https://www.peren-revues.fr/pollutionatmospherique/1072 Entrées d’index fr 0 https://www.peren-revues.fr/pollutionatmospherique/1265 La notion de représentativité est mentionnée dans de nombreux textes législatifs sur la qualité de l’air. Une définition de la représentativité spatiale d’une station de mesure de la pollution de l’air par le dioxyde d’azote est proposée et testée sur l’agglomération rouennaise pour l’année 2005. Partant d’une notion empirique – une station de mesure est représentative d’un point géographique si les concentrations en ces deux sites sont « suffisamment proches » – nous proposons une formalisation mathématique de la représentativité. La technique du krigeage permet de ramener le problème de la délimitation de la zone de représentativité d’une station à la majoration d’une somme de deux termes : le premier est l’écart entre la moyenne annuelle de la concentration mesurée à la station et son estimation par krigeage au point courant, le second fait intervenir l’écart-type de l’erreur d’estimation. La mise en œuvre pratique fournit la zone de représentativité de chaque station. Certaines de ces zones peuvent se chevaucher ; en tout point des zones de recouvrement, on détermine alors la station la plus représentative. Enfin, la notion de représentativité est examinée en relation avec la probabilité de dépassement d’un seuil fixé. The notion of representativeness is mentioned in many legislative texts on air quality monitoring. The work presented here aims at defining the spatial representativeness of a measurement station with regard to the annual average of NO2, and testing this definition at Rouen urban area for the year 2005.The work is carried out in the spatial probabilistic framework of geostatistics, which is well suited for such tasks. An intuitive definition of representativeness is written as follows : a measurement station is representative of a given geographical location if the concentrations at these sites are sufficiently similar. But pollution is only measured at the station, and remains unknown outside. The kriging technique gives an estimation of the pollution in each point and the standard deviation of the estimation error. Using these values, and a hypothesis of a normal-distributed-error, we can, for each station, select the points that satisfy the notion of representativeness. There will then be a good chance (at a fixed risk) that the concentrations within the “zone of representativeness” do not differ with more than a fixed deviation from the concentration measured at the station.The practical implementation of these theoretical definitions is to produce maps of areas representative of the stations. It is also possible to determine, for any point of a region, the most representative station and its associated deviation (by assigning an acceptable risk to it). Lastly, for each area, it is also possible to indicate the probabilities of overshooting a threshold. lun., 13 mai 2013 11:04:00 +0200 jeu., 05 nov. 2015 11:00:49 +0100 https://www.peren-revues.fr/pollutionatmospherique/1265 https://www.peren-revues.fr/pollutionatmospherique/1069 À Paris, les pollens sont recensés depuis 1985, à partir dʼun capteur situé sur le toit dʼun bâtiment de lʼInstitut Pasteur dans le 15e arrondissement. Afin dʼétudier la représentativité de ce site, un second capteur a été placé successivement à lʼest (rue Audubon dans le 12e arrondissement) en 2003 et 2004 puis au nord-ouest (rue Tocqueville, dans le 17e arrondissement) de 2005 à 2007.Les comptes polliniques totaux obtenus sur les capteurs Pasteur et Audubon ont des profils similaires : les tendances à la hausse ou à la baisse se retrouvent sur les deux endroits ; les pics de concentrations varient en intensité dʼun capteur à lʼautre, mais sont simultanés.Les taxons les plus représentés sur les sites Pasteur et Audubon sont les mêmes mais leur importance relative diffère quelque peu. Les dates de début et de fin de pollinisation sont, quant à elles, souvent proches ou identiques.Les comptes polliniques totaux sur les sites Pasteur et Tocqueville ont des profils similaires : les tendances à la hausse ou à la baisse se retrouvent sur les deux capteurs de façon simultanée. Cependant, Pasteur a tendance à surestimer les concentrations par rapport à Tocqueville : les concentrations totales obtenues sont supérieures de 12 %.Les concentrations journalières de grains des taxons les plus représentés sont globalement très ressemblantes. Lʼécart entre les dates de début et de fin de pollinisation est compris, dans la plupart des cas, entre 0 et 7 jours.On notera que dans environ 95 % des cas, lʼécart du risque allergique lié aux comptes polliniques entre les sites Pasteur, Audubon et Tocqueville est inférieur ou égal à 1 indice, ce qui est satisfaisant