Opened 8 years ago

Last modified 8 years ago

#220 new Data problem

2009 flights affected by varying offset

Reported by: benj Owned by: benj
Priority: alpha 5 Milestone:
Component: Processing: general Keywords:
Cc: jatt, mggr, mark1 Other processors:

Description

VOCALS 2008 flights 7-14 all have a small offset between Eagle and Hawk. Appears to be a boresight offset but is not consistent between flight days (or even between flight lines).

Coincidentally, Specim were asked to perform a calibration on the Eagle and Hawk after return from Chile, and they found an offset in the Eagle that appeared to have been caused by an external shock (though they couldn't say when it had occurred).

Email from Markku Koskela to Phil Goy below (edited 21/Sept/2009 per Specim's request to remove direct quotes - a summary has been inserted instead)

  • a spectral and spatial tilt was found in the Eagle sensor
  • the corner of the sensor looked like it had been impacted or stressed
  • a recalibration was performed and showed a 0.5nm difference at the edges of the image (+/- 0.25nm max change)
  • other damage noted was that the Hawk data cable pins were bent (replaced) and the power supplies / computers were loose and moving freely (fortunately no damage occurred).

Working hypothesis at the moment is that the sensor casing received a jolt between data collection for flight days 6 and 7 (hard landing, something bashed it, whatever) that jolted the Eagle slightly loose. This resulted in it moving around slightly (and slowly, or possibly only on hard turns such as between flightlines) during flight. Since the movement was only fractions of a degree, this was not detectable until the data was geocorrected.

Change History (3)

comment:1 Changed 8 years ago by benj

  • Cc mark1 added

Offset also affects flight days 1-6.

No longer convinced that this is a simple boresight offset - the size of the offset appears to remain fairly constant as the height of the aircraft (or the ground) is altered, you would expect it to change.

It also doesn't appear to be any of the following:

  1. Timing error - there is (for the most part) none of the distortion usually associated with this and trying to fiddle with the timing introduces distortion. There may be slight timing offsets on some lines, but not on most.
  1. Height changes in the aircraft or cloud layer (ie the clouds aren't on the ground) - as mentioned above, the offset remains constant as the height of the aircraft or clouds is varied. The offset also applies to the areas of the flightlines over water, which it shouldn't if it was a result of projection effects of treating the cloud as if it was on the ground.
  1. Differential penetration of the clouds by different wavelengths. There is no offset between short-wave and long-wave Eagle, or short-wave and long-wave Hawk, only between sensors. The offset is also a shift, not a size change, and the same offset that would "fix" a particular line this way fixes both cloud and water underneath (you'd expect it to throw the water off).
  1. Navigation errors - this would affect both sensors equally.
  1. Lever-arm offset - the offset is too big and not constant.
  1. Movement of the cloud under the aircraft between times when each sensor views it - still might just be this (if the plane is flying slowly, or is a long way above the clouds, it becomes slightly more plausible), but it looks like the clouds probably don't move fast enough to make this work, and the pitch offset probably isn't great enough.

comment:2 Changed 8 years ago by benj

  • Summary changed from Some VOCALS flights affected by varying offset to 2009 flights affected by varying offset

Issue also affects the rest of VOCALS as well and all data up to 2009 day 131 (when instrument was returned from SPECIM). SPECIM didn't fix this issue deliberately, but we think they probably tightened something somewhere that stopped the boresight drifting.

Have changed the title of the ticket to reflect this.

comment:3 Changed 8 years ago by benj

Should have said all 2009 data up to day 131. Data prior to deployment to Chile for VOCALS in 2008 is unaffected.

Note: See TracTickets for help on using tickets.