Services on Demand
Journal
Article
Indicators
- Cited by SciELO
- Access statistics
Related links
- Cited by Google
- Similars in SciELO
- Similars in Google
Share
Revista EIA
Print version ISSN 1794-1237
Abstract
REINA, Jhovana and OLAYA, Javier. CURVE FITTING NONPARAMETRIC METHODS FOR STUDYING BEHAVIOR FROM AIR POLLUTION PM10. Rev.EIA.Esc.Ing.Antioq [online]. 2012, n.18, pp.19-31. ISSN 1794-1237.
One of the main air pollutants is the particulate matter whose aerodynamic diameter is less than 10 micrometers, usually referred as PM10. It is a fact that the PM10 behavior in the air varies in an irregular way, and also in a temporal way in the atmosphere, mainly due to human activities, to unstable atmospheric conditions, and to meteorological phenomena. Our main purpose is to characterize through a nonparametric smooth model the PM10 daily behavior, taking into account the day of the week, and the precipitation levels. We illustrate the model using records on PM10 contamination, as well as on data on rain precipitation in the north side of Cali, Colombia. We estimate daily typical curves of the PM10 behavior using kernel and spline estimators. We processed these data using the free distribution statistical software R. The estimated curves allow us to observe a PM10 unimodal behavior during the morning hours, which varies from one day to another and from rainy to non-rainy days. The fitted models allow a robust characterization of the PM10 daily behavior, considering heteroscedastic observations on a multiple response per design point scenario.
Keywords : air pollution; heteroscedasticity; PM10; nonparametric regression; kernel smoothing; spline smoothing.