Prediction of PM2.5 via precursor method using meteorological parameters
Air pollution is one of the major environmental concerns faced by many countries including Pakistan. Being the major component of the pollution, particulate matters 2.5µm diameter (PM2.5), are known to highly raise health risks to people in the country. The present study investigates the modelling a...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
University of Bologna
2025-02-01
|
| Series: | EQA |
| Subjects: | |
| Online Access: | https://eqa.unibo.it/article/view/21009 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Air pollution is one of the major environmental concerns faced by many countries including Pakistan. Being the major component of the pollution, particulate matters 2.5µm diameter (PM2.5), are known to highly raise health risks to people in the country. The present study investigates the modelling and prediction of particulate matter 2.5µm using its precursor values and meteorological parameters; temperature, humidity levels, and wind speeds in Lahore and Karachi. The air quality of Lahore repeatedly plummets to hazardous levels in the winter season which is a severe threat to the public health and environment. An Artificial Neural Network (ANN) architecture was designed to predict PM2.5 by employing meteorological parameters (temperature, relative humidity & wind speed) and precursor values of PM2.5. The model consists of an input layer with four input variables, a hidden layer with 10 neurons, and an output layer consisting of PM2.5. The model was used for both the cities of Lahore and Karachi. The Root Mean Square Error (RMSE) value for Karachi was less than 18 and for Lahore, it was 39. The prediction of PM2.5 via ANN was good for Lahore and Karachi. However, the results of the modeling are better for Karachi. The accuracy of results were further verified by Mean Absolute Percentage Error (MAPE), Mean Absolute Bias Error (MABE), and Chi square statistics (Chi). |
|---|---|
| ISSN: | 2039-9898 2281-4485 |