• Annette John
• Priyansh Jain
• Sharon Raju
• Vinit Shah

Air pollution is a critical global environmental hazard, which is estimated to cause seven million deaths across the globe annually. Earlier studies have only focused on PM2.5 pollutants and have used data from costly sensors having sparse spatial resolutions. The proposed method tackles this with an interdisciplinary approach using Earth and Health sciences that integrates different datasets to study the causes and effects of air pollution by using freely available satellite data that have finer spatial resolutions; and also considers gases like NO2, CO, etc. The primary objective is to find the probable sources of air pollution in a cost-effective manner, and produce future projections based on the temporal data. We achieve this by deploying Artificial Intelligence techniques such as DBSCAN, K-Means, Time Series Analysis, and LSTM - on public datasets from the ESA and WHO. We performed an analysis of the Nationwide Lockdown on AQI levels. Our model acquired an accuracy of 88% on the training set and 90.7% on the validation set, respectively.

Deep learning, Artificial Intelligence, Kaggle Dataset, LSTM, KMeans, DBSCAN, Geospatial, WHO, ESA, Crowdsourcing, Satellite data, Sentinel 5P, AQI

"Detection of Air Pollution Hotspots and its Sources using Deep Learning", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.9, Issue 8, page no.f259-f268, August-2022, Available :http://www.jetir.org/papers/JETIR2208525.pdf

"Detection of Air Pollution Hotspots and its Sources using Deep Learning", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.9, Issue 8, page no. ppf259-f268, August-2022, Available at : http://www.jetir.org/papers/JETIR2208525.pdf