Closed oliversus closed 2 years ago
Talked with Carlos Horn, who is trying to find the source of these coefficients. He believes that the MetopC values stem from an intermin, unpublished PATMOS-X calibration dataset. We also tracked down the source calibration files he was once sent (without remembering who that was). For MetopC, that file says:
avhrr_3_instr.dat,v 1.1 2019/01/24 19:35:15 heidinger Exp $! heidinger made using EUMETSAT numbers
The coeficients in that file are identical to the ones in this PR. Except for S0, which we convert from the low and high gain estimates to one single value.
Hi @oliversus @sfinkens (and @carloshorn)
I checked my old emails and I can see that we received two sets of MetOp-C coefficients from Mike Foster (NOAA). The first one was received on 22 Sept 2020 and the second, improved one on 2 Feb 2021. The visible channel coefficients S0, S1 and S2 are different in these two files, while the everything rest seems to be the same. I am pasting them below so that Olli can check which CLARA-A3 version/processing corresponds to which set of coefficients.
22 Sept 2020
m03 !sat_name
4.366 !solar_3b
279.62 !ew_3b
0.00 0.00, 0.00, 0.00 !ch3b nonlinear rad coef (Ns,bo,b1,b2,b0,Ns)
-6.27, 6.58, -0.132030, 0.00065922 !ch4 nonlinear rad coef (Ns,b0,b1,b2)
-2.55, 3.23, -0.056920, 0.00024963 !ch5 nonlinear rad coef (Ns,b0,b1,b2)
2707.6457 !nu_3b
-1.7866821 !a1_3b
1.0023679 !a2_3b
931.89092 !nu_4
-0.56558177 !a1_4
1.0015104 !a2_4
832.69445 !nu_5
-0.39207223 !a1_4
1.0011504 !a2_5
40.41 !ch1_dark_count
40.94 !ch2_dark_count
40.57 !ch3a_dark_count
0.0543 0.000 -0.000 ! ch1 low gain S0,S1,S2
0.1592 0.000 -0.000 ! ch1 high gain S0,S1,S2
0.0541 0.000 -0.000 ! ch2 low gain S0,S1,S2
0.1584 0.000 -0.000 ! ch2 high gain S0,S1,S2
0.0258 0.000 -0.000 ! ch3a low gain S0,S1,S2
0.1812 0.000 -0.000 ! ch3a high gain S0,S1,S2
2018.848734 !date of launch
498.68 !channel 1 switch count
500.01 !channel 2 switch count
498.72 !channel 3A switch count
276.5862, 0.051051, 1.474208E-06, 0.00E-08, 0.00E-11 !PRT1
276.6136, 0.051029, 1.472138E-06, 0.00E-08, 0.00E-11 !PRT2
276.5975, 0.051065, 1.469268E-06, 0.00E-08, 0.00E-11 !PRT3
276.4595, 0.050990, 1.506223E-06, 0.00E-08, 0.00E-11 !PRT4
0.25 0.25 0.25 0.25 !PRT weighting
1.0159, 2.2285, 1.2289, -277.2571 !MCSST from OISST (day+night) (TBD)
1.0159, 2.2285, 1.2289, -277.2571 !MCSST from OISST (day+night) (TBD)
!$Id: avhrr_3_instr.dat,v 1.1 2019/01/24 19:35:15 heidinger Exp $!
!-heidinger made using EUMETSAT numbers (
2 Feb 2021
m03 !sat_name
4.366 !solar_3b
279.62 !ew_3b
0.00 0.00, 0.00, 0.00 !ch3b nonlinear rad coef (Ns,bo,b1,b2,b0,Ns)
-6.27, 6.58, -0.132030, 0.00065922 !ch4 nonlinear rad coef (Ns,b0,b1,b2)
-2.55, 3.23, -0.056920, 0.00024963 !ch5 nonlinear rad coef (Ns,b0,b1,b2)
2707.6457 !nu_3b
-1.7866821 !a1_3b
1.0023679 !a2_3b
931.89092 !nu_4
-0.56558177 !a1_4
1.0015104 !a2_4
832.69445 !nu_5
-0.39207223 !a1_4
1.0011504 !a2_5
40.41 !ch1_dark_count
40.94 !ch2_dark_count
40.57 !ch3a_dark_count
0.054 7.424 -2.579 ! ch1 low gain S0,S1,S2
0.163 7.424 -2.579 ! ch1 high gain S0,S1,S2
0.065 -0.900 1.278 ! ch2 low gain S0,S1,S2
0.195 -0.900 1.278 ! ch2 high gain S0,S1,S2
0.041 -45.535 20.122 ! ch3a low gain S0,S1,S2
0.284 -45.535 20.122 ! ch3a high gain S0,S1,S2
2018.848734 !date of launch
498.68 !channel 1 switch count
500.01 !channel 2 switch count
498.72 !channel 3A switch count
276.5862, 0.051051, 1.474208E-06, 0.00E-08, 0.00E-11 !PRT1
276.6136, 0.051029, 1.472138E-06, 0.00E-08, 0.00E-11 !PRT2
276.5975, 0.051065, 1.469268E-06, 0.00E-08, 0.00E-11 !PRT3
276.4595, 0.050990, 1.506223E-06, 0.00E-08, 0.00E-11 !PRT4
0.25 0.25 0.25 0.25 !PRT weighting
1.0159, 2.2285, 1.2289, -277.2571 !MCSST from OISST (day+night) (TBD)
1.0159, 2.2285, 1.2289, -277.2571 !MCSST from OISST (day+night) (TBD)
!$Id: avhrr_3_instr.dat,v 1.1 2019/01/24 19:35:15 heidinger Exp $!
!-heidinger made using EUMETSAT numbers (
Yes @abhaydd , those coefficients from the 2nd Feb 2021 are the new ones in this pull request. So now we have a source and a date, nice. Those coefficients have not been published by the PATMOS-X group yet, right?
I don't think they have officially published them yet. I don't see them at least on their website. https://cimss.ssec.wisc.edu/patmosx/calibration/
Carlos and me also had a look there before. I think this discussion is sufficient for the moment to serve traceability. I'd suggest to close this PR now and do the merge.
Nice research, thanks a lot! Can you please take a look at the failing tests?
Merging #109 (fdb6f17) into main (05ba35e) will not change coverage. The diff coverage is
n/a
.
@@ Coverage Diff @@
## main #109 +/- ##
=======================================
Coverage 72.93% 72.93%
=======================================
Files 32 32
Lines 2830 2830
=======================================
Hits 2064 2064
Misses 766 766
Impacted Files | Coverage Δ | |
---|---|---|
pygac/calibration.py | 94.21% <ø> (ø) |
Help us with your feedback. Take ten seconds to tell us how you rate us.
I updated the hashdigest for this new calibration file, replacing the old version saying "provisional MetopC coefficients".
Ready to merge, @sfinkens ?
@oliversus thanks for the update!
Nice work, thanks a lot @oliversus ! And also thanks @mraspaud for merging :)
Carlos Horn once provided me with these METOPC calibration coefficients. Unfortunately, I do not know the source of these coefficients. I will try to contact him, unless someone else knows where they might be coming from. I used these coefficients for release 1 of the PyGAC FDR.