Skip to content

Commit 380766b

Browse files
committed
oco round 2: correct forward model, TSIS-1 continuum, ILS kernels, pressure slider, cloud chip; publish oco/
- Spectra explorer: tau and E0 shipped UNCONVOLVED at 0.001/0.004/0.005 nm with one OCO-2 ILS kernel per band; the client forms exp(-m*tau) with psurf scaling, multiplies by E0, convolves with the ILS, then draws. Solar continuum from the TSIS-1 HSRS with Toon's lines. Line-depth check: model - measured = +0.016 / +0.000 / +0.013, continuum ratios 0.98 / 0.96 / 0.99 (was 0.44 vs 0.32 in the A-band). - Photon path: opaque-cloud chip. Footprint widget: pitch vs sampled width in caption and readout. - sounding_oco2.json: aod_l2 / quality_flag aliases. Surface label 'water (no glint)'. - _quarto.yml: /oco added to resources so the tree is published like /sif. - Hero: observation composite replaced by the JJA model field (script added; image follows).
1 parent 0a1ecd6 commit 380766b

24 files changed

Lines changed: 284 additions & 175 deletions

‎_quarto.yml‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -5,6 +5,7 @@ project:
55
resources:
66
- /publications
77
- /sif
8+
- /oco
89

910
website:
1011
title: "FraLab"

‎oco/data/README.md‎

Lines changed: 51 additions & 38 deletions
Original file line numberDiff line numberDiff line change
@@ -5,40 +5,50 @@ All files are plain JSON, numbers rounded to 6 significant digits, each with a `
55
`oco/scripts/` (Python: conda env `ee`, Python 3.12; Julia: the Google-RT project environment,
66
AtmosphericAbsorption.jl with ABSCO v5.2). Re-run order and exact commands are at the bottom.
77

8-
## 1. Spectra explorer
8+
## 1. Spectra explorer (round 2, 2026-09-16: unconvolved high-resolution inputs)
99

1010
| file | contents | check |
1111
|---|---|---|
12-
| `tau_aband.json` (134 kB) | 758–772 nm, 4000 pts; `tau_o2` (`tau_co2` = 0) `tau_h2o` | tau_o2 n=4000 min=1.643e-4 max=52.92 nan=0; tau_h2o max=9.76e-4; ILS FWHM 0.0418 nm |
13-
| `tau_wco2.json` (137 kB) | 1590–1621 nm; `tau_co2` at 420 ppm, `tau_h2o` (`tau_o2` = 0) | tau_co2 min=5.705e-5 max=0.4712; tau_h2o min=1.264e-4 max=0.03999; nan=0; ILS FWHM 0.0775 nm |
14-
| `tau_sco2.json` (173 kB) | 2042–2082 nm; `tau_co2`, `tau_h2o`, `tau_ch4` | tau_co2 min=0.01017 max=8.833; tau_h2o max=0.8333; tau_ch4 max=3.0e-5; nan=0; ILS FWHM 0.1011 nm |
15-
| `solar_{aband,wco2,sco2}.json` (64 kB each) | `e0` W m⁻² µm⁻¹ at 1 AU on the same `wl` | aband 738.4–1252; wco2 133.0–208.1; sco2 73.47–95.3; nan=0 |
12+
| `tau_aband.json` (360 kB) | 758–772 nm, 0.001 nm, 14 001 pts, **unconvolved**; `tau_o2`, `tau_h2o` (`tau_co2` = 0) | tau_o2 min=1.466e-4 max=528.9; tau_h2o max=3.667e-3; nan=0 |
13+
| `tau_wco2.json` (204 kB) | 1590–1621 nm, 0.004 nm, 7 751 pts; `tau_co2` at 420 ppm, `tau_h2o` | tau_co2 min=4.917e-5 max=2.131; tau_h2o max=0.1373; nan=0 |
14+
| `tau_sco2.json` (189 kB) | 2042–2082 nm, 0.005 nm, 8 001 pts; `tau_co2`, `tau_h2o` (CH₄ dropped, τ < 3e-5) | tau_co2 min=7.264e-3 max=38.14; tau_h2o max=2.791; nan=0 |
15+
| `solar_{aband,wco2,sco2}.json` (112 / 62 / 65 kB) | `e0` W m⁻² µm⁻¹ at 1 AU, unconvolved, same grids | 141.9–1265; 124.3–254.5; 71.47–107.9; nan=0; TSIS continuum at band centre 1245.5 / 250.7 / 103.7 |
16+
| `ils_{aband,wco2,sco2}.json` (13 / 8 / 11 kB) | OCO-2 L1b ILS kernel, footprint 4, band-centre sample, on the tau spacing, unit sum | FWHM 0.0415 / 0.0774 / 0.1012 nm; 583 / 351 / 547 pts (±0.28 / ±0.70 / ±1.37 nm); wings < 2e-4 of peak dropped (0.06 / 0.13 / 0.00 % of the sum) |
1617
| `sounding_oco2.json` (65 kB) | one OCO-2 L1b sounding, 3 bands × 1016 samples, L2 values | aband 757.66–772.57 nm, max 62.31 W m⁻² sr⁻¹ µm⁻¹, good 858/1016; wco2 1590.58–1621.74, max 14.32, good 878; sco2 2043.06–2083.20, max 5.518, good 855 |
1718

