* Corresponding Author: f.fedele@arpa.puglia.it ISSN: 2283-5954 © 2014 The Authors. Published by Digilabs Selection and peer-review under responsibility of DUST2014 Scientific Committee DOI:10.14644/dust.2014.026 Meteorological effects on PM10 concentrations in an urban industrial site: a statistical analysis Francesca Fedele 1,2* , Micaela Menegotto 1 , Livia Trizio 1 , Lorenzo Angiuli 1 , Anna Guarnieri Calò Carducci 1 , Roberto Bellotti 2,3 , Roberto Giua 1 , Giorgio Assennato 1 1 Apulia Region Environmental Protection Agency (ARPA Puglia), Corso Trieste 27, 70126 Bari, Italy 2 Università degli Studi di Bari, 70126 Bari, Italy 3 Istituto Nazionale di Fisica Nucleare Sezione di Bari, 70126 Bari, Italy Abstract This study deals with the analysis of 8 years time series of PM10 and meteorological data collected in the city of Taranto – Italy, which is characterized by the proximity to a large industrial area which includes the largest European integrated steel plant, an oil refinery and a cement plant. In particular we focus on a small neighbourhood called Tamburi characterized by several exceedances of regulatory limits with respect to PM10. This neighbourhood is located less than 1 Km away from the steel plant mineral stockyard, downwind at wind direction from North-West quadrant. The aim of the study is to identify specific wind conditions leading to deterioration of air quality with respect to PM10 concentrations. We chose two sampling sites of PM10 from ARPA Puglia Air Quality Monitoring Network, one located in Tamburi and the other in a Taranto neighbourhood called Talsano, similar to the first in population density and urban morphology but much farther from the mineral stockyard. Meteorological data are obtained from a station located in Taranto from the same Air Quality Monitoring Network. To identify the specific wind conditions, we looked for critical wind speed thresholds and wind speed permanence in terms of consecutive hours over the threshold. Combining thresholds and permanence we defined some meteorological criteria and applying them we divided into two classes every day of the observation period naming them “Wind Day” and “No Wind Day”. According to this classification we performed a statistical analysi s (Wilks, 2006) on the two PM10 data sets; we also performed a statistical hypothesis testing building two different distributions, one obtained from PM10 concentration and the other from the difference between the PM10 concentrations in two sites, and for each meteorological criterion we build a ROC curve. Days of Saharian advection, identified using a canonic method (EC, 2011), were eliminated from the dataset. The study shows that Talsano site exhibits a constant behaviour, that is a decrease of PM10 concentrations in the days classified as “Wind Day” with respect to the “No Wind Day” class, due to ProScience 1 (2014) 162-167 Available at www.scientevents.com/proscience/