ISIS, LISA and Flats a question?

[Andrew Smith]

I have been wondering about the use of flats with the latest ISIS software and LISA reduction process. The software averages the flat by column as the LISA’s built in calibration unit’s flat have a gradient even when adjusted optimally. Given this and the reduction process performed by ISIS which derives an instrumental response and then applies this to the target spectra I can’t see what the flat contributes.
If the flat was not averaged by column (and did not have a gradient) then it could compensate for pixel by pixel variation in response as the target or reference star was imaged at different places on the slit. However, this is negated by the averaging by column. I suppose it could compensate for high frequency variations in pixel response by row that would be filtered out when the instrumental response is smoothed but the binning by column of the spectra would reduce this in any case.

I would be happy to have others thoughts about this please.

Andrew

[Robin Leadbeater]

I have been wondering about the use of flats with the latest ISIS software and LISA reduction process. The software averages the flat by column as the LISA’s built in calibration unit’s flat have a gradient even when adjusted optimally.

Does it? That seems very odd to me.

Spectroscopic flats are normally applied in exactly the same way as an imaging flat and correct for:

Spacial variations eg pixel to pixel variation, dirt, vignetting etc
Wavelength dependent variations eg instrument response (camera, grating etc), fringes etc

Which leaves the flat lamp spectrum to be removed based on a calibration star spectrum.

Are you sure it is not just say normalising the values in each row to remove any vertical gradient, though I cannot see why this would be necessary (This would be ok provided the star spectrum is exactly horizontal. It could give errors though if there is any tilt in the spectrum)

Cheers
Robin

[Christian Buil]

The rationale for the equalization option along column for flat-field is associated with the tungsten lamp: some time
slit illumination along its long axis is not perfectly uniform during flat-field acquisition.

This procedure simulate after processing a uniform illumination
of the slit with tungsten lamp, i.e. an illumination similar to that given by the telescope along this axis.
Global flat field (2D) correction is better in this particular situation.

Christian

[Andrew Smith]

Thanks, Robin and Christian. Robin you may well be right and the column correction is probably a normalisation rather than an average which would not invalidate it's use as I thought.
Regards Andrew