EU/DG-XII, Contract : ENV4-CT97-0593

The overall goal of the MAMCS project was to improve our understanding of the mechanisms influencing the dynamics of mercury in the Mediterranean Sea region including:

emissions from natural and anthropogenic sources,

atmospheric transport and deposition to water and terrestrial receptors,

chemical and physical transformations of Hg⁰ and other Hg species in the atmosphere with changing meteorological conditions.

The modelling effort at the framework of MAMCS includes the state-of- the art meteorological and dispersion models chemical-physical transformation models, dry and wet deposition models and mercury emission inventory (MEI) database. In order to calibrate and validate the performance, environmental parameters (e.g. ambient concentrations of atmospheric mercury, deposition, fluxes, meteorological parameters) were measured at five sites during four field campaigns.

At the framework of MAMCS project two atmospheric numerical modelling system, namely RAMS and SKIRON/Eta, have been for development and simulation of the mercury physicochemical processes. The parallel development and implementation of the mercury processes in two atmospheric models allows the intercomparison of the results and consequently the more accurate development of the mercury modelling system. Both RAMS and SKIRON/Eta are 3-D full-physics limited-area models and they have similar capabilities. They can be used for high-resolution simulations and in this way they can represent satisfactorily regional and mesoscale features. Mesoscale and large-scale precipitation processes are important for the wet deposition of mercury. Also, both atmospheric models include highly accurate turbulence schemes. This is important since the dry deposition of mercury is strongly dependent on turbulence near the surface. Any uncertainties related to wet and dry removal processes were tested extensively.

RAMS includes a detailed cloud microphysical scheme and it has the capability of two-way interactive nesting. Sensitivity tests indicated that a very detailed cloud microphysical scheme is not absolutely necessary for the mercury removal processes because in this study emphasis is given on regional and synoptic scale features. SKIRON/Eta is an accurate model and is appropriate for long period simulations since it does not require significant computer resources.

The elemental, reactive and particulate mercury were taken into account in the model development. The model used the most detailed and accurate emission inventory created during the project. Moreover, it treated physico-chemical processes, atmospheric reactions, transformations, removal processes and especially the aqueous phase chemistry and gas-to-solid partitioning of elemental mercury.

Four different scenarios of mercury transport and dispersion, one for each season, have been analysed at the framework of the MAMCS project. Model simulations using RAMS and SKIRON/Eta were performed. The main results of this study are presented below:

The Mediterranean Sea region is not only affected by mercury released in its vicinity but also from air masses enriched in mercury from stations in northern and northeastern Europe. This suggests that local and remote emissions must be taken into account in mercury studies in the Mediterranean. This is particularly important for elemental mercury, which can be transported in long-ranges before deposition.

In general, a satisfactory agreement is evident between observations and model output. However, a systematic model evaluation is difficult unless some other controlling factors, like emission inventories and observation quality, are not improved significantly.

Higher values of mercury concentration over the Mediterranean Sea Region were usually observed during the summer period. This is attributed to the synoptic conditions prevailing during the warm period of the year. The dominant flow is from north to south during summer promoting the transfer of pollutants from Europe to the Mediterranean region. Moreover, the summer period in the Mediterranean is characterised by anticyclonic circulations, which are known to be associated with large-scale subsidence and no rain. Therefore, the lower troposphere is usually stably stratified and consequently wet and dry deposition processes are suppressed. The synoptic conditions prevailing during winter lead to higher concentrations of mercury over northern Europe than over the Mediterranean sea basin due to the dominance of quick removal processes over Europe. The annual wet deposition values of particulate and reactive mercury (Hg^P and Hg²) were one order of magnitude higher than the dry ones. The annual dry and wet deposition amounts of elemental mercury adsorbed are comparable. Also, they are about 4 orders of magnitude smaller than those of Hg² and Hg^P.

The deposition patterns showed that the largest amounts of mercury are deposited in eastern Europe and in the Mediterranean region, especially in its eastern part. Taking into account that the vast majority of the mercury sources is located over central and northwest Europe, two main paths of transport are indicated.The one path is from central to eastern Europe and the other is from Europe towards the Mediterranean sea, namely from north to south. This may have important negative implications not only for the fish and agricultural production of the nearby countries but also for the population directly exposed to mercury.

The difficulties in measuring the wet and dry deposition of mercury make the deposition patterns estimated by the model very useful. The models are also helpful in estimating the mercury concentration due to the lack of reliable and consistent measuring methods. A well-developed numerical model is also much cheaper than a dense observation network that is required for high-resolution estimations of the concentration and deposition. From this aspect the developed models should be considered as very useful tools for studying the mercury processes and therefore be used by policy makers.

This study focused on the regional and synoptic transport of mercury. The representation of mesoscale features was not in the aims of this research. This is the main reason that a relatively coarse resolution of about 50 km was used here. Very high resolution simulations with a grid spacing of about 5-10 km are required in order to represent mesoscale phenomena. This kind of simulations could resolve mesoscale transport and could provide an understanding of the effects of local versus remote sources. The significant computer resources required for very high resolution simulations will be available in the near future, and this will allow the study of the mercury cycle on the meso-scale.

Despite the significant modelling effort devoted so far, there is still need for further development. There is a need for a better representation of the gas-phase chemistry and the gas-to-particle conversions. A more accurate and systematic way of measuring the various mercury species and a better understanding of the air-water interactions is necessary.