SOI Logo Known Problems in Calibration of All MDI Data Products

· Image Quality ·

Datasets Affected

Description of Problem

There are several issues which affect MDI image quality. Topics included in this section include focus and MTF. In addition to the diffraction component of the MTF MDI full-disk images and High Res images have focus variations across the image. Due to thermal variations and front window variations these effects change with time.


The MDI full disk and high resolution images have different focus and MDT (Modulation Transfer Function) issues. The as-built optical configuration is set such that the full disk and high resolution images can not both be in focus with the same focus setting of the instrument. Focus is adjusted in MDI by inserting flat disks of glass with different index and thickness. There are two focus wheels each with three focus positions making a total of nine focus settings. By design the high resolution image is in focus two focus settings less than the full disk, where less is in the direction of less added optical path length. This was done because the imaging optics were sized for diffraction limited imaging at 0.6 arc-sec while the full disk pixels are 2 arc-sec. In order to reduce p-mode aliasing in an undersampled image we reduce the quality of the full-disk image by observing out of focus. In order to eliminate the need for changing the focus setting twice each minute, we set the instrument adjustments to have one image two steps out of focus when the other is in focus.

MDI actual focus is monitored on a regular basis and the results of a quick analysis are on the "MDI Calibration Page" accessible on the Operations page.

The following discussion concerns the full-disk image primarily. A detailed analysis of the high-resolution image at the beginning of the mission was made by XXX and published XXX. The high resolution image focus varies with time in step with the full disk focus discussed here.

A quick examination of these data shows that the best focus position of the instrument not only changes with time but is not uniform across the image. There are three separable issues: average focus change with time, East-West asymmetry of the focus map, symmetrical part of the focus map.

The change with time of the mean focus is due to the change in thermal conditions of the front window.

The East-West asymmetry is probably due to a tilt of the CCD chip away from normal to the optic axis of MDI.

The symmetrical component is probably due to the curvature of the image surface.

The combined shape of the focus map is very stable with time. The average position has two components, an annual term consistent with the SOHO-Sun distance variation and a drift term consistent with the average drift in temperature of the window and drop in instrument throughput with time.

The MTF or Point Spread Function (PSF) information is difficult to measure in flight for the full disk image. It can be estimated at the limbs and at the limbs when the image is off-center, but these data do not provide a full coverage of the image plane. It can be estimated from the method of XXX, but that method counts of pixel to pixel variation in the observed signal and the full disk pixels are a bit too large to use granulation for this purpose as was done for the high resolution field.

Nevertheless we have estimates for the PSF which shows XXX. We note that the PSF is not azimuthaly symmetric.

· Plate Scale ·

Datasets Affected

Description of Problem

There are several issues which affect MDI image plate scale. Topics included in this section include plate scale and distortion. MDI full-disk images and High Res images have plate scale variations with focus and distortions. Due to thermal variations and front window variations these effects change with time. The net effect of the focus and distortion can be more than a pixel in effective location vs tabular location on the solar disk.

The MDI data headers as of this date (20 Nov 2000) all show image scale parameters assuming that the basic plate scale is 1.97784 arc-sec per pixel. This is not only wrong but wrong in the 4th place. A better value is 1.982 +- 0.003. This number changes slowly for known reasons and will be corrected.


Items to note here: Plate scale changes with focus setting of the instrument and with the difference between the focus setting and the actual focus at each time. Thus there are combined changes due to the annual part of focus variation, due to the secular trend in focus, and due to commanded changes in focus position.

Each focus step changes the plate scale by about 0.001 or a tenth percent, or a half pixel in image radius. The sense is that the image gets larger when the focus setting is larger than the actual focus and the images is smaller when the setting is "less than" the actual focus. Since the nominal focus setting is two stops less than best focus in the full disk case, the image size tends to larger images.

The distortion of the image is probably close to that calculated from ray-trace analysis of the MDI optical design. Curves for this distortion are available in the "MDI Instrument Drawings" link on the Operations page.

The net effect of the distortion is that if one uses the observed limb to set the image plate scale, there will be a position error of about 0.3 pixels about half way between disk center and limb, with the actual location on the Sun being nearer the center than the apparent position.

General Info | Science Objectives & Proposals | Results, Images, & Documentation | Data Services & Info | Operations | Other Links | Press Releases | Solar-Center | Stanford Solar Group

Page last revised Monday, 20-Nov-2000 19:06:38 PST