Prediction of maximum of 24-h average of PM10 concentrations 30h in advance in Santiago, Chile

Pérez, Patricio; Reyes, Jorge Isaías Caicedo

doi:10.1016/s1352-2310(02)00419-3

Cited by 105 publications

(70 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In Perez and Reyes (2002), the exceedances of the 240 g/m 3 limit are detected about 75% of the times by FFNNs. One should be however aware that just 20 exceedances of such a limit are observed every year.…”

Section: Comparison With Results Available In Literaturementioning

confidence: 99%

“…In Zickus et al (2002), four different machine learning approaches are trained to predict whether the PM 10 daily average will exceed the limit of the (Kukkonen et al, 2003) and Santiago, Chile (Perez and Reyes, 2002), respectively.…”

Section: Comparison With Results Available In Literaturementioning

confidence: 99%

“…Remarkably, while we adopt -as already elucidated -the PM 10 daily average as forecast target, the study of Kukkonen et al (2003) tries to forecast the PM 10 hourly average, and that of Perez and Reyes (2002) the exceedance of the 240 g/m 3 limit on PM 10 daily maximum. Such different forecast targets do not allow a direct comparison of the findings.…”

Section: Comparison With Results Available In Literaturementioning

confidence: 99%

“…An attempt in predicting the time series of PM2.5 (Perez et al, 2000) and PM 10 (Perez and Reyes, 2002) in Santiago, Chile showed better performances of neural networks with respect to linear regressors and persistent predictors. 1 The EU-funded project APPE-TISE (http://www.uea.ac.uk/env/appetise) carried out an inter-comparison of different air pollution forecasting approaches; the application to PM 10 prediction in Helsinki (Kukkonen et al, 2003) showed a better forecast accuracy for neural networks than for other approaches, such as linear regressors and also deterministic physically-based models.…”

Section: Introductionmentioning

confidence: 98%

See 3 more Smart Citations

Air quality prediction in Milan: feed-forward neural networks, pruned neural networks and lazy learning

Corani

2005

Ecological Modelling

222

View full text Add to dashboard Cite

Ozone and PM 10 constitute the major concern for air quality of Milan. This paper addresses the problem of the prediction of such two pollutants, using to this end several statistical approaches. In particular, feed-forward neural networks (FFNNs), currently recognized as state-of-the-art approach for statistical prediction of air quality, are compared with two alternative approaches derived from machine learning: pruned neural networks (PNNs) and lazy learning (LL). PNNs constitute a parameterparsimonious approach, based on the removal of redundant parameters from fully connected neural networks; LL, on the other hand, is a local linear prediction algorithm, which performs a local learning procedure each time a prediction is required. All the three approaches are tested in the prediction of ozone and PM 10 ; predictors are trained to return at 9 a.m. the concentration estimated for the current day.No strong differences are found between the forecast accuracies of the different models; nevertheless, LL provides the best performances on indicators related to average goodness of the prediction (correlation, mean absolute error, etc.), while PNNs are superior to the other approaches in detecting of the exceedances of alarm and attention thresholds. In some cases, datadeseasonalization is found to improve the prediction accuracy of the models.Finally, some striking features of lazy learning deserve consideration: the LL predictor can be quickly designed, and, thanks to the simplicity of the local linear regressors, it both gets rid of overfitting problems and can be readily interpreted; moreover, it can be also easily kept up-to-date.

show abstract

Section: Comparison With Results Available In Literaturementioning

confidence: 99%

Section: Comparison With Results Available In Literaturementioning

confidence: 99%

Section: Comparison With Results Available In Literaturementioning

confidence: 99%

Section: Introductionmentioning

confidence: 98%

See 2 more Smart Citations

Air quality prediction in Milan: feed-forward neural networks, pruned neural networks and lazy learning

Corani

2005

Ecological Modelling

222

View full text Add to dashboard Cite

show abstract

“…Other typical approaches for PM prediction are neuronal networks (cf. Pe´rez and Reyes, 2002;Hooyberghs et al, 2005), discriminant analysis (cf. Silva et al, 2001) or Kalman filtering (cf.…”

Section: Introductionmentioning

confidence: 99%

Quality and performance of a PM10 daily forecasting model☆

Stadlober¹,

Hörmann²,

Pfeiler³

2008

Atmospheric Environment

108

View full text Add to dashboard Cite

Air quality forecasting for winter‐time PM_2.5 episodes occurring in multiple cities in central and southern Chile

Saide

Mena‐Carrasco²,

Tolvett

et al. 2016

JGR Atmospheres

View full text Add to dashboard Cite

Episodic air quality degradation due to particles occurs in multiple cities in central and southernChile during the austral winter reaching levels up to 300-800 μg/m 3 hourly PM 2.5 , which can be associated with severe effects on human health. An air quality prediction system is developed to predict such events in near real time up to 3 days in advance for nine cities with regular air quality monitoring: Santiago, Rancagua, Curicó, Talca, Chillan, Los Ángeles, Temuco, Valdivia, and Osorno. The system uses the Weather Research and Forecasting with Chemistry model configured with a nested 2 km grid-spacing domain to predict weather and inert tracers. The tracers are converted to hourly PM 2.5 concentrations using an observationally based calibration which is substantially less computationally intensive than a full chemistry model. The conversion takes into account processes occurring in these cities, including higher likelihood of episode occurrence during weekends and during colder days, the latter related to increased wood-burning-stove activity for heating. The system is calibrated and evaluated for April-August 2014 where it has an overall skill of 53-72% of episodes accurately forecasted (61-76% for the best initialization) which is better than persistence for most stations. Forecasts one, two, and three days in advance all have skill in forecasting events but often present large variability within them due to different meteorological initializations. The system is being implemented in Chile to assist authority decisions not only to warn the population but also to take contingency-based emission restrictions to try to avoid severe pollution events.

show abstract

Prediction of maximum of 24-h average of PM10 concentrations 30h in advance in Santiago, Chile

Cited by 105 publications

References 10 publications

Air quality prediction in Milan: feed-forward neural networks, pruned neural networks and lazy learning

Air quality prediction in Milan: feed-forward neural networks, pruned neural networks and lazy learning

Quality and performance of a PM10 daily forecasting model☆

Air quality forecasting for winter‐time PM_2.5 episodes occurring in multiple cities in central and southern Chile

Contact Info

Product

Resources

About

Prediction of maximum of 24-h average of PM10 concentrations 30h in advance in Santiago, Chile

Cited by 105 publications

References 10 publications

Air quality prediction in Milan: feed-forward neural networks, pruned neural networks and lazy learning

Air quality prediction in Milan: feed-forward neural networks, pruned neural networks and lazy learning

Quality and performance of a PM10 daily forecasting model☆

Air quality forecasting for winter‐time PM2.5 episodes occurring in multiple cities in central and southern Chile

Contact Info

Product

Resources

About

Air quality forecasting for winter‐time PM_2.5 episodes occurring in multiple cities in central and southern Chile