EASLEE v1.1: an Efficient AtmoSpheric Lagrangian Easy-to-usE particle dispersion model written in Julia

Abstract. Lagrangian particle dispersion models (LPDMs) are widely used tools in atmospheric sciences, from the interpretation of in situ measurements to source–receptor analysis and inverse modeling of surface fluxes. Their practical use, however, still requires a level of technical expertise (compilation of Fortran code, per-dataset preprocessors, fragmented configuration files) that is unrelated to the physics of the problem being studied. In this paper, we present EASLEE (Efficient AtmoSpheric Lagrangian Easy-to-usE model) version 1.1, a new Lagrangian particle dispersion model of the FLEXPART family, written entirely in the Julia language. Because Julia is a compiled language with just-in-time compilation, no compilation step is exposed to the user, and the same source code runs unmodified under Linux, macOS and Windows, and a run is defined by a single configuration file or by a web-based graphical user interface. EASLEE ingests meteorological fields from ERA5, NCEP GFS, MERRA-2, AROME, Meso-NH and WRF through a common level-index vertical transport formulation, and includes boundary layer turbulence (Hanna and Thomson well-mixed Langevin schemes), a mass-flux convection scheme that is well mixed by construction, an isentropic–kinematic hybrid vertical transport option for the stratosphere, dry and wet deposition, gravitational settling, radioactive decay, size-resolved aerosols with hygroscopic growth and aerosol optical depth diagnostics, biomass-burning plume rise, and mineral-dust and sea-spray emission. Forward and backward (source–receptor) transport are supported. The gridded output uses an adaptive kernel in which each parcel deposits its mass over a support set by the local inter-parcel spacing, so that the spatial resolution of the model is determined by the number of trajectories rather than by the output grid. A companion resolution field reports this true local resolution at every grid cell. The model is evaluated against the ETEX-1 and CAPTEX-83 tracer experiments, for which EASLEE obtains composite ranks of 2.76 and 2.43, respectively, comparable to FLEXPART and other established LPDMs. In terms of computation efficiency, the one-million-parcel, two-day ETEX-1 reference run completes in 12 min on 16 threads of a workstation, and in 5 min on its GPU. The source code, configuration examples and a step-by-step description of the installation are provided to the reader.