Overview¶

GENOA (GENerator of Optimized Atmospheric chemical mechanisms) series provides automated, algorithmic tools for reducing detailed atmospheric chemical mechanisms while preserving key processes such as secondary organic aerosol formation and important gas-phase oxidants and pollutants.

Key features in GENOA v3¶

GENOA v3 is a scalable, graph-aware framework for the reduction, analysis, and visualization of atmospheric chemical mechanisms. It offers a suite of integrated tools:

Mechanism reduction
Reduces detailed VOC degradation mechanisms into compact, semi-explicit schemes with user-defined control of size and accuracy.
Mechanism interpretation
Parses detailed mechanisms from multiple sources/generoators (e.g., GECKO-A, MCM, SSH) for adaptation and conversion.
0-D Box modeling
Interfaces directly with the GECKO-A and SSH-aerosol models for simulation-based reduction, testing, and validation. And provides options to output mechanisms in formats for other models.
Post-processing and visualization Provides quantitative analysis and plotting tools to compare mechanism size, performance, and error statistics across reduction levels.

Schematic overview of the GENOA v3 framework.

Reduction at a glance¶

GENOA v3 combines two reduction processes that enable efficient reduction of very large chemical mechanisms containing up to millions of reactions and species.

Threshold-based reduction (TBR) – fast, rule-driven pruning using physical, structural, and kinetic thresholds to trim large mechanisms.
Simulation-based reduction (SBR, in previous versions referred to as training) – accuracy-controlled, iterative refinement guided by 0-D box-model feedback and quantitative error metrics.

Together, these processes form a hybrid-reduction workflow that balances mechanism size, accuracy, and computational cost.
Reduced mechanisms can be evaluated and visualized using built-in tools for mechanism interpretation, box-model simulation, and post-processing.

GENOA v3 hybrid reduction workflow.

Threshold-based reduction (TBR)¶

TBR performs a fast, rule-driven pruning of the mechanism. It screens species and reactions using configurable strategies (TBR strategies) based on intrinsic or condition-dependent properties.

Implemented TBR strategies fall into two categories:

Environmentally insensitive strategies Apply intrinsic filters insensitive of environmental conditions, such as:
- Fast-degrading species jumping, bypassing short-lived intermediates to reduce stiffness and improve solver efficiency.
- Volatility-based classification, removing the gas-particle partitioning of semi-volatile compounds (SVOCs) with saturation vapor pressures above (SVOC -> non-volatile) or below (SVOC -> volatile) given thresholds.
- Reaction pathway removal, modifying entire reaction channels from a specific species based on user-defined criteria, such as:
  - Node depth limit – removing species and reactions beyond a certain step in the degradation pathway from a precursor.
  - NVOC downstream removal – removing further degradation of non-volatile products.
  - VOC branch removal – removing further gas-phase degradation from volatile products when focusing only on SOA formation.
Environmentally sensitive strategies Account for referenced environmetal conditions such as oxidant concentrations and temperature, including:
- Branch concentration removal, removing low-impact pathways based on their contribution to overall species concentrations.
- Yield-to-depth removal, removing high-generation low-yield products within competing reaction channels.
- Absolute product yield removal, removing products with negligible absolute yields from the entire mechanism.
- Reaction branching ratio removal, removing minor competing channels based on their effective branching ratios.

This process is computationally efficient and can typically reduce mechanism size by 40–90%.

Simulation-based reduction (SBR, Training)¶

SBR refines the mechanism iteratively.

In each step, GENOA identifies a set of reduction candidates: species and reaction modifications proposed for removal or alteration based on four reduction strategies and targeting the same species (or groups of species).

Each candidate is evaluated for two key aspects:

Reduction efficiency: Measured by changes in the number of species, reactions, and condensable compounds.
Accuracy: Quantified using reduction errors computed across evalaution targets and evaluation scenarios.

Candidates are applied only if they satisfy user-defined accrucy constraints while maximizing reduction efficiency. Users can thus generate families of reduced mechanisms tuned for different accuracy levels (e.g., low-, medium-, and high-tolerance schemes).

This process can reduce mechanism size by orders of magnitude (e.g. reduced to <0.1% of the reference size), depending on the desired accuracy level and mechanism complexity. Because SBR can be computationally intensive for large mechanisms, it is often preceded by TBR to improve efficiency.

Version Differences¶

The GENOA series¶

The GENOA framework has evolved through three major versions, each extending its capabilities and scalability:

GENOA v1 – introduced a series-reduction workflow targeting SOA-forming pathways.
- It reduced the Master Chemical Mechanism (MCM) sesquiterpene degradation scheme from over 1,600 reactions and 579 species to just 23 reactions and 15 species, with less than 3% error in simulated SOA mass.
GENOA v2 – implemented a parallel-reduction design to handle multiple precursors simultaneously.
- It reduced an MCM mechanism with 3,001 reactions and 1,227 species describing three monoterpenes was reduced to 197 reactions and 110 species, with about 3 % error in simulated SOA mass.
GENOA v3 – redesigned as a hybrid-reduction workflow capable of reducing fully explicit mechanisms.
- It reduced a six-generation GECKO-A mechanism for limonene (762,747 reactions and 125,314 species) to 98,794 reactions and 14,417 species after threshold-based reduction (87% size reduction with 8% error in SOA and key gas-phase tracers).
- Subsequent simulation-based reduction with 15 % error tolerances further reduced this mechanism with 15% error tolerance to 27 reactions and 20 species (99.994% size reduction) when focusing only on SOA formation, or to 78 reactions and 44 species (99.937%) when also tracking important gas-phase oxidants and pollutants.

Originally, GENOA v1 and v2 were referred to as the GENerator of reduced Organic Aerosol mechanisms, emphasizing their focus on SOA chemistry. In v3, the acronym was redefined as the GENerator of Optimized Atmospheric chemical mechanisms, reflecting its broader applicability to atmospheric chemistry.

Comparison Table¶

Feature	v1	v2	v3
Mechanism scale	~10³	~10³	>10⁶
Mechanism format	MCM	MCM	GECKO-A, MCM
Multiple precursor	No	Yes	Yes
Parallel reduction	No	Yes	Yes
Error species	SOA	User	User
Box model coupling¹	SSH	SSH	SSH, GCK_box
Code extensibility	Low	Low	High

"SSH": SSH-Aerosol model; "GCK_box": GECKO-A box model ↩