Assessment of regional metal levels in ambient air by statistical regression models

Abstract

The assessment of the particulate matter (PM) levels and its constituents presented in the atmosphere is an important requirement of the air quality management and air pollution abatement. The heavy metal levels in PM10 are commonly evaluated by experimental measurements; nevertheless, the EC Directives also allow the Regional Governments to estimate the regulated metal levels (Pb in Directive 2008/50/EC and As, Ni and Cd in Directive 2004/107/EC) by objective estimation and modelling techniques. These techniques are proper alternatives to the experimental determination because the required analysis and/or the number of required sampling sites are reduced. The present work aims to estimate the annual levels of regulated heavy metals by means of multivariate linear regression (MLR) and principal component regression (PCR) at four sites in the Cantabria region (Northern Spain). Since the objective estimation techniques may only be applied when the regulated metal concentrations are below to the lower assessment threshold, a previous evaluation of the determined annual levels of heavy metals is conducted to test the fulfilment of the EC Directives requirements. At the four studied sites, the results show that the objective estimations are allowed alternatives to the experimental determination. The annual average metal concentrations are well estimated by the MLR technique in all the studied sites; furthermore, the EC quality requirements for the objective estimations are fulfilled by the developed statistical MLR models. Hence these estimations may be used by Regional Governments as a proper alternative to the experimental measurements for the regulated metal levels assessment.