核心概念
Machine learning models can effectively predict air quality index (AQI) by analyzing air pollutant concentrations, and the influence of Punjab's stubble burning on AQI variability across different cities and states in India is significant.
摘要
This research focuses on predicting the Air Quality Index (AQI) across diverse cities and states in India, including Delhi, Haryana, and Punjab, using various machine learning models. The study investigates the influence of stubble burning in Punjab on the variability of AQI in the region.
The researchers used a dataset from the Central Pollution Control Board (CPCB) website, which includes air pollutant concentrations from 22 different monitoring stations in the three states. The dataset was preprocessed to handle missing values and outliers, and feature engineering techniques were applied.
Several machine learning models were employed for AQI prediction, including CatBoost, XGBoost, Random Forest, and Support Vector Regression (SVR). Additionally, a time series model, SARIMAX, and a deep learning model, LSTM, were also used. The performance of these models was evaluated using metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and R-squared (R²).
The results showed that the Random Forest Regressor model outperformed the other models, with an R² value close to 1, indicating its ability to accurately predict AQI. The study also found that the air pollutants PM2.5 and PM10 have the highest positive correlation with AQI, while meteorological factors like temperature, solar radiation, and relative humidity are negatively correlated.
The research also explored the spatial and temporal patterns of AQI across the three states. The heatmap analysis revealed that AQI was at its peak during the months of November and December, even during the COVID-19 pandemic year. The study highlighted the significant impact of stubble burning in Punjab on the air quality of the neighboring states, particularly Delhi.
統計資料
PM2.5 (ug/m3) and PM10 (ug/m3) have the highest positive correlation with AQI.
The burning of 63 Mt of crop stubble releases 3.4 Mt of CO, 0.1 Mt of NOx, 91 Mt of CO2, 0.6 Mt of CH4, and 1.2 Mt of PM into the atmosphere.
About 84 Mt (23.86%) of the stubble is burnt on-field each year in India.
引述
"The intensive rice-wheat rotation system which generates a large amount of stubble is the main reason for bad air quality in India. Each year about 352 Mt of stubble is generated in India straight after the harvesting season."
"During the rice stubble burning season the impact of stubble burning is more severe as the lower winter temperature leads to a more stable atmosphere (Inversion conditions). There is a fact that pollutants stay longer in the atmosphere during this time and that the amount of rice stubble burned is quite higher than that of wheat."