Toward stable, general machine-learned models of the atmospheric chemical system

Atmospheric chemistry models—components in models that simulate air pollution and climate change—are computationally expensive. Previous studies have shown that machine-learned atmospheric chemical solvers can be orders of magnitude faster than traditional integration methods but tend to suffer from numerical instability. Here, we present a modeling framework that reduces error accumulation compared to previous work while maintaining computational efficiency. Our approach is novel in that it: 1) uses a recurrent training regime that results in extended (>1 week) simulations without runaway error accumulation, and 2) can reversibly compress the number of modeled chemical species by >80% without further decreasing accuracy. We observe a ~260× speedup (~1900× with specialized hardware) compared to the traditional solver. We use random initial conditions in training to promote general applicability across a wide range of atmospheric conditions. For ozone (concentrations ranging from 0-70 ppb), our model predictions over a 24-hour simulation period match those of the traditional solver with median error of 2.7 ppb and <19 ppb error across 99% of simulations initialized with random noise. Error can be significantly higher in the remaining 1% of simulations, which include extreme concentration fluctuations simulated by the reference model. Results are similar for total particulate matter (median error of 16 mg/m³ and <32 mg/m³ across 99% of simulations with concentrations ranging from 0-150 mg/m³). The machine learning models described here are not yet replacements for traditional chemistry solvers but represent a step toward that goal.