Inverse modelling of microplastics sources and sinks in the Mediterranean by assimilating observational data and Lagrangian models

Plastics entering the ocean face a fate which is still relatively unknown: how much of it will stay afloat, how quickly will it degrade, and how much of it will end up on the beach? Estimates of plastic inputs are orders of magnitude larger than the quantities found in the surface waters. In order to get a better understanding of the fate of these plastics, an inverse modelling methodology is presented here for a Lagrangian ocean model, in order to predict floating microplastic quantities in the Mediterranean.

While measurements of microplastic quantities match relatively well to larger scale patterns predicted by numerical models for the oceans on a global scale, extensive validation of numerical models with actual measurements for a smaller scale basin such as the Mediterranean is lacking. The Mediterranean is an interesting test case for modelling microplastics because of two reasons. First of all, numerical studies and field measurements suggest that there are no steady regions where microplastics accumulate. This absence of steady attractors means that it is even more important to take in account transient processes in the model, such as beaching and sinking of microplastics, seasonal changes in e.g. river run-off, and the microplastic sources. Secondly, a large number of field studies measuring microplastic concentrations in the Mediterranean are available, giving us valuable information that can be used to tune our numerical models.

In this research, field measurements of plastic concentrations in the Mediterranean are used to inform parameterizations of processes affecting the distribution of microplastics. Parameterizations are implemented in a Lagrangian framework for beaching, fragmentation, sinking, and for various sources of microplastics. The parameters of the models are then found using inverse modelling techniques to best fit the measurements of floating microplastic concentrations. It is shown that this leads to a better match of the model with respect to the measurements compared to a baseline model without these parametrizations. Furthermore, the inferred parameters can be used to rank the processes affecting microplastic dispersal, and thus improve our understanding of the plastic mass budget.