Fenix Calibration, 2017

[lah]

This ticket records the 2017 calibration of the loan Fenix sensor (ID 350006R), so shouldn't be used for year on year comparisons. The process is also different as a blue filter was not used for the radiometric calibration and the scripts were altered to use the mylar data to increase the number of SWIR anchor points.

Data location: ~arsf/arsf_data/2017/misc/loan_Fenix_cal_may_cambridge

As this is a loan system, Specim have provided their own calibration. Files and reports are stored in:
~/arsf/calibration/loan_fenix_2017

[lah]

"Initial checks"

Initial checks were performed at Cambridge on the wavelength scale to find a small offset of 0.281 in the VNIR and 0.914 in the SWIR. These were based on a lamp file not using Oxygen or Mylar so need to be redone.

The scripts have been altered to include Mylar and increase the number of SWIR anchors. These changes are part of merge request ​https://gitlab.rsg.pml.ac.uk/arsf/libarsfcal/merge_requests/15. Tests of these scripts are stored in "processed_at_pml_mylar_test".

Some tips for proceeding with the calibration are:

• The files will probably need renaming (for the pm = d-2) as Gary added an underscore for no apparent reason.
• There were some files without dark frames in the morning, but these were repeated, so not a problem. They are in the reject_files directory. These shouldn't be a problem in the afternoon, but if something goes wrong check for this.
• Although the data were acquired in one day the 2 sets are named d-1 and d-2 for consistency. They may be renamed.
• Since adding O back in 2 lines have failed so will need checking. Make sure all the lamps are correct in the spectral lines file you use and rerun with different offsets. The scripts now automatically ignore ridiculous FWHM but there may still be a ~10 nm wide peak in the VNIR that shouldn't be there.
• Should check with Chris that the method implemented for Mylar is correct before using.
• For the data quality report you can compare results with Specim's instead of last year's. Their reports are in /users/rsg/arsf/calibration/loan_fenix_2017
• Oddly this sensor has an extra band so need to specify number_of_pixels=623 in config file
• Be prepared to redo when Chris sends us the new calibration files for the integration sphere

There were no data collected with the blue filter, so there are no F4106 files. The purpose of the blue filter is to enhance the SWIR spectrum, so this impacts the SWIR calibration multipliers. Could use the unfiltered filter as the filtered file, but the relative values between the VNIR and SWIR aren't correct - have emailed Chris for his opinion

[lah]

Incorrect ID

Note that the files have the incorrect sensor ID. They have our ID instead of 350006R. Specim replied:

"I have learned that the ssp files were not updated when the new
calibration files were loaded to your computer, and that's why the *.cal
and *.hdr files ID is according to your own sensor. The actual
calibration files should be for the sensor id 350006R, i.e. the sensor
that was delivered to you. So you should be good to go and operate the
system. However, we will still double check this, unfortunately it will
take couple of days as the personnel that calibrated the system are not
available today." Katja LEFEVRE

[lah]

'Sensor Problem

Further investigation into the SWIR data shows that the problem is not with the scripts or the lack of blue filter. There is a problem with the raw SWIR data as the signal level is not as high as it should be for the recorded integration time. Specim have been emailed.

Plots illustrating the problem have been saved to /users/rsg/arsf/arsf_data/2017/misc/loan_Fenix_cal_may_cambridge/sensor_problem.

With an unknown integration time a radiometric calibration for the SWIR data cannot be conducted. Furthermore, if the problem persists in flight data then the SWIR data may be impossible to calibrate.

Comparison of the VNIR region with Specim's data can still be conducted and Specim's files can be used to calibrate the flight data.

[lah]

Bad Pixels

Manually identified bad pixels (methodE) from lines 1, 5 and 20 of flight 167a. Updated files in /users/rsg/arsf/calibration/2017/fenix and tested as working fine.