Fenix Calibration, end of season, october 2018

[asm]

This ticket records calibration of the Fenix sensor for the final season. As a memento, the Fenix sensor was repaired by Specim in February 2018 and the detector array was replaced with a new one so timeline comparisons should not be made with calibrations made before 2018

Data location:
~arsf/arsf_data/2018/misc/calibration_october/original_data

[asm]

A season final calibration was performed at BAS Cambridge during 17-19 October 2018. Most likely the last NERC-ARF calibration. asm attended to along with Chris from NERC-FSF and Gary from Ops.

• Collected 3 datasets for Fenix
• Problems with some Fenix data. The frame counter will go to negative values sometimes! Usually, first flightline data after warming up. The data can be used but has less number of frames.
• asm wrote a script to compare uniforme versus absolute fenix data (close or far from integrating sphere)
• asm to test latest changes to librarfscal and get a better tracking with the most up to date files
• Discussed with Chris the need of uncertainties in report. Chris also trying to get a better documentation of processing procedures.
• Better documentation for the scripts is also needed so will do that as we process the data.
• Further characterisation using the double monochromator to be performed during 5 days starting on 29th October. asm to go back to BAS those days.

[asm]

Wavelength accuracy.

Started at Cambridge, some problems were found and resumed at PML. No Mylar filter used for any of the datasets
-Dataset 1: No peaks found in the SWIR using the Kr lamp. Only one peak found for the 2 available He emissions in the SWIR. No fitting can be done with only one peak.

-Dataset 2: None of the Kr peaks were found in the SWIR, only the 2 peaks for He could be used.

-Dataset 3: The 2 peaks for He and 4 peaks for Kr where found in the SWIR region. This automatically makes this dataset the more trustworthy. Some of the other peaks in the csv file are very close to others not noted in the same file. Their contribution are likely to be mixed as fityk is reporting peaks in between of those values when analyising that lamp data alone. Related to that, there are a few peaks in the VNIR showing large values of FWHM that could be caused for something similar (such as 837.8 that was identified as 837.64 returning a FWHM of 25.15 when it seems that the Ne has one peak at 836.6 and another at 837.8). Since I am going back to the lab at Cambridge next week, I will put further processing on hold and will discuss this issue with Chris. So far the offset for this dataset is: 0.330852 and 0.386719 with respect to the original values of the Specim's calibration (good values)

[asm]

Wavelength accuracy.

Resumed. Got rid of peaks that were providing dubious values: 841.8 nm for Ar and Hg-AR and 607.9 nm for Ne.
Final offset with the Specim wavelenghts is good: 0.273926 and 0.386719

[asm]

Radiometric Calbration

Updated the latest "ARF Radiance Sphere Cal Data" to mW rather than W and commit that one to the git repository. Also extrapolated the data (ended in 2500nm and the Fenix new wavelenghts over that number). Calibrated the files withouth using the blue filter files (no calibrated updated file). Used data from the 3rd dataset, same as for wavelength calibration.

Radiometric differences look good compared to the latest calibration, everything makes sense.

[asm]

Bad Pixel mapping

Used same settings than the previous calibration as commented in ticket 627. Used the same Method E as that one but double checked again band by band using fastQC. Everything looking good.

Processed to level 1 a couple 179 and 174b 2019 fenix lines to make sure everything is OK. Everything looks good. Also processed Also compared with py6s against vegetation pixels; spectra looks sensible.

Created the calibration files for the different binnings. Placed everything under ~arsf/calibration/2019/fenix
Calibration is completed with the exception of creating a new data quality report.

[asm]

Final Remarks

A new script "plot_fenix_sensitivity.py" have been added for the hyperspectral data quality report, section "method F". The calibration processing have been carried out using the version libarsfcal-1.0.5 and the quality report is currently under review. I will close this ticket now.