Skip to content

validation.md

Latest commit

 

History

History
105 lines (70 loc) · 5.81 KB

validation.md

File metadata and controls

105 lines (70 loc) · 5.81 KB
Raw

Validation report (auto-generated by build/validation_report.py)

Reference: IPCC AR6 WG1 Ch.7 ERF series (Smith et al.) and AR6 Annex III concentrations. Forcing is relative to 1750, year 2019.

Quantity radforce (W m-2) AR6 (W m-2) difference
CO2 2.156 2.156 +0.0000
CH4 0.544 0.544 +0.0000
N2O 0.208 0.208 +0.0000
Halogenated (computed species only) 0.406 0.408 -0.0024
Total WMGHG 3.314 3.317 -0.0024

2xCO2 ERF: 3.934 W m-2 (AR6: 3.93 +/- 0.47).

Whole-series agreement 1850-2019 (max |error|, W m-2)

  • CO2: 0.0000
  • CH4: 0.0000
  • N2O: 0.0000

Non-GHG emulators vs the AR6 series 1750-2019 (max |error|, W m-2)

Agent max error AR6 2019
O3 0.0764 0.474
H2O_strat 0.0000 0.050
Aerosol_ari 0.0000 -0.216
Aerosol_aci 0.0077 -0.843
BC_snow 0.0126 0.080

Aerosol-cloud shape factors are a least-squares fit to the AR6 median series (see constants.py).

Skipped halogenated species (no RE or no 2019 concentration): c-C4F8, n-C4F10, n-C5F12, n-C6F14, i-C6F14, C7F16, C8F18, CFC-112, CFC-112a, CFC-113a, CFC-114a, HCFC-31, HCFC-124.

Spectral module (line-by-line + HITRAN cross-sections, MT_CKD continuum, US-standard atmosphere)

Comparison with observation (independent of the model tuning)

Quantity model observed source
clear-sky OLR, present day 264.0 W m-2 268.1 CERES EBAF Ed4.0 (Loeb et al. 2018, Table 5)
all-sky OLR, present day 236.8 W m-2 240.1 same
longwave cloud radiative effect 27.9 W m-2 27.9 same (this is the tuning target)

A single US-standard profile is not a global mean, so differences of a few W m-2 are expected. Without the water-vapour continuum and the 25 cm-1 line cut-off the model clear-sky OLR was 275 W m-2, 7 W m-2 too high.

Cloud model

All-sky = (1 - 0.67) x clear + 0.67 x [grey deck at 5-6 km, optical depth 0.685]. Cover is the satellite global-mean cloud fraction; the opacity is the single tuned parameter; the deck height is an assumption (sensitivity in docs/cloud_sensitivity.md).

Greenhouse-gas forcing (instantaneous, top of atmosphere, vs 1750 atmosphere)

  • 2xCO2: 2.76 W m-2 clear sky, 2.01 all-sky (AR6 effective forcing 3.93, which includes stratospheric adjustment).
  • Successive CO2 doublings add 2.76, 2.79 and 2.83 W m-2 (clear sky): logarithmic law from line physics.
  • 1750 to 2019, all-sky estimate: CO2 1.12, CH4 0.51, N2O 0.19, all three 1.80 W m-2.

Halocarbons (17 molecules, HITRAN cross-sections)

All-sky radiative efficiency / AR6 Table 7.SM.7: 1.02-1.31, mean 1.15 (clear sky: 1.62-1.95). Full table: docs/halocarbon_check.md. The AR6 values were not used in tuning. The remaining excess over AR6 is plausibly due to the single-profile, uniform-vertical-distribution and one-deck cloud simplifications; stratospheric adjustment would push our values the other way, so it does not explain it.

Emission metrics (GWP and GTP), against AR6 Table 7.SM.7

Method: AR6 WG1 Section 7.6.1 (impulse-response CO2 of Joos et al. 2013, two-timescale temperature response, climate-carbon feedback of Gasser et al. 2017, AR6 adjustments for CH4 and N2O). Code: physics/radforce/metrics.py; tests: physics/tests/test_metrics.py.

  • 136 halogenated gases with a lifetime of at least a year, recomputed from lifetime, radiative efficiency and a molar mass derived from the formula: all five metrics (GWP-20/100/500, GTP-50/100) within 1.6 % of AR6; 135 of 136 within 1 %.
  • Gases living under a year (and two entries with no usable formula) are not recomputed; AR6 values are shown instead, because the result there is dominated by the carbon-cycle term and by inputs the public table rounds away.
Gas metric this model AR6
CH4 GWP-100 27.86 27.9
N2O GWP-100 273.4 273
CFC-12 GWP-100 1.248e+04 1.25e+04
HFC-134a GWP-100 1525 1530
SF6 GWP-100 2.521e+04 2.52e+04

Stratospheric adjustment (fixed dynamical heating)

Method and approximations: physics/radforce/spectra/adjust.py; computation: build/stratospheric_adjustment.py; tests: physics/tests/test_adjust.py, physics/tests/test_adjustment_results.py.

Change (all-sky) instantaneous, top of atmosphere (W m-2) stratosphere-adjusted, this model AR6 stratosphere-adjusted difference
2 × CO₂ (278.3 → 556.6 ppm) 2.01 3.82 3.75 +2 %
CH₄ 1750 → 2019 (729 → 1866 ppb) 0.51 0.55 0.64 -14 %
N₂O 1750 → 2019 (270 → 332 ppb) 0.19 0.20 0.20 -2 %

For 2 x CO2 the adjustment adds 1.81 W m-2 at the top of the atmosphere and removes 0.48 W m-2 at the tropopause (instantaneous 4.30); both ends agree after adjustment (3.824 and 3.825). With AR6's +5 % tropospheric adjustment the model effective forcing is 4.02 W m-2 against AR6's 3.93. The methane forcing is 14 % below AR6's; the cause was not isolated.

Warming and scenarios (FaIR v2 with the median AR6-constrained climate parameters)

Climate response: three-layer energy balance model, ECS 3.00 K, TCR 1.72 K (AR6: 3 K [2.5-4], 1.8 K [1.4-2.2]). The Python and TypeScript implementations both reproduce the FaIR run (physics/tests/test_ebm.py, web/tests/ebm.test.ts).

Scenario 2081-2100 warming, this run (K) AR6 SPM Table SPM.1: best estimate [very likely range]
SSP1-1.9 1.36 1.4 [1.0-1.8]
SSP1-2.6 1.73 1.8 [1.3-2.4]
SSP2-4.5 2.59 2.7 [2.1-3.5]
SSP3-7.0 3.44 3.6 [2.8-4.6]
SSP5-8.5 4.27 4.4 [3.3-5.7]

Present-day check: modelled 2011-2020 warming (1850-1900 baseline) 1.00 K; HadCRUT5 1.12 K; AR6 assessed 1.09 K [0.95-1.2]. CO2 in 2024: SSP2-4.5 424.4 ppm, Mauna Loa 424.6 ppm.

Limits: one deterministic run per scenario (no ensemble spread), global-mean series only, no regional climate.