Prediction of daily PM10 concentration using machine learning

2020 ◽  
Author(s):  
Oumaima BOUAKLINE ◽  
Khadija ARJDAL ◽  
Kenza KHOMSI ◽  
Noureddine SEMANE ◽  
Abdelhak ELIDRISSI ◽  
...  
Keyword(s):  
Machine Learning ◽  
Pm10 Concentration ◽  
Daily Pm10

Machine Learning-based Prediction Maps for the Tomorrow’s PM10 Concentration in Korea

2020 ◽  
Vol 15 (4) ◽  
pp. 269-285
Author(s):  
Yemin Jeong ◽  
◽  
Subin Cho ◽  
Youjeong Youn ◽  
Seoyeon Kim ◽  
...  
Keyword(s):  
Machine Learning ◽  
Pm10 Concentration

scholarly journals Monitoring the contribution of desert dust intrusion to PM10 concentration in Northern Cyprus

2017 ◽  
Vol 27 (2) ◽  
Author(s):  
HY Sulaiman ◽  
S Çakir
Keyword(s):  
Pm10 Concentration ◽  
Hysplit Model ◽  
Desert Dust ◽  
Regional Background ◽  
Northern Cyprus ◽  
Local Soil ◽  
Dust Events ◽  
Standard Limit ◽  
Urban Sites ◽  
Daily Pm10

Air quality in the Mediterranean basin has been affected by PM10 pollution induced by transported desert dust and local emission. The study used PM10 data from Nicosia, Kyrenia, Guzelyurt and Famagusta urban representatives, Kalecik rural background and Alevkayasi regional background. HYSPLIT model and satellite data were used to identify dust days and dust input was quantified using the method suggested by the European Commission. Anthropogenic background contribution of each site was then estimated by subtracting the regional background concentrations. A total of 35 dust days occurred on Cyprus island within the 3-years period; mostly during winter and spring. Daily PM10 concentration on dust days can reach up to 400 μg/m3. After removing dust background, annual PM10 concentrations were 48-58 μg/m3 in Nicosia, 42-47 μg/m3 in Famagusta, 40-50 μg/m3 in Kyrenia, 33-41 μg/m3 in Guzelyurt, 21-28 μg/ m3 in Alevkayasi, and 32-38 μg/m3 in Kalecik. PM10 concentrations were higher during winters in the urban sites. Despite the high frequency of dust events, only a fraction of exceedances of the standard limit in the urban sites were attributable to dust. Anthropogenic background sources contributions were 12.3 μg/m3 in Guzelyurt, 18 μg/m3 in Kyrenia, 18.4 μg/m3 in Famagusta, 27.8 μg/m3 in Nicosia and 9.7 μg/m3 in Kalecik. Effects of other natural sources that the study did not assess, such as sea salt and local soil resuspension, could be the reason for exceedances.

Modeling of Daily PM10 Concentration Occurrence Using Markov Chain Model in Shah Alam, Malaysia

2017 ◽  
Vol 10 (2) ◽  
pp. 96-106
Author(s):  
Norsalwani Mohamad ◽  
Sayang Mohd Deni ◽  
Ahmad Zia Ul-Saufie Japeri
Keyword(s):  
Markov Chain ◽  
Markov Chain Model ◽  
Pm10 Concentration ◽  
Chain Model ◽  
Daily Pm10

scholarly journals Evaluation of Boosted Regression Tree for the Prediction of the Maximum 24-Hour Concentration of Particulate Matter

2021 ◽  
Vol 12 (4) ◽  
pp. 126-130
Author(s):  
Wan Nur Shaziayani ◽  
◽  
Ahmad Zia Ul-Saufie ◽  
Syarifah Adilah Mohamed Yusoff ◽  
Hasfazilah Ahmat ◽  
...  
Keyword(s):  
Air Quality ◽  
Regression Tree ◽  
Pm10 Concentration ◽  
Boosted Regression Trees ◽  
Concentration Data ◽  
Boosted Regression Tree ◽  
Independent Test ◽  
Multiple Regression Models ◽  
Daily Pm10 ◽  
Fold Cross Validation

