An additive geostatistical model for mixing total and partial PM10 observations with CHIMERE rCTM
| Autor: | Chantal de Fouquet, Maxime Beauchamp, Laure Malherbe, Frédérik Meleux, Bertrand Bessagnet, Anthony Ung |
|---|---|
| Přispěvatelé: | Institut National de l'Environnement Industriel et des Risques (INERIS), MINES ParisTech - École nationale supérieure des mines de Paris, Université Paris sciences et lettres (PSL), Centre de Géosciences (GEOSCIENCES), Université Paris sciences et lettres (PSL)-Université Paris sciences et lettres (PSL) |
| Rok vydání: | 2018 |
| Předmět: |
Atmospheric Science
education.field_of_study 010504 meteorology & atmospheric sciences Population Univariate CHEMISTRY-TRANSPORT MODEL COKRIGING Particulates Spatial distribution 01 natural sciences 010104 statistics & probability PARTICULATE MATTER 13. Climate action Kriging [SDE]Environmental Sciences Statistics Environmental science Fraction (mathematics) Limit (mathematics) 0101 mathematics education Air quality index 0105 earth and related environmental sciences General Environmental Science |
| Zdroj: | Atmospheric Environment Atmospheric Environment, Elsevier, 2018, 189, pp.61-79. ⟨10.1016/j.atmosenv.2018.06.035⟩ |
| ISSN: | 1352-2310 |
| DOI: | 10.1016/j.atmosenv.2018.06.035 |
| Popis: | International audience; European Legislation (Directive, 2008/50/EC) requires the Member State to identify the areas where limit or target values are exceeded and to estimate the population exposed to these exceedances. The monitoring network of particulate matter with diameter less than 10 μs (PM10) and 2.5 μm (PM2.5) is homogeneous enough in Western Europe to describe accurately their daily spatial distribution. Daily mapping is required to correctly quantify the population exposed to the exceedances. Geostatistical methods are commonly used in Air Quality to interpolate the observations by taking into account the spatial dependencies between the data. The case of particulate matter is specific: in addition to the total PM10 observations, there are, for historical reasons, measurements of the so-called non-volatile fraction of the particles. The non-volatile fraction dataset is corrected and used as an observation of the total PM applied for mapping. It provides therefore a meaningful information on the chemical composition of the particles. Along this line, PM2.5 data, which are a subset of PM10, also bring an important information on the spatial distribution of the PM10, and conversely. This work demonstrates the importance of keeping an information considered as secondary in the monitoring network and how it is possible to improve the estimation of PM concentrations. An additive modelling for PM10 and its subset is used to link the explanatory variables between them and the related cokriging is presented. Maps and scores are shown and confronted to univariate kriging estimations throughout the first three months of 2015 in metropolitan France. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect