Abstract
Amidst the rapid expansion of the economy, a discernible decline in air quality has become increasingly evident. This surge in air pollution raises significant concerns regarding potential health hazards for the public. The ability to anticipate air quality conditions would offer substantial advantages in implementing stringent measures to mitigate public exposure and curtail emissions from specific sources. While current climate models offer insights into air quality forecasts, they impose considerable computational demands, especially when aiming for enhanced spatial precision. Furthermore, these model-derived results often exhibit significant biases attributed to various factors, including uncertainties in emissions and the underlying physics/chemistry. Conversely, the integration of machine learning (ML) techniques within climate science has witnessed remarkable growth in recent years. ML models can play a pivotal role in monitoring and predicting air quality by leveraging historical data, sensor readings, model data, and other relevant factors to generate accurate forecasts. In this chapter, we show a case study of the United Kingdom to highlight the importance of the ML model in improving the concentration of ammonia (NH3) as well as the computing time. Intercomparison analysis is carried out between WRF-Chem and the ML model in simulating spatial distribution of NH3 during the month of April 2016. Our results indicate the ML model significantly reduces the computing time by half compared to WRF-Chem. Spatial patterns indicated that ML captures NH3 hotspots over the Midlands compared to WRF-Chem where maximum emission occurs due to intense agricultural activity. However, there are regions (e.g., the west and south parts of UK) that are underestimated by the ML model implying the need of a greater number of filters to overcome the underestimation. The results of this test case study encourage the possible usage of ML models with proper realizations to enhance the prediction of air quality. As a consequence this will help in framing climate policies.
Original language | English |
---|---|
Title of host publication | Artificial Intelligence for Air Quality Monitoring and Prediction |
Editors | Amit Awasthi, Kanhu Charan Pattnayak, Gaurav Dhiman, Pushp Raj Tiwari |
Place of Publication | Boca Raton |
Publisher | CRC Press |
Chapter | 3 |
Pages | 39-47 |
Number of pages | 9 |
Edition | 1 |
ISBN (Electronic) | 9781040131183, 9781032683805 |
ISBN (Print) | 9781032683799 |
DOIs | |
Publication status | Published - 1 Jan 2024 |