Lagrangian particle dispersion models are indispensable tools to study atmospheric transport processes. Here, we will discuss the development, evaluation, and applications of the Lagrangian particle dispersion model MPTRAC (Massive-Parallel Trajectory Calculations) developed at the Jülich Supercomputing Centre.

Next to calculating air parcel trajectories, MPTRAC allows to simulate the effects of turbulent diffusion and sub-grid-scale winds, convection, sedimentation, radioactive decay, hydroxyl chemistry, dry deposition, and wet deposition. Simulations are driven by meteo data sets provided by ECMWF, NASA, or NCAR/NCEP. Various output methods for particle, ensemble, gridded, sample, and station data have been implemented.

So far, the model was mostly applied to study transport processes in the upper troposphere and lower stratosphere (UT/LS) region. In particular, we used MPTRAC to estimate volcanic sulfur dioxide emissions and to simulate their long-range transport. We also used the model to compare transport simulations conducted with different reanalyses and for comparisons with superpressure balloon trajectories.

Long-term and large-scale Lagrangian transport simulations require efficient algorithms and implementations. MPTRAC features an MPI/OpenMP/OpenACC hybrid parallelization for use on HPC systems. So far, the largest simulations have been conducted on the JUWELS and JURECA supercomputers in Jülich and on Tianhe-2 in China. Our current work is focusing on making MPTRAC fit to run with comprehensive meteo data sets such ECMWF's ERA5 reanalysis and to speed up the simulations by utilizing graphics processing units (GPUs).

MPTRAC is open-source and free to use: https://github.com/slcs-jsc/mptrac