18-
Optical depths: vertical (airmass 1), US Standard Atmosphere 1976 (T, p) on 60 layers from
19-
1013.25 to 0.05 hPa, H₂O = 50 % relative humidity in the troposphere (column 15.7 kg m⁻²),
20-
XCO₂ 420 ppm (dry), O₂ 0.20946, CH₄ 1.9 ppm. CO₂, O₂, H₂O cross sections from **ABSCO v5.2**
21-
(`~/data/ABSCO/v5.2_final/*_v52.hdf`, pressure/temperature/H₂O-broadener interpolation via
22-
`AtmosphericAbsorption.compute_cross_section`); CH₄ line-by-line from the HITRAN2020 `.par`
23-
artifact (Voigt, 40 cm⁻¹ cutoff) because no ABSCO CH₄ table is on disk. Computed at 0.01 cm⁻¹,
24-
**convolved at that resolution** with the OCO-2 L1b instrument line shape (per-spectral-sample
25-
`ils_delta_lambda` / `ils_relative_response`, footprint 4, of granule
26-
`oco2_L1bScND_53642a_240801_B11205r_240906004756.h5`), then linearly interpolated to 4000
27-
points per band. Note the client's `exp(-m·tau_conv)` is an approximation of the ILS-convolved
28-
transmittance (requested form); CH₄ in this band is negligible (τ ≤ 3e-5) and can be dropped.
29-
30-
Solar irradiance: Geoff Toon's disk-integrated solar transmittance
31-
(`solar_merged_20240731_600_33300_100`, 0.01 cm⁻¹) × Planck continuum at 5772 K scaled by
32-
(R☉/AU)². The Planck continuum is within ~1 % of measured irradiance at 0.76 µm but ~10 % low at
33-
1.6 µm; stated in `meta`.
34-
35-
Sounding: OCO-2 `2024080120532372`, 2024-08-01 20:53:34 UTC, nadir, footprint 2, Railroad
36-
Valley playa (38.382 N, 115.857 W), SZA 24.9°, VZA 0.52°, L2 Lite v11.2 XCO₂ 422.89 ppm,
37-
albedo 0.353 / 0.409 / 0.394, surface pressure 857 hPa, AOD 0.071, quality flag 0. Radiance
38-
converted from photon units with hc/λ; wavelengths from the footprint's L1b dispersion polynomial
39-
(1-based sample index); `good` = `bad_sample_list == 0`. **OCO-2 measures one linear
40-
polarization**, so the L1b radiance is ≈ ½ of the total intensity of an unpolarized scene
41-
(`meta.polarization_factor = 0.5`; the explorer scales the forward model accordingly).
19+
Grids are uniform and stored compactly: `wl[i] = meta.wl_start_nm + i·meta.wl_step_nm`, `i = 0 … meta.n_points−1`
20+
(no `wl` array; this keeps the A-band file under 400 kB). Optical depths are vertical (airmass 1) at 1013.25 hPa
21+
for the US Standard Atmosphere 1976 (60 layers, H₂O 50 % RH in the troposphere, 15.7 kg m⁻²), XCO₂ 420 ppm dry,
22+
O₂ 0.20946, from **ABSCO v5.2** (`~/data/ABSCO/v5.2_final`) at 0.01 cm⁻¹, linearly interpolated to the grids (the
23+
0.01 cm⁻¹ computation is finer than every grid: 0.0006 / 0.0025 / 0.0042 nm). **The client does the physics in
24+
the right order**: `T = exp(−m·[psurf/1013.25·(XCO₂/420·tau_co2 + tau_o2) + tau_h2o])`, `L = E₀·cos(SZA)·a/π·T`
25+
at full resolution, then convolution with the ILS kernel, then sampling every 4th point for drawing; a
26+
surface-pressure slider (700–1030 hPa) is set to the L2 value when the sounding overlay is on.
27+
28+
Solar irradiance: **TSIS-1 Hybrid Solar Reference Spectrum** (Coddington et al. 2021; v2 0.005 nm file,
29+
`hybrid_reference_spectrum_p005nm_resolution_c2022-11-30_with_unc.nc`) provides the continuum — its lines are
30+
divided out with Toon's transmittance at 0.2 nm and the ratio smoothed to 1 nm — and Geoff Toon's disk-integrated
31+
transmittance (`solar_merged_20240731_600_33300_100`, 0.01 cm⁻¹) provides the Fraunhofer structure.
32+
33+
**Line-depth check** (`oco/scripts/check_line_depths.py`: the shipped sounding forward-modelled exactly as the
34+
widget does, at the L2 XCO₂ 422.89 ppm, albedos 0.353 / 0.409 / 0.394, psurf 857.9 hPa, SZA 24.9°, polarization
35+
factor 0.5; depth = 1 − L/continuum envelope, averaged over good samples):
36+
37+
| band | mean line depth measured | mean line depth model | model − measured | continuum meas/model |
38+
|---|---|---|---|---|
39+
| A-band | 0.291 | 0.307 | +0.016 | 0.980 |
40+
| weak CO₂ | 0.078 | 0.078 | +0.000 | 0.959 |
41+
| strong CO₂ | 0.301 | 0.314 | +0.013 | 0.985 |
42+
43+
(Round 1, convolved-τ form with Planck continuum: 0.437 / 0.070 / 0.419 vs measured 0.317 / 0.060 / 0.321 and
44+
continuum ratios 0.97 / 1.18 / 1.07.) All bands are now within the 0.02 target and no fudge factor is applied.
45+
46+
Sounding: OCO-2 `2024080120532372`, 2024-08-01 20:53:34 UTC, nadir, footprint 2, Railroad Valley playa
47+
(38.382 N, 115.857 W), SZA 24.9°, VZA 0.52°, L2 Lite v11.2 XCO₂ 422.89 ppm, albedo 0.353 / 0.409 / 0.394,
48+
surface pressure 857.9 hPa, AOD 0.071 (`meta.aod_l2`), quality flag 0 (`meta.quality_flag`). Radiance converted
49+
from photon units with hc/λ; wavelengths from the footprint's L1b dispersion polynomial (1-based sample index);
50+
`good` = `bad_sample_list == 0`. **OCO-2 measures one linear polarization**, so the L1b radiance is ≈ ½ of the total
51+
intensity of an unpolarized scene (`meta.polarization_factor = 0.5`).
4252

