The Atmospheric Dispersion And Parallel Transport (ADAPT1) program can be used to evaluate various OMP parallel statements in the context of atmospheric dispersion calculations to optimize the number of cores available in a processor. Program elements including reading ARL packed meteorological data files, converting meteorological data to computational units, computing particle advection and dispersion, and the accumulation of particle mass on a concentration grid. Mass transformations or deposition are not available. The current version only supports the use of conformal projection meteorological data on sigma surfaces for transport and TKE fields for mixing. The computer code and data files described here are distributed under the GNU LGPL license.


Various download options are available, including a sample Windows executable or the actual source code. Use the following statements to compile the code with gfortran:
OPT="-ffree-form -fconvert=big-endian -frecord-marker=4"
gfortran -oadapt -fopenmp ${OPT} ${MAP} adapt.f

The compilation requires the CMAPF routines (version 1.0) to transform between earth and model grid coordinates. Version 1.0 covers the standard conformal map projections centered at the North and South Pole, namely the Polar Stereographic, the Mercator, and the Lambert Conformal projections. Version 1.0 is described more fully in an article, Conformal Map Transformations for Meteorological Modelers, by Albion D. Taylor, published in the February, 1997 issue of Computers and Geosciences (v23, no 1) and can be downloaded from NOAA or alternately from Computers and Geosciences.

Obtaining meteorological data

Calculations can be made using WRF, NAMS, or HRRR meteorological data if processed for use by HYSPLIT. This WRF data archive is described in more detail in A Long-Term WRF Meteorological Archive for Dispersion Simulations: Application to Controlled Tracer Experiments, by F. Ngan and A.F. Stein. Daily WRF data files can be downloaded from directories {avg | inst} as instantaneous or time-averaged data files. WRF output that has been converted to the ARL format in these directories follows the file name convention of wrfout_d01_ccyymmdd.ARL. Only the time-averaged data files contain the TKE field required for dispersion calculations, otherwise ADAPT will estimate the TKE. Note that the HRRR archived data (a variant of WRF) can also be found on the ARL server. In fact, any conformal sigma level data set should be compatible for computations. The sample calculation discussed below can be performed using the meteorological data provided with the HYSPLIT tutorial.

Processing ADAPT output files

There are four types of output files created by ADAPT and three of them can be processed by software provided with the HYSPLIT package. Air concentrations are output in the HYSPLIT concentration binary format, directly in DATEM format, or as a SVG graphic. Trajectory output is in the HYSPLIT trajectory endpoints file format. Provided with the code is a sample SETUP.CFG that can be used to create the concentration output file for CAPTEX release #2 as discussed in the HYSPLIT tutorial. All the post-processing options described for HYSPLIT output in the Tutorial can be applied to the ADAPT output.

Creating the simulation name list: SETUP.CFG

The Atmospheric Dispersion and Parallel Transport (ADAPT) calculation uses the maximum number of available cores unless the environment has been adjusted: export OMP_NUM_THREADS={x}. The code is run according to the variables set in the name list file SETUP.CFG. The User's Guide describes the file contents in more detail. If the file is not found in the startup directory, a default version of the file will be created that sets the parameters to perform the calculation for CAPTEX release #2. All numeric inputs are integers unless indicated as real (R).


Although the ADAPT code and HYSPLIT share some algorithms because they both read the same meteorological data format available through the NOAA ARL READY website, and the output from ADAPT is compatible with many of the HYSPLIT post-processing programs, the internal code structure of the two models is completely different. ADAPT is designed to run with multiple processors in a shared memory environemnt (OPENMP) while HYSPLIT can be run with multiple processors through a distributed memory structure (MPI). The ADAPT code loops through the entire meteorological data grid in space and time, while HYSPLIT only processes a subgrid of the data required to meet the computational requirements of the simulation. Further, ADAPT can only handle single trajectories and basic dispersion calculations, while HYSPLIT supports a wide range of simulation configurations such as radionuclide releases to the atmosphere.