top of page

Tools

We use models alongside observational datasets to study atmospheric composition and climate.

CESM2 family of models

We are using different configurations of the Community Earth System Model version 2 (Danabasoglu et al., 2020), a CMIP6-generation model, with options to include full tropospheric and stratospheric chemistry, and a modal aerosol scheme.  The following configurations are currently used in our group:

  • ​CESM2-WACCM6 is the most comprehensive configuration of CESM2, with tropospheric and stratospheric chemistry and a modal aerosol scheme fully coupled to ocean and sea ice models.  This configuration was used to generate a 13-member initial-condition ensemble over the period 1950-2014, adding to the three ensemble members conducted by NCAR for CMIP6. These simulations serve as a basis for multiple ongoing projects. [Qindan, Xinyuan, Steph]

  • WACCM6 (atmosphere-only climate model; Gettelman et al., 2019).  This configuration offers a computationally less expensive approach to the fully coupled climate model.  For example, we are prescribing sea surface temperatures and sea ice fields taken from the CESM2-WACCM6 historical ensemble to conduct a series of sensitivity simulations for attributing tropospheric ozone trends to specific sources.  [Xinyuan]

  • MUSICA (spectral element dynamical core, variable resolution grid) is nudged to reanalysis meteorology for air quality applications (Schwantes et al., 2022).  We are currently using the standard configuration with 1 degree horizontal resolution globally with two-way nesting to 14km over the contiguous United States. [Tao-ma]

  • CESM2-CAM6 is the standard "out-of-the-box" CESM2 configuration (Danabasoglu et al., 2020), featuring the "low-top" CAM6 atmosphere with a simple online aerosol scheme driven by offline oxidants (generated by WACCM6), coupled to ocean and sea ice models. [Steph]

  • CAM-Chem includes fully interactive tropospheric chemistry within the CAM6 atmosphere (Emmons et al., 2020).  CAM-Chem is used for simulations of global tropospheric composition. The standard (default) configuration is with the MOZART-T1 chemical mechanism, along with a volatility basis set (VBS) parameterization for the formation of secondary organic aerosols (SOA) (Tilmes et al., 2019). We are prescribing sea surface temperature and sea ice at present day to conduct model simulations to investigate the OH and methane lifetime responses to climate warming. [Qindan]

  • CESM-GC leverages the modularity of the CESM framework by integrating key components from GEOS-Chem (Lin et al., 2024). Specifically, CESM-GC incorporates HEMCO (The Harmonized Emissions Component), the GEOS-Chem chemistry mechanisms, and photolysis schemes like FAST-JX and Cloud-J. This integration enables comparisons between different model configurations, enhancing our ability to evaluate the impact of uncertain atmospheric processes. [Lucas]

GEOS-Chem model

The GEOS-Chem chemistry-transport model is a global model driven with assimilated meteorology and has been extensively used in atmospheric chemistry.  It was originally developed and maintained by the Harvard University Atmospheric Chemistry Modeling Group.  Different configurations are available, including two-way regional nesting (variable grid resolution). We have used GEOS-Chem extensively in past work to examine trends in atmospheric composition as well as for source attribution during regional pollution events.

Svante High-Performance Computing Cluster

We joined the Svante cluster at MIT, housed at the Massachusetts Green High Performance Computing Center.  Svante includes 36 new compute nodes based on the recently released 3rd generation scalable Intel Xeon Gold chip 6336Y, in a dual-core configuration (48 cores/node). Briefly, the 6336Y operates at 2.4 Ghz and boasts 8 channels of DDR4/3200 Mhz memory bandwidth, minimizing potential memory bandwidth bottleneck issues. Svante includes file servers with over 3.5 PB.  More information on high-performance computing within PAOC available here.  Several different configurations of CESM and GEOS-Chem are run routinely on Svante.

In progress: Plume-tracking tool

 

 

 

 

 

 

 

 

Tracking pollutants as they disperse in the atmosphere is challenging because of highly nonlinear interactions between physics and chemistry. In addition, observations are not always available, and models overestimate how fast pollutants disperse. We are developing a set of tools that incorporates both observational data and simulations to track pollution over long distances. Our approach relies on locating dynamic barriers (in black in the illustration) that serve as channels for the dispersion of pollutants (in the illustration, carbon monoxide emitted by wildfires in California). [Louis]

​​

​

​

​

Louis_image.png

Brewer

Brewer was our cluster at LDEO, named to honor of Alan Brewer and Gordon Dobson, discoverers of the Brewer-Dobson circulation pattern in the stratosphere.  Brewer has 32 compute nodes with a total of 1024 2.4MHZ AMD Opteron processors and 4096GB aggregate memory , a head node, a storage node with RAID-6 72TB capacity, Infiniband QDR 40Gb/s fast interconnect for parallel computations, and a wide variety of scientific software installed.

GFDL CM3/AM3 Chemistry-Climate Model

Work at LDEO/Columbia used the CMIP5-generation NOAA GFDL general circulation model that includes fully coupled tropospheric and stratosphere chemistry in a single mechanism and aerosol-cloud interactions, coupled to a dynamic vegetation land model (LM3), either coupled to ocean and sea-ice models (CM3) or forced with sea surface temperatures and sea ice distributions (AM3).   AM3 includes an option to nudge the meteorology to “real” winds, enabling us to interpret “snapshot” observations and to evaluate model processes directly with measurements at specific locations and times in the same model used to simulate changes in atmospheric composition and climate.

bottom of page