Tropospheric Ozone Prediction in Mexico City
DOI:
https://doi.org/10.29356/jmcs.v49i1.1313Keywords:
Ozone measurements, Mexico City, autoregresive method, embedded space methodAbstract
Abstract. Two techniques are applied to forecast time series for the hourly ozone measured at Pedregal’s recording station of the Automatic Network for Environmental Monitoring (RAMA for itsacronym in Spanish) located in the metropolitan area of Mexico City. The techniques have been widely applied since last century: the autoregressive (AR) and the method of delays in an embedded space. The predicted values by the autoregressive method are somewhat less precise than those forecasted by the embedded space method, as presented below. It is intended to predict the maximum ozone daily concentration in advance to be able to alert the citizenship or for taking appropriate control measures. In this presentation, the models have as main limitation to be based only on the ozone time series; more robust models should take into consideration meteorological variables to increase precision. If it is roughly considered only the series formed by the hourly ozone series from January to May 1999; the series of daily ozone maxima have an standard deviation of around 0.05 ppm of ozone. In the most precise —the embedded space method— shown at the end of the article, the error standard deviation between predicted and real maximum daily ozone data is around
0.027 ppm of ozone, which shortens the gap, considering the total of the ozone maximum as a normal distribution.
Resumen. Se aplican dos técnicas para el pronóstico de series en el tiempo para el caso de la serie formada por las lecturas de ozono horario en la estación de Pedregal de la Red Automática de Monitoreo Ambiental (RAMA) en el Valle de México. Estas técnicas
han sido usadas desde el siglo pasado; el modelo autorregresivo (AR) y el método para series caóticas. Los valores pronosticados por el modelo caótico son más precisos que el modelo autorregresivo. Se pretende pronosticar los valores máximos de concentración de ozono durante el día. Sin embargo, se requiere estudiar las variantes que tienen estos modelos para tener mayor precisión en el pronóstico, de manera que sea de utilidad práctica para tomar medidas preventivas pertinentes. En esta exposición los algoritmos tienen como principal carencia usar sólo el contaminante ozono en la serie del tiempo, dado que modelos más complejos deberán considerar variables meteorológicas que mejoren la precisión del pronóstico. Así a grosso modo, si se considera la serie formada por los máximos diarios de ozono en el período estudiado (enero a mayo de 1999) la desviación estándar es de 0.05 ppm de ozono en la variante más precisa del modelo caótico que se muestra al final del artículo, la desviación estándar del error entre los valores máximos pronosticados y los valores reales es de alrededor de 0.027 ppm de ozono, lo que acorta la banda de valores pronosticados a algo más de la mitad, considerando el total de los máximos diarios como una distribución normal.
Downloads
References
2. INE, Almanaque de datos y tendencias de la calidad del aire en ciudades mexicanas, 2004, México.
3. Pao-Wen Grace Liu, Forecasting Peak Daily Ozone Levels I. A Regression with Time Series Errors Model Having a Principal Component Trigger to fit 1991 Ozone Levels, J. Air & Waste Manag. Assoc., 2002, 52, 1064-1074.
4. Brown, M. J. Mexico City Ozone Concentration as a Function of Readily Available Parameters, February 1994, Los Alamos National Laboratory, Report 87545, Los Alamos,
5. Comrie, A. C., Comparing Neural Networks and Regression Models for Ozone Forecasting, J. Air & Waste Manag. Assoc., 1997, 47, 653.
6. Mattheij, R.M.M.; Molenaar, J., Ordinary Differential Equations in Theory and Practice, 1996, J. Wiley and Sons, New York.
7. Press, W.H.; Flannery, B.P.; Teukolsky, S.A.; Vetterling, W.T., “Numerical Recipes: The Art of Scientific Computing (FORTRAN Version)” 1989, Cambridge University Press, Cambridge.
8. Wolfram Research Inc., “Mathematica”, 2004.
9. Shaw, W.T.; Tigg, J. Applied Mathematica: Getting Started, Getting it done, 1994, Addison Wesley Pub. Co., New York.
10. Farmer, J.D.; Doyne, J.; Sidorowich, J.J., Phys. Rev. Letl. 1987, 59, 845.
11. Jian-Long Ch.; Islam, S.; Biswas, P. Atm. Envir. 1998, 32, 1839-1848.
Downloads
Published
Issue
Section
License
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.