A Multivariate Approach for the Analysis of Spatially Correlated Environmental Data

Journal of Environmental Informatics 5 (1) 9-16 (2005)

05JEI00041 1726-2135/1684-8799 © 2005 ISEIS www.iseis.org/jei doi:10.3808/jei.200500041

A Multivariate Approach for the Analysis of Spatially Correlated Environmental Data
A. Lamberti1* and E. Nissi2
2 1 ISTAT - Via C. Balbo, 16 - 00184 Roma, Italy Dipartimento di Metodi Quantitativi e Teoria Economica, Viale Pindaro, 42 - 65127 Pescara, Italy

ABSTRACT. The formulation and the evaluation of environmental policy depend upon a general class of latent variable models known as multivariate receptor models. Estimation of the number of major pollution sources, the source composition profiles and the source contributions are the main interests in multivariate receptor modelling. Many different approaches have been proposed both when the number of sources is unknown (explorative factorial analysis) and when the number and the type of sources are known (regression models). The objective of this work is to propose a flexible approach to the multivariate receptor models that incorporates the extra variability due to the spatial dependence. The method is applied to Lombardia air pollution data. Keywords: Covariance modelling, environmental data, latent variable models, multivariate receptor models, spatio-temporal modelling

1. Introduction
In the past few years interest in air quality monitoring has increased, specifically pertaining to the identification of pollution sources and their information needed to implement air pollution control programs. Since observing the quantity of various pollutants emitted from all potential pollution sources is virtually impossible, receptor models are used to analyze concentrations of pollutants or particles measured over time in order to gain insight concerning the unobserved pollution sources. Multivariate receptor modeling aims to identify the pollution sources and assess the amounts of pollution by resolving the measured mixture of chemical species into the

References: Anderson, T.W. (1984). An Introduction to Multivariate Statistical Analysis, 2nd Edition, John Wiley & Sons, New York, USA. Bartholomew, D.J. and Knot, M. (1999). Latent Variable Models and Factor Analysis, 2nd Edition, Oxford University Press, New York. Gleser, L.J. (1997). Some thoughts on chemical mass balance models. Chemom. Intell. Lab. Syst., 37, 15-22. Guttorp, P. and Sampson P.D. (1994). Methods for estimating heterogeneous spatial covariance functions with environmental applications, in G.P. Patil and C.R. Rao (Eds.), Handbook of Statistics XII: Environmental Statistics, Elsevier/North Holland, New York, pp. 663-690. Henry, R.C. (1987). Current factor analysis models are ill-posed. Atmos. Environ., 21, 1815-1820. Henry, R.C. (1997). History and fundamentals of multivariate air quality receptor models, Chemom. Intell. Lab. Syst., 37, 37-42. Henry, R.C., Park, E.S. and Spiegelman, C.H. (1999). Comparing a new algorithm with the classic methods for estimating the number of factors. Chemom. Intell. Lab. Syst., 48, 91-97. Henry, R.C. (2002). Multivariate receptor models: current practice and future trends. Chemom. Intell. Lab. Syst., 60, 43-48. Henry, R.C. (2003). Multivariate receptor modelling by N-dimensional edge detection. Chemom. Intell. Lab. Syst., 65, 179-189. 15 A. Lamberti and E. Nissi / Journal of Environmental Informatics 5 (1) 9 - 16 (2005) Hopke, P.K. (1991). An introduction to receptor modelling. Chemom. Intell. Lab. Syst., 10, 21-43. Hopke, P.K. (1997). Receptor modelling for air quality management, in R.E. Hester and R.M. Harrison (Eds.), Issues in Environmental Science, Issue 8, Royal Society of Chemistry, Cambridge UK, pp. 95-117. Hopke, P.K. (2003). Recent developments in receptor modelling. J. Chemom., 17, 255-265. Javitz, H.S., Watson, J.G., Guertin, J.P. and Mueller, P.K. (1988). Results of a receptor modelling feasibility study. J. Air Pollut. Control Assoc., 38, 661-667. Kim, E., Hopke, P.K., Paatero, P. and Edgerton, E.S. (2003). Incorporation of parametric factors into multilinear receptor model studies of Atlanta aerosol. Atmos. Environ., 37, 5009-5021. Loader, P.S. (1992). Spatial covariance estimation for monitoring data, in A. Walden and P. Guttorp (Eds.), Statistics in Environmental and Earth Sciences, Edward Arnold, London, pp. 52-70. Meiring, W., Sampson, P.D. and Guttorp, P. (1998). Space-time estimation of grid-cell hourly ozone levels for assessment of a deterministic model. Environ. Ecol. Stat., 5, 197-222. Nott, D.J., Dunsmuir, W.T.M., Speer, M.S. and Glowacki, T.J. (1998). Non-stationary Multivariate Covariance Estimation for Monitoring Data, Technical Report S98-14. Paatero, P. and Hopke, P.K. (2002). Utilizing wind direction and wind speed as independent variables in multilinear receptor modelling studies. Chemom. Intell. Lab. Syst., 60, 25-41. Paatero, P., Hopke, P.K., Hoppenstock, J. and Eberly, S.I. (2003). Advanced factor analysis of spatial distributions of PM2.5 in the eastern United States. Environ. Sci. Technol., 37, 2460-2476. Park, E.S., Henry, R.C. and Spiegelman, C.H. (1999). Determining the Number of Major Pollution Sources in Multivariate air Quality Receptor Models, NRCSE, TSR No.34. Park, E.S., Henry, R.C. and Spiegelman, C.H. (2000). Estimating the number of factors to include in a high-dimensional multivariate bilinear model. Commun. Stat., 29(B), 723-746. Park, E.S., Guttorp, P. and Henry, R.C. (2001). Multivariate receptor modelling for temporal correlated data by using MCMC. J. Am. Stat. Assoc., 96, 1171-1183. Park, E.S., Oh, M.S. and Guttorp, P. (2002). Multivariate receptor models and model uncertainty. Chemom. Intell. Lab. Syst., 60, 49-67. Sampson, P.D. and Guttorp, P. (1992). Nonparametric estimation of nonstationary spatial covariance structure. J. Am. Stat. Assoc., 87, 108-119. Spiegelman, C.H. and Dattner, S. (1993). Multivariate chemometrics, a case study: applying and developing receptor models for the 1990 El Paso winter PM10 receptor modelling scoping study, in G.P. Patil and C.R. Rao (Eds.), Multivariate Environmental Statistics, Elsevier Science publishers, New York, pp. 509-524. 16