Author(s)

P. Perez, R. Palacios & A. Castillo

Abstract

Prediction of 8 hour average of carbon monoxide concentrations, in Santiago, Chile P. Perez a, R. Palacios b & A. Castillo b a Departamento de Fisica b Departamento de Ingenieria Geograjka, Universidad de Santiago de Chile. Abstract Carbon Monoxide (CO) is considered a harmful pollutant emitted mainly by motor vehicles. In large cities like Santiago, CO concentrations may reach values that exceed the norm that health organizations have established as a safe limit to which the population may be exposed (10 mg/m3, 8 hour moving average). It is important to be able to predict with at least 30 hours in advance, when concentrations will exceed this limit, because restriction to vehicle circulation may imply a relevant decrease with respect to expected values when no actions are taken. We show here a study on the possibility to predict maximum values of 8 hour moving average of CO concentrations using past values of CO concentrations and meteorological forecasts as input to multi linear regressions and neural network models. The neural network model seems to leave more room to adjust free parameters with one year data in order to predict the following year values. We have worked with data of three years measured in two of the monitoring stations located in the urban area of Santiago. Introduction Topographic and meteorological conditions and the activity of a large population in Santiago, Chile make this city is one of the most polluted in the world. At present, health authorities have defined air quality by the value of the

Keywords