Fractional order Lorenz based physics informed SARFIMA-NARX model to monitor and mitigate megacities air pollution

Air Pollution is an emerging disaster and considered one of the biggest challenges of the world to effectively con-trol, mitigate and forecast due to abrupt variability, stochastic, and chaotic pattern of particulate matter (PM) in terms of time and space of the pollutants. Composition of ambient PM not only causes serious damage to public health but also emerging as a global hazard particularly for urban environment with negative impact on human health including morbidity. Mortality and ultimately towards unstable economy. In this study, hourly short-term trends of PM2.5 and air quality index (AQI) of Lahore city of Pakistan is monitored and mitigated by the design of fractional order Lorenz based physics informed hybrid computing paradigm SARFIMA-NARX for forecasting hourly pattern of next two days. The complex dynamics of earth system and its weather forecast are character-ized by combination of biological, physical, and chemical processes governed by the different laws of science that provides additional information for the climate variation in terms of physics inform intelligence. The perfor-mance index based on statistical indicator of RMSE confirmed the high accuracy and efficiency of designed model to predict the pattern. The early predictions based on computational intelligence paradigm may serve as a surveillance system to reduce the air pollution through cost-effectiveness planning by environmental monitor-ing agencies.(c) 2022 Elsevier Ltd. All rights reserved.

Automated summary

Air pollution is an emerging disaster that presents a significant challenge in terms of controlling, mitigating, and forecasting due to the complex variability of particulate matter. This study focuses on monitoring and predicting the short-term trends of PM2.5 and air quality index in Lahore, Pakistan using a hybrid computing paradigm. The results show high accuracy and efficiency in predicting the hourly pattern for the next two days. Early predictions based on computational intelligence can aid in reducing air pollution through cost-effective planning by environmental monitoring agencies.