couple2 for NEI2016InlineEmis (elevated point sources) - #233
Open
xk-y wants to merge 12 commits into
Open
Conversation
Restore the disabled het/cloud sinks as parameter coupling seams (default 0 = inert, mirroring k_cld/k_mt) and wire them to Aerosol.jl through a new AerosolExt: - geoschem_fullchem.jl: expose k_het_N2O5/HO2/NO2/NO3 (s^-1) and enable the first-order het uptake reactions N2O5->2HNO3, HO2->(sink), NO2->0.5HNO3+0.5HNO2, NO3->HNO3 (the dominant nighttime NOx / HOx heterogeneous sinks). - ext/AerosolExt.jl: couple2(GEOSChemGasPhase, AerosolDistribution) sets each k_het = S_t*c_bar*gamma/4 (GEOS-Chem Ars_L1k kinetic limit); couple2( GEOSChemGasPhase, CloudChemistryFixedpH) sets k_cld1 from the in-cloud H2O2 S(IV) oxidation rate (S&P Eq. 7.75). - Project.toml: Aerosol weak dependency + AerosolExt extension; add Aerosol and OrdinaryDiffEqNonlinearSolve to the test target. - test/AerosolExt_test.jl: het params/reaction present, het rates driven by S_t, in-cloud SO2+H2O2->SO4 with sulfur conserved. Defaults are no-ops, so base GEOSChemGasPhase behaviour is unchanged (the existing mechanism sensitivity tests pass unmodified). Requires an Aerosol.jl providing the AerosolDistributionCoupler / CloudChemistryFixedpHCoupler types. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
FC=1.0 (100% cloud everywhere) over-depleted H2O2. 0.15 is a prescribed climatological mean; interim until coupled from GEOSFP A3cld CLOUD. GC cloud-couple2 FC hunk only. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The O1D finite-difference sensitivity is a near-cancellation quantity: at the default tolerance + 10% perturbation the signal sat below the solver accuracy floor (reltol * O3 ~ 2e-9) and drifted with structural changes such as the new (default-off) het uptake reactions. Tight tolerances (abstol=reltol=1e-10) and a 50% perturbation make it a converged, reproducible value; rtol=0.01. This is the O1D hunk only — the OH/HO2 hunks are already in upstream main via PR EarthSciML#222. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- NH3/NH4 species in GEOSChemGasPhase + NEI ammonia emission coupling (NH3 + NH3_FERT sectors; Vector-of-pairs translate table so duplicate LHS species are preserved) - IsorropiaOp (src/isorropia_operator.jl): k_mt = 1/300 s^-1 mass-transfer relaxation toward the NH4-SO4-NO3 metastable equilibrium (matches the common 300 s Strang step) - AerosolExt get_odefunction: faithful ISORROPIA II regime split + bisection solve (_iso_ternary, robust across all composition regimes), per-cell met read with tri-level H2O resolution (observed / state / constant-species parameter), convergence-gated graceful fallback, condensed-phase electroneutrality anion cap on aerosol NH4 - fixes swapped sulfate/nitrate totals in the solver call; documents k_mt*dt semantics and the per-cell pressure source - tests: isorropia_op_test.jl (regimes, conservation, met-coupled RH sensitivity, robustness) + NEI-NH3 coupled invariants The tri-level H2O read is written so this coupling is independent of merge order with respect to the met-driven H2O change. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
38 tasks
Also exempt 'Strang' (Gilbert Strang, operator splitting) in _typos.toml. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Species: re-enable SO2, add SO4/NH3/NH4/NIT/N2O5/NO3 (7 new states). Hooks
k_cld1/k_het_{N2O5,HO2,NO2,NO3} default 0 => bare SuperFast chemistry unchanged.
Couplings (ext/AerosolExt.jl, mirroring the GEOS-Chem templates):
* SuperFast<->CloudChemistryFixedpH: aqueous S(IV)+H2O2 -> k_cld1 (SO2+H2O2->SO4);
p_NH3 now fed from the mechanism's real NH3.
* SuperFast<->AerosolDistribution: S_t-driven k_het_* (kinetic-limit uptake).
SF<->NEI coupling gains SO2/SULF->SO4/NH3/NH3_FERT emissions (MW_SULF=98.079e-3),
Vector-of-pairs translate table (duplicate left species preserved).
H2O treatment: H2O stays a met-coupled isconstantspecies PARAMETER (the
GEOS-Chem met-driven fixed-species design); SuperFast<->GEOSFP couple2 keeps
param_to_var(:T,:P,:H2O) + RH-driven c.H2O; IsorropiaOp reads it through its
tri-level path (already on this base).