Air pollution is a considerable health danger to the environment. The objective of this study was to assess the characteristics of air quality and predict PM10 concentrations using boosted regression trees (BRTs). The maximum daily PM10 concentration data from 2002 to 2016 were obtained from the air quality monitoring station in Kuching, Sarawak. Eighty percent of the monitoring records were used for the training and twenty percent for the validation of the models. The best iteration of the BRT model was performed by optimizing the prediction performance, while the BRT algorithm model was constructed from multiple regression models. The two main parameters that were used were the learning rate (lr) and tree complexity (tc), which were fixed at 0.01 and 5, respectively. Meanwhile, the number of trees (nt) was determined by using an independent test set (test), a 5-fold cross validation (CV) and out-of-bag (OOB) estimation. The algorithm model for the BRT produced by using the CV was the best guide to be used compared with the OOB to test the predicted PM10 concentration. The performance indicators showed that the model was adequate for the next day’s prediction (PA=0.638, R2=0.427, IA=0.749, NAE=0.267, and RMSE=28.455).

scholarly journals Mass concentrations and indoor-outdoor relationships of PM in selected educational buildings in Nis, Serbia

2015 ◽  
Vol 21 (1-2) ◽  
pp. 149-158 ◽  
Author(s):  
Renata Kovacevic ◽  
Visa Tasic ◽  
Marija Zivkovic ◽  
Nenad Zivkovic ◽  
Amelija Djordjevic ◽  
...  
Keyword(s):  
Occupational Safety ◽  
Ambient Air ◽  
Pm10 Concentration ◽  
Strong Correlations ◽  
Outdoor Air ◽  
Limit Value ◽  
Guideline Value ◽  
Educational Buildings ◽  
Daily Pm10 ◽  
Mass Concentrations

Mass concentrations of particulate matter (PM) fractions were measured in educational buildings in the city of Nis, Republic of Serbia. Two sampling campaigns were conducted in winter periods. The first campaign was in the period from 21 February to 15 April 2010 at the Faculty of Occupational Safety (FOS) and the second campaign was from 20 March to 4 April 2013 at the primary school Vozd Karadjordje (VK). PM measurements were carried out with low volume samplers Sven/Leckel LVS3. The average daily PM10 concentration inside the FOS (47.0 ?21.8 ?g/m3) was lower than PM10 concentration in outdoor air (50.7 ?28.1 ?g/m3). The average daily PM10 concentration inside the VK (54.6 ? 17.6 ?g/m3) was higher than in outdoor air (47.9 ? 22.8 ?g/m3). The 24 hours average PM10 concentrations at FOS exceeded the EU limit value (50 mg/m3) during 34 % of days outdoors, and 39 % of days indoors. The 24 hours average PM10 concentrations at VK exceeded the limit value during 35 % of days outdoors, and 53 % of days indoors. The 24 hours average PM2.5 concentrations at VK exceeded the WHO daily mean guideline value (25 mg/m3) during 71 % of days outdoors, and 88 % of days indoors. The average PM10 I/O ratio at VK was 1.57 during teaching hours, and 1.00 during no teaching hours. Similarly, average PM2.5 I/O ratio at VK was 1.11 during teaching hours and 0.90 during no teaching hours. Average daily PM2.5/PM10 ratio in the ambient air at VK was 0.87, and 0.82 at FOS. Very strong correlations between the indoor and outdoor PM concentrations were observed at VK during no teaching hours (r>0.8). Moderate to strong negative correlations were found between the wind speed and PM at both schools. High outdoor PM concentrations and resuspension of particles are probably the most possible reasons for the elevated indoor PM concentrations found in the study.

scholarly journals Machine Learning by Data Mining REPTree and M5P for Predicating Novel Information for PM10

2020 ◽  
pp. 40-48
Author(s):  
Yas Alsultanny
Keyword(s):  
Machine Learning ◽  
Data Mining ◽  
Decision Tree ◽  
Meteorological Parameters ◽  
Pm10 Concentration ◽  
Divide And Conquer ◽  
Time Processing ◽  
Elapsed Time ◽  
Tree Algorithms

We examined data mining as a technique to extract knowledge from database to predicate PM10 concentration related to meteorological parameters. The purpose of this paper is to compare between the two types of machine learning by data mining decision tree algorithms Reduced Error Pruning Tree (REPTree) and divide and conquer M5P to predicate Particular Matter 10 (PM10) concentration depending on meteorological parameters. The results of the analysis showed M5P tree gave higher correlation compared with REPTree, moreover lower errors, and higher number of rules, the elapsed time for processing REPTree is less than the time processing of M5P. Both of these trees proved that humidity absorbed PM10. The paper recommends REPTree and M5P for predicting PM10 and other pollution gases.

Sign in / Sign up

Export Citation Format

Share Document