Home  >  Solar Archives  >  STEREO  >  STEREO data  >  HI processing documentation

STEREO HI data processing documentation

Unlike most camera systems, the HI cameras have no shutters, therefore HI exposures are smeared vertically, both as the CCD is cleared prior to an exposure and as the CCD is read out. A small number of single HI camera exposures (one exposure per 24 hours) are sent to the ground with no processing and at the full 2048*2048 pixel resolution of the camera CCD but most are scrubbed for cosmic rays, binned to 1024*1024 resolution and then summed on board the spacecraft before being downlinked [30 exposures are summed into a single HI-1 image and 99 exposures are summed into a single HI-2 image]. An estimate of the DC bias of the CCD is also subtracted from the binned, summed image. For more detail on this on-board processing see "The Heliospheric Imagers Onboard the STEREO Mission, Eyles, C.J. et al, Solar Phys. 254. 387 [ DOI 10.1007/s11207/008-9299-0 ] and in particular section 10.1.2.

In order to optimise the communication to the ground these 32-bit summed images are then split into two 16-bit images (high word and low word). HI data are transmitted using 'Rice' compression - a lossless technique. Image parameters and spacecraft status data are also downlinked. Note that some images are transmitted using a highly compressed "beacon mode", these are primarily intended for space weather alerts.

HI Level 0 data

The first stage of the production of usable data is conducted at NRL. The split summed images are recombined and coded values from the spacecraft and instrument housekeeping and status parameters are turned into physical values. Fits files incorporating these values are then generated.

The output is written as "fits" files with a naming convention  yyyymmdd_hhmmss_LATTS.fts.
yyyymmdd_hhmmss The nominal start time of the exposure/summed image
LA n4 - un-scrubbed single exposure at 2048 x 2048 pixel resolution
s4 - scrubbed summed images at 1024 x 1024 resolution [normal science image]
s7 - scrubbed summed highly compressed image
TT h1 - hi_1 camera
h2 - hi_2 camera
S A - STEREO-A spacecraft
B - STEREO-B spacecraft

The summed s4 images also include extended fits headers that provide header information for all the single exposures in the sum.

Some of the pixels in a s4 image are replaced with a report from the cosmic ray scrubbing routine. This affects 31 pixels in a HI-1 image and 100 pixels in HI-2 images - when conventionally plotted these pixels are in the lowest row at the far left on the STEREO-A images and in the highest row at the far right for STEREO-B images.

Occasionally, as better determination of the spacecraft ephemeris data become available the Level 0 HI data is refreshed.

HI Level 1 data

The HI Level 1 data provided by the UKSSDC are derived from the aforementioned s4 and n4 files and again are stored as "fits" files with their naming convention as above except for the "LA" field in  yyyymmdd_hhmmss_LATTS.fts.
yyyymmdd_hhmmss The nominal start time of the exposure/summed image
LA 14 - units of DN/sec/pixel
1b - units of mean solar brightness (MSB) - diffuse source correction applied
1t - units S10 [flux of 10th magnitude star of solar spectral type distributed over 1 square degree]- diffuse source correction applied
TT h1 - hi_1 camera
h2 - hi_2 camera
S A - STEREO-A spacecraft
B - STEREO-B spacecraft

Processing L0 images to L1 involves a sequence of steps

  1. Rejecting corrupted images - a list is maintained of images that were badly corrupted in processing and transmission - these images are not processed to level 1.
  2. Images from the behind spacecraft are rotated 180° so that solar north is towards the top of the images.
  3. Extract cosmic ray report (present in s4 images only). The pixels are replaced by a duplicate of the adjacent row to allow subsequent processing.
  4. Removal of columns where saturation has occurred on the CCD - due to the HI cameras not having shutters it is not generally possible to recover columns where saturation has occurred. Bright stars and planets often cause saturation to occur. Values in the affected columns are replaced by "NaN",
  5. De-smearing the image - by applying an algorithm for matrix multiplication remove the smearing that occurs during CCD clearing and readout. This process automatically normalises the total image exposure time to 1 second and removes the exposure summing effect in the s4 images.
  6. Removing pixels marked as having missing telemetry from the image.
  7. Applying a calibration flat field to the image - this allows for the geometric effect in the lenses and for a linear production feature in the CCD. Additionally for some products, photometric calibration is applied to yield data in units of MSB and S10; note that for these products a correction is also made for the differing solid angles on the sky corresponding to pixels across the CCD.

At RAL another step is performed to fit the observed image to a known star map to derive optimal pointing information for the cameras on a image-by-image basis. Data from this fit is distributed via solarsoft updates and gets included in the processing above where available.

HI Level 2 data

Derived from the HI Level 1 data, the Level 2 data provided by the UKSSDC has a background removed (that mainly includes the contribution from the F-corona). The file naming convention has been expanded as yyyymmdd_hhmmss_LATTS_bbll.fts.
yyyymmdd_hhmmss The nominal start of the exposure
LA 24 - units of DN/sec/pixel - point source calibration used
2b - units of mean solar brightness (MSB) - diffuse source calibration used
2t - units S10 [flux of 10th magnitude star distributed over 1 square degree]- diffuse source calibration used
TT h1 - hi_1 camera
h2 - hi_2 camera
S A - ahead spacecraft
B - behind spacecraft
bb br - running window background estimate, background computed as mean of lowest quartile of data in the window
ll Length of window (in days) used to compute background
`

Normally 1 and 11 day backgrounds are removed for hi-1 and 3 and 11 day backgrounds are removed for hi-2.

Further Information

A worked example of these processing steps can be found here