Explicit stochastic advection algorithms for the regional-scale particle-resolved atmospheric aerosol model WRF-PartMC (v1.0)

Jeffrey H. Curtis, Nicole Riemer, Matthew WestGeoscientific 2024 · DOI

TL;DR

This paper introduces a novel stochastic advection algorithm for regional-scale atmospheric aerosol models, enabling particle-resolved simulations. By interpreting finite-volume fluxes as particle transport probabilities, the method accurately captures aerosol mixing states, crucial for climate modeling, and shows expected convergence rates as particle numbers increase.

What Problem Does This Paper Address?

Accurately modeling the impact of atmospheric aerosols on climate requires understanding their "mixing state" – how different chemical species are distributed within individual particles. Traditional models often simplify this by assuming internal or external mixtures, which can lead to significant uncertainties in aerosol properties like optical behavior and cloud-nucleating potential. Particle-resolved models can capture this complexity but have historically been computationally too expensive for regional-scale simulations. This paper addresses this gap by developing a computationally efficient stochastic advection scheme that allows particle-resolved aerosol models to be used at regional scales.

Methods

The core innovation is a stochastic advection algorithm that bridges the gap between deterministic finite-volume methods and particle-based simulations. Instead of directly calculating the movement of a continuous concentration, the algorithm treats the fluxes between grid cells as probabilities. Computational particles within a grid cell are then stochastically sampled to determine how many cross the boundary.

Key aspects of the method include:

  • Finite-Volume Interpretation: Standard finite-volume advection schemes are used to calculate fluxes, but these fluxes are then reinterpreted as probabilities for particle movement.
  • Stochastic Sampling: A binomial distribution is used to determine the number of particles crossing a grid cell boundary, introducing randomness that mimics the continuous nature of advection.
  • Order of Accuracy: The authors explore third- and fifth-order advection schemes, noting that dissipative odd-order schemes are preferred as they help dampen the high-frequency noise introduced by the stochastic sampling.
  • Monotonic Limiters: To ensure numerical stability and prevent unphysical results, monotonic limiters are applied, similar to those used in standard WRF-Chem simulations.
  • WRF-PartMC Integration: The developed algorithms are implemented within the WRF-PartMC model, a regional-scale particle-resolved atmospheric aerosol model.

Datasets

The paper does not use external datasets for training. Instead, it employs idealized initial conditions and realistic meteorological data from the North American Mesoscale Forecast System (NAM) model for evaluation. The "datasets" are essentially the simulated meteorological fields and the initial conditions for the aerosol particles.

Key Insights and Ideas

  • Bridging the Scale Gap: The primary insight is that a stochastic interpretation of finite-volume fluxes can make particle-resolved aerosol modeling computationally feasible for regional scales, a significant advancement for climate modeling.
  • Noise and Dissipation: Stochastic sampling inherently introduces high-frequency "noise." The paper demonstrates that dissipative (odd-order) finite-volume schemes are crucial for effectively damping this noise and achieving stable simulations, unlike conservative (even-order) schemes.
  • Convergence is Key: The stochastic method shows expected convergence rates (proportional to 1/√Np, where Np is the number of particles) towards the deterministic finite-volume solution as the number of computational particles increases. This validates the approach.
  • Mixing State Matters: The underlying motivation is the importance of aerosol mixing state for climate impact. This work provides a tool to investigate these effects more accurately at regional scales.

Key Results

  • Successful Implementation: The stochastic advection algorithms were successfully implemented and tested in WRF-PartMC for 1D, 2D, and 3D simulations.
  • Convergence Rates: Numerical experiments confirmed the theoretical convergence rate of 1/√Np for the stochastic error.
  • Odd-Order Schemes Preferred: Odd-order advection schemes (third and fifth) showed superior performance by effectively suppressing high-frequency noise introduced by stochastic sampling, leading to smoother and more accurate solutions compared to even-order schemes.
  • Realistic Simulations: The 3D simulations using WRF-simulated meteorology demonstrated that the method can handle complex terrain and evolving wind fields, producing results comparable to finite-volume solutions with sufficient particles.
  • No Loss of Information: In tests involving particle size distributions, the stochastic method did not lead to a systematic loss of information for rare particle types, even with varying grid cell volumes and air densities.

Limitations and Future Work

  • Computational Cost: While more efficient than traditional particle-resolved methods, the stochastic approach still requires a significant number of computational particles (e.g., 1000 per grid cell) for accurate results, which can still be demanding for very large-scale simulations.
  • Numerical Diffusion: Similar to finite-volume methods, the stochastic approach can introduce numerical diffusion, although this is managed by the choice of scheme and particle number.
  • Future Work: The authors suggest that this work enables direct comparisons between particle-resolved models and those with simplified aerosol representations, isolating differences due to aerosol representation. Future applications could involve quantifying the impact of aerosol mixing state on climate-relevant properties like aerosol absorption and CCN concentration. The code is made publicly available.

Summary updated

Abstract

This paper presents the development of a stochastic particle method to simulate advection in regional-scale models with a particle-resolving aerosol representation. The new method is based on finite-volume discretizations with the flux terms interpreted as probabilities of particle transport between grid cells. We analyze the method in 1D and show that the stochastic particle sampling during transport injects energy at high spatial frequencies, which can be partially compensated for with the choice of a dissipative odd-order finite-volume scheme. We then apply the stochastic third- and fifth-order advection algorithms with monotonic limiters in WRF-PartMC, using idealized and realistic wind fields in 2D and 3D. In all cases we observe the expected convergence rates of the stochastic particle method to the finite-volume solution as the number of computational particles is increased. This work enables the use of particle-based aerosol models on the regional scale.
BibTeX
@inproceedings{curtis2024explicit,
  author = {Jeffrey H. Curtis and Nicole Riemer and Matthew West},
  title = {{Explicit stochastic advection algorithms for the regional-scale particle-resolved atmospheric aerosol model WRF-PartMC (v1.0)}},
  booktitle = {Geoscientific},
  year = {2024},
  doi = {10.5194/gmd-17-8399-2024},
}
Figure