A source attribution system based on Lagrangian simulations, emission inventories and satellite data: an example of application to the MOSAiC campaign

We present here a semi-automatic system of source apportionment analysis for long-term observations sites (e.g. measurement stations) and measurements campaigns. We will use as an example the application to the Multidisciplinary drifting Observatory for the Study of Arctic Climate (MOSAiC) campaign, which involved a one-year-long ship expedition into the Central Arctic (September 2019 - October 2020)

The system is based on the Lagrangian particle dispersion model FLEXPART (Stohl et al., 2005; Pisso et al., 2019). The simulations are driven with hourly ERA5 data at 0.5° horizontal resolution. The cluster of back-trajectories can be released at any time resolution along the period of interest (the higher the resolution, the higher the computational cost). For the one-year campaign, we have chosen a 3-hours resolution, which will therefore represent the resolution of our timeseries of source contribution. The system can also be adjusted for the number of particles released, length of simulations, and aerosol or gas species that need to be simulated, which will therefore be applied to each simulation release. For our example application we use 100000 particles and a maximum time of transport of 30 days, and CO, SO2, BC and a generic air tracer as chosen species.

The simulations can then be coupled with emissions fluxes, to give a description of the transport conditions of a specific species from its source to the point of measurements. We use in our case the ECLIPSE v4 database for anthropogenic emissions and the GFED one for the fires emissions. With a similar approach, any gridded information from model output or satellite data can also be coupled with the air tracer back-trajectories. In our example, the analysis has been coupled to daily-resolved satellite data of sea ice cover to provide an estimate of the sea ice influence and its seasonal variability vs. the influence from the open ocean and continental land surface. The results of the whole set of simulations, including the quick-looks and resulting time series can also be easily automatically organized in directories. In our application, the data and the plots have been collected and distributed on a dedicated website which allow for an easy browsing of the results for the MOSAiC campaign.