Hawk data very noisy in some bands

[benj]

Some Hawk bands are noisy to the point of uselessness. For example band 71 (1403nm), raw data from Nigg Bay line 3 (attached)

Also see over-water spectrum produced in level 3 image:

Spectrum cannot really be used for anything much - there is no means of quality controlling the data

Data issue occurs over more than one flight, possibly there are sensor problems with some Hawk bands. See eg. Latera band 71, similar to Nigg Bay:

[benj]

Hawk 8-pixel average spectrum over water in Nigg Bay

[benj]

Hawk band 71 (line 1) raw data from Latera

[benj]

Phil Goy said that he thinks the noisy hawk bands might be because these bands coincide with water absorption maxima - water absorbs most of the signal, so the signal to noise ratio is low for these bands. He's a bit concerned about the spikes in the spectrum though - those aren't explained by water absorption and look more like sensor underflows. He's going to send the Hawk calibration data down for us to see if that does the same thing.

[benj]

8-pixel average over-water spectra from the same pixel on two different scan lines (raw data)

[benj]

Spectra from different pixels on the same scan line (8-pixel average, raw data)

[benj]

Spikes seem to occur in the raw data as well as the processed data, though not as many (trying to integrate bad data may cause more spikes in the processed data). If you take two points in the raw data from the same pixel on different scan lines they both have very similar spikes, suggesting bad pixels for specific bands on the sensor itself, eg:

If you take two points in the raw data from different pixels on the same scan line, they both have spikes but in different places:

This suggests that most spatial pixels on the Hawk CCD have at least one bad spectral pixel, but that it's not always the same spectral pixel that is bad.

[benj]

Issues of azspec causing spiky spectra and bad pixels/calibration have been hived off to tickets #110 and #111 respectively - this ticket should now only be for noisy bands

[mggr]

Better picture of noisy hawk bands - with azspec that's corrected for the spikes.

Band 87 (1504.61nm) vs band 71 (1403.66nm)

[benj]

We pretty much concluded that this was just to do with water absorption - any band in a part of the spectrum where water absorbs a lot is noisy. Closing this ticket.