Isis3: clemhirescal: Add support for all MCP gain settings

[dpmayerUSGS]

Description
clemhirescal is meant to radiometrically calibrate Clementine HIRES camera images to I/F by applying flat field correction and then scaling pixel values by a calibration constant, _k_, that is chosen from a lookup table based on the image filter and gain mode. The values of _k_ themselves have been empirically derived in order to scale I/F values of the HIRES camera to be consistent with the Clementine UVVIS camera.

The k values currently used by clemhirescal are taken from Tables 5 and 6 in Robinson et al. (2003), but the _k_ values in those tables only work for a tiny minority of the possible MCP gain states, and even then only apply to the A and D filters of the camera. The good news is that on page 11 of the paper, the authors state that _k_ values for arbitrary MCP gain states of the A and D filters can be derived by fitting a line to the values in Tables 5 and 6, respectively.

Just doing a quick line fit in LibreOffice Calc, the calibration function for the A filter would be
k_A = (-5.33333333333333*10^-5*MCP_gain) + 0.00937

and for the D filter,
k_D = (-9.75301204819275*10^-5*MCP_gain) + 0.0163866265

(The paper also describes how a LUT of _k_ values can be established for the B and C filters, but this would require comparing a sample of HIRES images to co-registered UVVIS images and applying equations 5 and 6 from the paper [I don't know why the authors didn't just do this in the paper]. This would require considerable effort and it would only yield a first order approximation, so maybe not worth implementing at this time.)

tl;dr
clemhirescal currently uses a LUT of k values that only supports a minority of Clementine HIRES products, but the relevant source paper makes it sound like it's easy to derive k values to support all MCP gain states, at least for the A and D filters. Adding this feature would allow users to calibrate nearly all of the Clementine HIRES images from the A and D filters without necessarily having to derive calibration factors themselves.

Acceptance Criteria

• [ ] clemhirescal applies calibration function (as described on Page 11 of Robinson et al. (2003)) to Clementine HIRES images taken through the A and D filters for arbitrary MCP gain states
• [ ] The updated version of clemhirescal must produce rasters with pixel values +/- 1% of results obtained with ISIS 3.8.0 for images collected in the subset of MCP gain states that are currently supported (no code regression)
• [ ] Update comments in the clemhirescal source code to correctly describe how the k values are determined in the updated code

Example Data
Processing consists of running clem2isis on Clementine HIRES EDRs (such as mentioned below), followed by spiceinit and then clemhirescal.
A Filter
An example image that is currently supported is /pds_san/PDS_Archive/Clementine/cl_0051/lun224/lhxxxxxx/lhxxxxxe/lha0547e.224. When this file is passed as input, an updated version of clemhirescal should yield a raster whose pixel values are each within 1% of a raster obtained by running the current version of clemhirescal.

An example image that is not currently supported is /pds_san/PDS_Archive/Clementine/cl_0034/lun155/lhxxxxxx/lhxxxxxm/lha4988m.155. This image should yield a calibrated output raster after clemhirescal is modified to use the calibration function described in Robinson et al. (2003)

D Filter
An example image that is currently supported is /pds_san/PDS_Archive/Clementine/cl_0054/lun237/lhxxxxxx/lhxxxxxn/lhd3924n.237. When this file is passed as input, an updated version of clemhirescal should yield a raster whose pixel values are each within 1% of a raster obtained by running the current version of clemhirescal.

An example image that is not currently supported is /pds_san/PDS_Archive/Clementine/cl_0018/lun088/lhxxxxxx/lhxxxxxh/lhd3062h.088. This image should yield a calibrated output raster after clemhirescal is modified to use the calibration function described in Robinson et al. (2003)

Product Owner
@dpmayerUSGS

[dpmayerUSGS]

@jessemapel Could you add the "Products" label to this issue?

I don't have permissions to add labels

[dpmayerUSGS]

@AustinSanders I updated the original issue to include example data from the A filter

[AustinSanders]

The latest test results indicate a difference of ~.58% for "D" filter (original/new implementation). Original/new implementations yield an exact match for "A" filter.

[dpmayerUSGS]

@AustinSanders The following comment block should be updated to reflect the changes in the code:
https://github.com/AustinSanders/ISIS3/commit/3e79dc7c43dc8ba74eea1d7adc6dd9abe3ba1b03#diff-4b1c297d512780bc6b6834b3bbc890adL51-L53

The updated comment should read

Compute the K value to convert to I/F. The functions for K value per MCP gain state and filter are based on Robinson, M. S., Malaret, E., and White, T. ( 2003), A radiometric calibration for the Clementine HIRES camera, J. Geophys. Res., 108, 5028, doi:10.1029/2000JE001241, E4.
The functions were determined by fitting a line to the data in Table 5 (A filter) or Table 6 (D filter) of Robinson et al., as described on page 11.

The following 2 comments can be deleted:
https://github.com/AustinSanders/ISIS3/commit/3e79dc7c43dc8ba74eea1d7adc6dd9abe3ba1b03#diff-4b1c297d512780bc6b6834b3bbc890adL58
https://github.com/AustinSanders/ISIS3/commit/3e79dc7c43dc8ba74eea1d7adc6dd9abe3ba1b03#diff-4b1c297d512780bc6b6834b3bbc890adL71

[jessemapel]

@dpmayerUSGS @AustinSanders Can specific code feedback be left on a PR? Issues are not a very good venue for this type of discussion.

[dpmayerUSGS]

:+1:

On Nov 6, 2019, at 15:05, Jesse Mapel [e-mail] wrote:


@dpmayerUSGS @AustinSanders Can specific code feedback be left on a PR? Issues are not a very good venue for this time of discussion.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.