dans le cadre de lʼinformation sur le risque allergique.Lʼétude montre que la couverture du capteur historique de lʼInstitut Pasteur est proche de celle des capteurs situés dans lʼest et le nord de Paris. On peut considérer que le capteur situé à Pasteur est représentatif de la couronne parisienne pour lʼinformation sur le risque allergique lié à lʼexposition au pollen. In Paris, pollens are listed since 1985, starting from a pollen trap located on the roof of the Pasteur Institute. In order to study the representativeness of this pollen trap, a second one was successively placed at the east (Audubon street) in 2003 and 2004 then in the North-West (Tocqueville street) from 2005 to 2007.The total pollen counts obtained on the Pasteur and Audubon traps have similar profiles: the increases and decreases are found on the two traps; the peaks of concentrations vary in intensity from one trap to another, but are simultaneous. Taxa the most represented on the sites Pasteur and Audubon are the same but their relative importances differ. For Urticaceae, Betula and Castanea, the results obtained on the Pasteur and Audubon traps are equivalent, with close pollen index on the two traps. On the other hand, in 2004, the number of grains of Cupressaceae-Taxaceae collected on Pasteur is 1.5 times more important than on Audubon. For Platanus, there are 3 times more pollens on Audubon than on Pasteur. Lastly, for Poaceae, the results obtained on the two traps are not statistically equivalent over the seasons 2003-2004, with higher pollen counts (18%) on the Audubon trap. For these six taxa, the observation of the daily concentrations obtained on the two traps shows that their profiles are very connected. The global aspects of the two curves are the same, whatever the sensor considered. The dates for the beginning and the end of pollination are often close, and identical on several occasions.The total pollen counts on the Pasteur and Tocqueville traps have similar profiles: the increases and decreases are found on the two traps in a simultaneous way. However, the Pasteur trap tends to over-estimate the concentrations compared to the Tocqueville trap: the total concentrations obtained on Pasteur are higher (12% more) than those of Tocqueville. The observation of the daily concentrations of grains of Cupressaceae-Taxaceae, Urticaceae, Betula, Castanea, Poaceae and Platanus shows that the profiles of the daily concentrations obtained on the two traps, for each of the six taxa studied, are overall very similar. The difference between the starting dates of pollination and those of end of pollination varies, in most of cases, between 0 and 7 days. However, an important peak of concentration of Urticaceae was detected on the Pasteur sensor between the 13th and 16th on April of 2007 whereas the number of grains collected on Tocqueville remained weak. In the same way, on April 23, 2006, a sudden increase in the number of grains of Platanus was observed only on the Tocqueville trap.It is also interesting to compare the differences in allergic risks obtained from the pollen counts of the various traps during these 5 years. It will be noted that in 93% of the cases, the variation of the allergic risk related to the pollen counts of birch between the Pasteur and Audubon sites (2003 and 2004) are lower or equal to 1 index, which is satisfactory. This figure goes up to 96% of the cases for the grasses pollens. Concerning the years 2005 to 2007 (comparison between the Pasteur and Tocqueville sites), in 97% of the cases the difference in index is lower or equal to 1 for pollens of birch whereas that accounts for 95% of the cases for the grasses.The study shows that the cover of the historical trap of the Pasteur Institute is close to the traps located in the east and the north of Paris. There is not any significant difference between the Pasteur and Audubon sites from the point of view of the accounts of pollens. The results obtained on Pasteur are higher of 10% than those obtained on Tocqueville but, just as between Pasteur and Audubon, the principal taxa collected are the same (Urticaceae, Poaceae, Cupressaceae-Taxaceae, Betula, Castanea…) and their order of appearance per quantity of grains of pollen is very close for the two studies. For all taxa with allergic interest, the temporal variations of concentrations are close. It is also the case for Cupressaceae-Taxaceae of which the quantities of grains collected on Pasteur are however 1.5 times higher than on the two other traps. The trap located at Pasteur can be considered as representative of the Parisian area for information on the allergic risk related to the exposure to pollen. jeu., 02 mai 2013 12:00:09 +0200 ven., 09 oct. 2015 10:11:50 +0200 https://www.peren-revues.fr/pollutionatmospherique/1069