azexhdf -h a195041b.hdf -BSe a195041b_test.raw

Should produce an envi .hdr for a bseq file (azexhdf help states:

  -B  fn      : convert image data to BIL binary file
  -BS fn      : convert image data to BSEQ binary file
  -Be fn      :
  -BSe fn     :
              :  B and BS may be followed by  e  to request an ENVI
              :  compatible header to be output with extension fn.hdr
              : For ENVI compatibility fn should be given as  fn.raw or fn.bil
              : fn  = output filename

Actual header file produced claims it's a BIL, not a BSEQ: arsf@pmpc889:~/support/20070515-GB06_05-195a-RachelGaulton<1040>cat a195041b_test.hdr

ENVI
description = { File output by azexhdf }
file type = ENVI
interleave = bil
samples = 938
lines = 3802
bands = 11
header offset = 0
data type = 12
byte order = 0

Correcting the interleave line to read "bseq" results in a correct load into ENVI.

Currently obtained via requests to Andrew Wilson.

We may be able to gain access via the Pan-Government Agreement, according to Tony Clifton @ BGS. We need to sign up at ​http://www.iggi.gov.uk/signUp.php to see if we're eligible.

Steve is doing this.

Just a tracker ticket for an issue found in #2. azgcorr wasn't handling the SCbedit vector correctly when BSEQ/BIL files were used.

Reported to Bill 18/May and fixed over the weekend in release 106. Some notes from Bill's email:

Updated azgcorr.481 by separate email.

SCbedit now works as advertised with the limitation that it only works for the first 16 bands of a data set (because it uses an unsigned int). It was originally only put in for the ATM anyway.

The "distortion" was because I had forgotten for BIL/BSEQ input to move the file when a line was skipped, so after each skipped line the image would shuffle up.

So now for CASI and Specim data the only option is the whole frame is skipped if it is flagged. This makes sense because with a CCD sensor it would be pretty unlikely to have a whole row dead, columns maybe but not rows.

The default in azgcorr.48* is if there is NO -be  the SCbedit vector is used;  if there is a -be then SCbedit is not used and no lines/scans are ignored. 

This needs testing