4353
## 2. Footprint scale explorer
4454

@@ -79,12 +89,13 @@ snow 0.955/0.156/0.052, conifer 0.652/0.261/0.112, dry grass 0.517/0.597/0.452.
7989

8090
## 4b. Hero image
8191

82-
`img/xco2_hero.jpg` (2400 × 1080, 148 kB): per-cell **median** XCO₂ of all good-quality OCO-2
83-
Lite v11.2 soundings of 2023 on a 3° grid (29.0 M soundings, 5855 cells), nearest-neighbour
84-
gap-filled, σ = 0.7 cell smoothing, drawn over land only (ocean left dark), Antarctica removed entirely for now,
85-
bold light coastlines, Robinson projection, colour range 413.4–420.8 ppm. There is no OCO L3
86-
product on disk, so this is our own composite; annual medians at high latitudes are biased low by
87-
summer-only sampling (the blue Siberia).
92+
`img/xco2_hero.jpg` (2400 × 1080, dark background, < 400 kB): **JJA 2019 mean XCO₂ from the AtmosTransport CATRINE
93+
C90 run** (`co2_natural`, dry-air-mass-weighted column, 8 snapshots/day × 92 days, nearest-cell regridded to 0.25°,
94+
σ = 1 cell smoothing, dark-blue → gold ramp, subtle coastlines, no axes/labels). `img/xco2_hero_jja2019.gif` is the
95+
same field day by day at 12 UTC (92 frames). An observation-based alternative is produced by
96+
`make_xco2_composite.py` + `make_xco2_hero.py` (OCO-2 Lite v11.2, JJA 2021–2024, 1° medians incl. ocean glint,
97+
σ = 2 cells, Antarctica removed) — smoother than round 1 but still streaky from the glint sampling, which is why the
98+
model field is used.
8899

