Printable Version of Topic

Click here to view this topic in its original format

Unmanned Spaceflight.com _ MRO 2005 _ HiRISE Expert Question: Radiometric Calibration

Posted by: Nirgal Mar 7 2010, 12:47 PM

Hi,

I have a question regarding the radiometric calibration of HiRISE (greyscale) images that I have not been able to find an conclusive answer to from looking at the specs & docs so far.

EDIT: ok, I just found the following excellent document:
http://hirise.lpl.arizona.edu/pdf/HiRISE_RDR_v12_DTM_11_25_2009.pdf
that I'm going to digest now ... should answer most questions

However, another thing that is not quite clear to me yet is the handling of look-up table (LUT) during calibration. The LUT is used to convert the original 10 bit to 8 bit on-board needed by the lossless compression algorithm before transmission to earth.
However, as the LUT is nonliner (something square-rootish) the data has to be expanded again somewhare during calibration chain to the original range of 0-1024. However as many of the published images contain D/N values only up to 255 I'm courious if the inverse-lookup correction is always applied or if the LUT-compressed is possibly retained for some of the browse and/or QLOOK images ??

So would it be necessary for those images to apply the inverse-LUT to obtain the real radiometrically calibrated values ?

In an older thread about a similar subject, tuvas wrote:

QUOTE

I was under the impression that the RDR's were going to be released deLUTing the images, restoring them to their 10 bit glory, but it appears that that is not the case... Odd... For anyone who is hard core, I believe the LUT information is included in the EDRs, as might have been mentioned here... I should be paying more attention, but due to my long absence, well, I'm having to only skim topics of interest...

Maybe someone of the experts could shed some light onto this (also with respect to the new, since then improved calibration chain)

Posted by: GuyMac Mar 7 2010, 06:10 PM

QUOTE (Nirgal @ Mar 7 2010, 05:47 AM)

However, another thing that is not quite clear to me yet is the handling of look-up table (LUT) during calibration. The LUT is used to convert the original 10 bit to 8 bit on-board needed by the lossless compression algorithm before transmission to earth.
However, as the LUT is nonliner (something square-rootish) the data has to be expanded again somewhare during calibration chain to the original range of 0-1024. However as many of the published images contain D/N values only up to 255 I'm courious if the inverse-lookup correction is always applied or if the LUT-compressed is possibly retained for some of the browse and/or QLOOK images ??

The EDRs contain the uncalibrated, LUTted values. hi2isis is used in the calibration steps after that. The lossless RDR's (no QLOOK) are the only products that are both calibrated and can be used to get absolute I/F values (our accuracy is found to be within 20% of the true values). All the extras have things done to them like color stretches and conversion to 8-bit (per band) values and lossy JP2 compression. Hope that helps!

Posted by: Bjorn Jonsson Mar 7 2010, 11:58 PM

There is some interesting general discussion on albedo variations in the paper http://astrogeology.usgs.gov/Projects/ISPRS/MEETINGS/Houston2003/abstracts/Kirk_isprs_mar03.pdf but I assume you must have probably already seen that document.