loading page

Updated aerosol optical properties for the Goddard Earth Observing System (GEOS) Earth system model
  • +1
  • Osku Kemppinen,
  • Peter Colarco,
  • Reed Espinosa,
  • Patricia Castellanos
Osku Kemppinen
NASA Goddard Space Flight Center

Corresponding Author:[email protected]

Author Profile
Peter Colarco
NASA Goddard Space Flight Center
Author Profile
Reed Espinosa
NASA Goddard Space Flight Center
Author Profile
Patricia Castellanos
NASA Goddard Space Flight Center
Author Profile

Abstract

The Goddard Earth Observing System (GEOS) global atmospheric model tracks and transports several individual aerosol species. A key contribution by these aerosols is to interact with solar radiation. When calculating the aerosol-radiation interactions, pre-computed look-up tables of aerosol optical properties are used.We have recently finished an effort to update the aerosol optical properties used in the GEOS model. This has been accomplished with a full rewrite of the optical property simulation code into a Python-based version. In addition to structural changes to the code (outlined below), there are some concrete changes to the aerosol particle definitions. First, truncated lognormal size distributions have been replaced with non-truncated ones, and sub-bin size resolutions of all simulations have been enhanced. Additionally, non-spherical dust accuracy has been improved due to switching to a different spheroidal kernel database. Finally, dust size distribution has been changed from a continuous power law distribution to a bin-specific lognormal distributions.There are also significant changes in the convenience and flexibility of adding or modifying aerosol types for future work and other investigations. Particle properties are now defined exclusively in resource files rather than in the code. This allows for easy addition or modification of aerosol particles, with no need to touch the code. Relatedly, the code can be run either with a provided command-line interface, or via a custom script or notebook. That is, no Python knowledge is needed to generate aerosol particle optics, making the tool easy to use. Additionally, particle types are fully decoupled from specific size distributions, hydration schemes, and other type-specific properties. This also means changing, e.g., a particle size distribution type for a given particle is simple. Performance is significantly improved due to novel Mie simulation optimizations. This enables the aforementioned high size resolutions to be used without concerns for processing time.