Exploratory study for estimating atmospheric low level particle pollution based on vertical integrated optical measurements
Résumé
We present a method for retrieving atmospheric particulate matter (PM10) from sun-sky photometer measurements (AOT). As PM10 is a "surface parameter" and AOT is an "integrated parameters", we first determined whether a "functional relationship" linking these two quantities exists. Since these two parameters strongly depend on atmospheric structures and meteorological variables, we classified the meteorological situations in terms of weather types by using a neuronal classifier (Self organizing Map). For each weather type, we found that a relationship between AOT and PM10 can be established. We applied this approach to the Lille region (France) for the summer 2007 and then extended to a five summer period (summers of the years 2003-2007) in order to increase the statistical confidence of the PM10 retrieval from AOT measurements. The good performances of the method led us to envisage the possibility of deriving the PM10 from satellite observations.