Evaluation of air pollution control policies in Mexico City using finite Markov chain observation model

Abstract

This paper proposes a Markov observation based model, where the transition matrix is formulated using air quality monitoring data for specific pollutant emissions, with the primary objective to analyze the corresponding stationary distributions and evaluate sceneries for the air quality impact of pollution control policies. The model is non predictive and could be applied to every source of pollutant emissions included in air monitoring data. Two cases of study are presented, ozone and sulfur, over central zone of Mexico City for a seven years span from 2000 to 2006. For presentation purposes each year were divided in two semesters. In ozone case, the stationary distribution for both semesters shows a probability diminution of the higher ozone concentrate levels, with tendency to "piston effect". In the sulfur case, the first semester displays an oscillatory behavior with a little tendency to diminution of the higher sulfur concentrate levels, the second semester had decreasing probabilities of the higher sulfur levels. The results support an small improvement of air quality and then a favorable evaluation of the diverse pollution control policies that had been implemented in Mexico City over the last several years. Keywords ResumenEn este trabajo se propone un modelo de Markov basado en observaciones, donde la matriz de transición se formula empleando información del monitoreo de la calidad del aire para ciertos contaminantes específicos, con el objetivo principal de analizar las distribuciones estacionarias correspondientes y evaluar escenarios del impacto de las medidas de control anticontaminantes de la calidad del aire. El modelo no es predictivo y puede aplicarse a cualquier fuente de emisiones contaminantes incluidas en el sistema de monitoreo ambiental. Se presentan dos casos de estudio, ozono y azufre, sobre la zona central de la Ciudad de México para un intervalo de siete años del 2000 al 2006. Para propósitos de presentación, cada año se dividió en dos semestres. En el caso del ozono, la distribución estacionaria para ambos semestres mostró una disminución en la probabilidad de ocurrencia de los niveles de alta concentración de ozono, con cierta tendencia al "efecto pistón". En el caso del azufre, el primer semestre mostró un comportamiento oscilatorio con una pequeña tendencia a la disminución de los niveles de alta concentración y en el segundo semestre disminuyeron las probabilidades de altos niveles de azufre. Los resultados establecen una ligera mejoría en la calidad del aire y por lo tanto una evaluación favorable de las diversas medidas de control de contaminantes que han sido implementadas en la Ciudad de México en losúltimos años.Palabras clave: Cadenas de Markov; distribución estacionaria; contaminación del aire; emisiones de azufre; emisiones de ozono; Ciudad de México.Mathematics Subject Classification: 60J20, 91B76.

Prediction of Nevşehir Province Air Quality with Markov Chain

“…Zamandan bağımsız bir stokastik süreç olan Markov zinciri [28], kategorik verileri analiz etmek için birçok alanda yaygın olarak kullanılmaktadır [11] Markov zincirleri, çevresel izleme verilerinin gerçek davranışını analiz etmek, rüzgar gücü dağılımını tahmin etmek [29], çevresel izleme verilerinin zaman serisi verilerini tahmin etmek [30], hava kirleticilerinin konsantrasyonunu tahmin etmek [31] ve hava kirliliği kontrol politikalarını değerlendirmek [32] için kapsamlı bir şekilde uygulanmıştır.…”

Comparing the Number of Ozone Exceedances in Different Seasons of the Year in Mexico City

“…We also have from [48] that even though ozone peaks have been reduced significantly, maximum concentrations still persist. Hoyos et al [22] presents an analysis corroborating the decrease in ozone levels, but now using Markovian models. In [13] a study is performed to analyse the car use restriction in Mexico City.…”

Bayesian Information Criterion for Fitting the Optimum Order of Markov Chain Models: Methodology and Application to Air Pollution Data

“…They worked with the high-order chains that can be estimated by examining correlation plots, and transition probabilities have been estimated using the maximum likelihood method. Hoyos et al [15] employed the MC model for modeling O 3 and SO 2 for evaluating the effect of air pollution events in Mexico City. Rodrigues and Achcar [16] have proposed a discrete-time MC model for ozone air pollution based on the maximum daily observations.…”