89100
## Reproduce
90101

@@ -94,14 +105,16 @@ PY=~/.conda/envs/ee/bin/python; S=~/code/gitHub/RSFraLab.github.io; W=/path/to/w
94105
julia -t 8 $S/oco/scripts/tau_lbl.jl $W # tau_{aband,wco2,sco2}_lbl.txt
95106
$PY $S/oco/scripts/make_solar_lbl.py /kiwi-data/Data/groupMembers/cfranken/solar/toon/solar_merged_20240731_600_33300_100.out.gz $W
96107
L1B=~/data/OCO2_L1b_full/oco2_L1bScND_53642a_240801_B11205r_240906004756.h5 # GES DISC OCO2_L1B_Science.11.2r/2024/214
97-
$PY $S/oco/scripts/make_tau_solar_json.py $W $S/oco/data $L1B 3
108+
$PY $S/oco/scripts/make_tau_solar_hr_json.py $W $S/oco/data $L1B 3 ~/data/solar/hybrid_reference_spectrum_p005nm_resolution_c2022-11-30_with_unc.nc # TSIS-1 HSRS v2 from LASP LISIRD
109+
$PY $S/oco/scripts/check_line_depths.py $S/oco/data
98110
$PY $S/oco/scripts/make_sounding_json.py $L1B /kiwi-data/Data/satellite/OCO2/L2_Lite_FP/11.2r/raw/2024/oco2_LtCO2_240801_B11211Ar_240924224233s.nc4 2024080120532372 $S/oco/data/sounding_oco2.json "Railroad Valley playa, Nevada"
99111
# 2. footprints and track (local O2-band L1bSc granule), OCO-3 SAM (Lite), basemaps
100112
$PY $S/oco/scripts/make_footprints_json.py /kiwi-data/Data/satellite/OCO2/L1bSc/11.2r/raw/2024/196/oco2_L1bScND_53380a_240714_B11205r_240814182247.h5 $S/oco/data 34.05 -118.25
101113
$PY $S/oco/scripts/make_oco3_sam_json.py /kiwi-data/Data/satellite/OCO3/L2_Lite_FP/11r/raw/2022/oco3_LtCO2_220218_B11072Ar_240915212336s.nc4 fossil0005 $S/oco/data/footprints_oco3_sam_la.json
102114
$PY $S/oco/scripts/make_basemaps.py $S/oco/img $S/oco/data/basemaps.json
103115
# 3. albedo
104116
$PY $S/oco/scripts/make_albedo_surfaces.py <dir with the ECOSTRESS .spectrum.txt files> $S/oco/data/albedo_surfaces.json
105-
# 4b. hero
106-
$PY $S/oco/scripts/make_xco2_composite.py 2023 $W/xco2_2023_3deg.npz 3 && $PY $S/oco/scripts/make_xco2_hero.py $W/xco2_2023_3deg.npz $S/oco/img/xco2_hero.jpg
117+
# 4b. hero (model field) and the observation-based alternative
118+
$PY $S/oco/scripts/make_model_xco2_hero.py ~/data/AtmosTransport/output/catrine_c90_2019_rt_co2 $S/oco/img/xco2_hero.jpg $S/oco/img/xco2_hero_jja2019.gif
119+
$PY $S/oco/scripts/make_xco2_composite.py $W/xco2_jja_2021_2024_1deg.npz 1 2021-2024 6,7,8 && $PY $S/oco/scripts/make_xco2_hero.py $W/xco2_jja_2021_2024_1deg.npz $W/xco2_hero_obs.jpg
107120
```

0 commit comments

Comments
 (0)