IsorropiaOp perf + diagnostics: cell loop Threads.@threads :static with per-thread
obscaches (du bitwise-identical regardless of thread count), skip gate
(<1e-4 ppb combined inorganic N), _iso_ternary nbisect 60->30 + warm +-1.5-decade
bracket + width break (equivalence vs the exact solve <=6.4e-6
rel at trace extremes = <2e-6 ppb absolute; physically negligible vs k_mt), and a
per-sim-hour non-convergence counter (@info).
N2O5/NO3 are dead-by-construction in reduced SuperFast (no gas-phase production;
IC=0 stays 0) — uptake channels wired but inert until a nighttime-NOx
mini-mechanism lands.
Tests: new superfast_aerosol_test.jl (inert defaults, bare-SuperFast O3
equivalence [baseline filled next commit], cloud S conservation, S_t/k_het
observed exposure, triple-coupling compile).
The GC<->NEI Dict->Vector hunk and the tri-level H2O read slice are submitted
with the NEI-31-pair and ISORROPIA PRs respectively.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ory checkpoints) Measured on the reduced SuperFast baseline: hourly bare-SF box O3. The additions are inert (zero-rate/constant with defaults) so bare-SuperFast chemistry is unchanged. Checkpoints at h2/h4/h8 replace the 24h endpoint, which decays to ~1e-7 ppb where adaptive-step noise dominates relative comparisons. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…ip gate * SuperFast cloud-couple2 FC 1.0 -> 0.15: same interim climatological-mean fix as the GEOS-Chem block (FC=1.0 over-depleted H2O2). * ISORROPIA skip gate (b): compositions with no aerosol-forming anion material (SO4 + total nitrate + existing NH4 < 1e-3 ppb) have a trivial all-gas equilibrium and are exactly the degenerate compositions where the activity/water fixed point burned all 40 outer iterations without converging (a large fraction of early solves in these cells); their |du| bound is k_mt*1e-3 ~ 3e-6 ppb/s ~ 0. The GC cloud FC hunk is in the het/cloud coupling PR. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…olver The SuperFast aerosol-species addition grows the state vector 12->19; update the two coupled-system unknowns-count assertions accordingly (breaking, minor bump). superfast_aerosol_test.jl used the OrdinaryDiffEq umbrella (not a declared test dep); it only needs Rosenbrock23, which OrdinaryDiffEqRosenbrock (already in [extras]/[targets]) provides. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The cloud/het couple2s reference SuperFast's T/P in variable form by design (they are bound by the met coupling in production; see the NOTE in ext/AerosolExt.jl), so coupling them WITHOUT a met source is intentionally unbalanced — newer MTK correctly rejects it. Add SFTestMet, a minimal met stand-in mirroring the production GEOSFP binding (param_to_var(c,:T,:P) + connector equations), to the three standalone-couple testitems. The added testitems assert all three couples compile (19 unknowns) and check in-cloud S conservation. Two MTK v11 compat fixes in the same tests: convert(System, csys) already compiles (an outer mtkcompile now raises 'double simplification'); positional prob.u0 indexing is no longer guaranteed to match unknowns(sys) ordering, so u0 overrides and readouts use symbolic indexing. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…uperFast + GEOS-Chem Maps the inline (elevated) point-source emission variables (SO2/NO/NO2/CO/NH3, already in kg/kg/s mixing-ratio rate at the plume-injection layer) into both mechanisms with the same uconv/MW pattern as the surface NEI coupling. The vertical placement lives entirely in the emissions System's wrapper, so these couplings are layer-agnostic. Companion to the NEI2016InlineEmis feature in EarthSciData (feature + validation details there). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
opt-in · non-breaking
Change
Maps the inline point-source variables (SO2/NO/NO2/CO/NH3, already a mixing-ratio rate at the plume-injection layer) into SuperFast and GEOS-Chem with the same
uconv/MWpattern as the surface NEI coupling. Layer-agnostic — vertical placement lives entirely in the emissions System's wrapper. Pure addition: two newcouple2methods, no existing code or defaults touched.Dependencies
EarthSciData
NEI2016InlineEmis— EarthSciML/EarthSciData.jl#218 (closes EarthSciData#211) merged + tagged; the GasChem SuperFast-aerosol-species PR (#231) for the SF SO2/NH3 targets and — via its stack — the GasChem ISORROPIA PR (#230) for the GEOS-Chem NH3 species (which also inherits the Aerosol 0.4.2 gate).Review order - part of the coordinated train (#225)
Base =
main. Stacked on SuperFast (#231); net-new = the couple2-wiring commit. CI is red until the EarthSciData NEI-inline PR (EarthSciML/EarthSciData.jl#218) merges + EarthSciData tags (and, inherited via the parent stack, the Aerosol 0.4.2 gate).Part of the coordinated cross-repo update tracked in #225.