Air Pollution in India

The KaruSphère laboratory contributed its expertise to the article entitled ” Prediction of PM2.5 Concentration of four Indian cities using a Novel Hybrid Decomposition Approach “. The study proposes a hybrid approach combining Multivariate Empirical Mode Decomposition (MEMD) with Machine-Learning (ML) algorithms to predict daily PM2.5 concentrations in four major Indian cities: Thiruvananthapuram, New Delhi, Mumbai, and Visakhapatnam. The models integrate meteorological and air-quality parameters. The results show that MEMD-ML hybrid approaches generally outperform standalone machine-learning models, with MEMD-KNN performing particularly well across several cities and input configurations. These findings highlight the potential of hybrid methods to better capture the nonlinear and nonstationary dynamics of atmospheric pollution and improve air-quality prediction.