Détermination à l'aide d'un modèle récepteur des zones sources à l'origine des concentrations mesurées dans les précipitations collectées en trois sites du réseau MERA (France) Source receptor modeling using precipitation data collected at three rural French sampling stations of MERA network

Résumé

Ces travaux s'inscrivent dans le cadre du programme national de MEsure des Retombées Atmosphériques (MERA). Ils portent sur la recherche de l'origine des précipitations collectées entre 1997 et 1999 dans trois (Morvan, Iraty, Le Casset) des onze stations du réseau MERA localisées en différents points du territoire français. Deux méthodes statistiques ont été utilisées dans cette étude. Les régions à l'origine des fortes concentrations mesurées au site récepteur ont d'abord été déterminées à l'aide d'un modèle (méthode de Seibert) combinant les mesures réalisées sur site et les rétrotrajectoires de masses d'air puis, dans un second temps les différents profils de transport atmosphérique, leur fréquence et concentrations associées ont été évaluées à l'aide d'une classification par Nuées Dynamiques (méthode K-means/distance Euclidienne simple) des rétrotrajectoires de masses d'air. Le test de Kruskal-Wallis a été utilisé pour vérifier si les médianes des concentrations associées à chaque classe sont statistiquement différentes. L'étude réalisée à Iraty (Pyrénées) et au Casset (Alpes) a montré que ces deux stations sont influencées différemment du Morvan. Plus exactement, ces deux sites ne sont pas, ou pratiquement pas, influencés par les zones d'Europe centrale ou du Nord-Ouest fortement émettrices de SO2, de NOx et de NH₃. Seul le pH des précipitations collectées à Iraty semble dépendre des émissions de SO2 et de NOx d'une de ces zones. Iraty et le Casset sont très influencées par les émissions anthropiques et par les poussières d'origine terrestres en provenance d'Afrique du Nord. Néanmoins, les niveaux de concentrations mesurés dans les flux en provenance d'Afrique du Nord sont similaires pour Iraty, le Casset et le Morvan (sauf en ions calcium, pour lequel le Casset et Iraty montrent de fortes concentrations). Une autre région européenne peut influencer les niveaux en composés acidifiants mesurés au Casset, il s'agit de l'Italie et de la zone localisée au niveau de l'ex-Yougoslavie. Mais, les niveaux de concentrations qui en résultent sont faibles par rapport à ceux mesurés dans certains flux arrivant au Morvan.

Mots-clés:

• Précipitations acides,
• zones rurales,
• transport à longues distances,
• retro trajectoires,
• modèle récepteur

Abstract

The chemistry of precipitation in France was examined using data from the French atmospheric deposition network (MERA). In order to examine the source-receptor relationships responsible for acid rain at three background sites in France, a receptor-oriented model was applied to the precipitation data collected from 1997 to 1999. This methodology combined precipitation and chemical data with air parcel backward trajectories to establish concentration field maps of likely contributing sources. Then, a clustering technique using partitioning methods (K-means/Euclidian distance) was performed to backward trajectories and the distributions of mixing samples associated with backward trajectories in each cluster were compared. The Kruskal-Wallis test was used to verify that the concentration medians associated with each cluster were statistically significant. The results of this study demonstrated that two stations (Iraty and le Casset) were not influenced by the same sources as Morvan. Specifically, these sites were less influenced by high emissions from Central or Northwestern Europe when compared to Morvan. Only the pH seemed under the influence of SO₂ and NOx emissions from one of these areas. Iraty and Le casset are very influenced by anthropogenic emissions and the crustal sources around the Mediterranean Basin and North Africa. Other European areas (e.g. Italy) can influence the concentrations recorded at Le Casset but the levels of concentration are lower than those measured at Morvan.

This paper represents a complete statistical analysis of wet-only deposition chemistry data for three stations (Iraty, Le Casset and Morvan). Two statistical methods were used in this study. In order to examine the source-receptor relationships responsible for acid rain at these three background sites in France, a receptor-oriented model was applied to the precipitation data collected from 1997 to 1999. This methodology combined chemical data with air parcel backward trajectories to establish concentration field maps of likely contributing sources. This receptor-oriented model was developed by Seibert and it assumes that if a trajectory endpoint falls in a grid cell (i,j), the air mass is assumed to collect components emitted in this cell and once the components are incorporated, they are transported along the trajectory to the receptor site. This model doesn't take into account the atmospheric diffusion and the removal mechanisms occurring during the trajectory from the sources to the receptor. Finally, a concentration field map for the selected species was calculated taking into account all grid cells. For mapping, the grid cells counting fewer than 10 endpoints were not taken in consideration because the confidence of their results was considered too low. The role of three-dimensional backward trajectories is fundamental, so we used three different information sources: the French Institute of Meteorology, Météo-France; the British Atmospheric Data Centre (BADC); and the Atmospheric Environment Service Long Range Transport model of Air Pollution (AES-LRTAP), Canada. These trajectory models were compared for different chemical species. All data were projected in the EMEP grid (150 x 150 km) for establishment of the concentration field map. A clustering technique by partitioning methods (K-means/Euclidian distance) was performed on backward trajectories and the distributions of mixing samples associated with backward trajectories in each cluster were compared. The Kruskal-Wallis test was used to verify that the median concentrations associated with each cluster were statistically significant.

The results of this study for Morvan determined five classes of backward trajectories associated with the precipitation collected at this station located in the centre of France. The fluxes from SW and WSW sectors contribute for 52% of events, while the fluxes of NW and E contribute for 31% of events but are mainly responsible for high concentrations of sulphates, nitrates, ammonium and hydronium ion. Regions found to be responsible for rain events coincide with European regions known for their high anthropogenic emissions of SO₂ and NOx (Great Britain, North of France, Belgium, The Netherlands and the North of sea).

The results for Iraty (South of France) yielded five classes of backward trajectories associated with the precipitation collected in this station. The fluxes from W sectors (NNW, NW, W and WSW) were responsible for 71% of events, while the flux of S (low wind) was responsible for 29% of events but is mainly responsible for high concentrations of sulphates, nitrates, ammonium and calcium. High concentrations of hydronium ion were identified in the NNW sector.

The results for Le Casset (East region and mountainous) gave four classes of backward trajectories associated with the precipitation collected in this station. The fluxes from W and WSW sectors were responsible for 35% of events, while the flux of SSW was responsible for 43% and the flux from the SE was responsible for 22% of events. This last sector was mainly responsible for high concentrations of sulphates, nitrates, ammonium and calcium. The concentrations measured at this station were low. Regions found to be responsible for rain events coincide with southern and eastern areas known for their high anthropogenic emissions of SO₂ and NOx (north Africa, northern Italy, Yugoslavia).

All these results demonstrate that the Iraty and Le Casset stations were not influenced by the same sources as Morvan. Specifically, these sites were less influenced by the high emissions from central or northwestern Europe than Morvan. Only the measurement of pH seemed to be under the influence of SO₂ and NOx emissions of one of these areas. Iraty and Le Casset were very influenced by the anthropogenic emissions and the crustal sources around the Mediterranean Basin and North Africa. Other European areas (e.g., Italy) can influence the concentrations recorded at Le Casset but the levels were lower than those measured at Morvan. A relation between sulphates, nitrates and ammonium was identified for Morvan and Le Casset. This observation suggests that aerosol transport of NH₄ HSO₄, (NH₄)₂ SO₄ and NH₄ NO₃ is occurring.