 96-minute) data series.
To provide an uninterrupted record, 96m magnetograms are collected on board the spacecraft on a regular cadence and stored until they can be downlinked. There are two basic kinds of on-board magnetograms, one-minute and five-minute. Five-minute sums are only collected for the 96m program. Depending on the MDI observing schedule, several magnetograms of each kind may be collected on any given day. A keyword in the FITS header describing the data product code, DPC_OBSR, indicates which kind of magnetogram the data file contains. DPC_OBSR='FD_Magnetogram' indicates a one-minute magnetogram, while DPC_OBSR='FD_Magnetogram_Sum' indicates a 5-minute onboard averaged magnetogram. The summed magnetograms have lower noise and different systematic errors. See Liu et al. (2004) for more detail.
In the thumbnail plots of the 96m magnetogram time series, five-minute summed magnetograms are labelled with an asterisk.
MDI can make magnetograms more frequently during campaigns. Extended sequences of one-minute magnetograms are grouped into hourly chunks, and are commonly referred to as campaign or 01h magnetograms. These magnetograms are available for export
Page last revised
Friday, 18-Jun-2010 13:19:52 PDT