A Two-Sample Test for Functional Data Applied to Fine Particulate Matter Measurements on Air (303200)

Keywords: PM2.5, Air pollution, Smoothing, Data Analysis

We made a comparison between the contamination levels due to fine particulate matter on two monitoring stations of air quality in Cali, Colombia. Environmental authorities have defined fine particulate matter as particles suspended in air whose aerodynamic diameter is less than 2.5 micrometers (µm), usually denoted as PM2.5. Measurements of PM2.5 are taken every ten seconds, but just the hourly averages are reported. It means that we have available as much as 24 observations per day. So, although this variable is continuous in nature, we only have a finite number of observations during one single day. This is exactly the right setting on which Functional Data Analysis plays a key role on recent developments in statistical studies. We have information corresponding to one year calendar, but we have kept only days with complete information due to missing data. That means, we kept only those days having information on each of the 24 hours of the day. We also analyzed separately workdays and weekend days. We constructed functional data, one per day, and compared the functional means to find out that differences are not statistically significant between the two stations.