The impact of crop residue burning (CRB) on the diurnal and seasonal variability of the ozone and PM levels at a semi-urban site in the north-western Indo-Gangetic plain

Journal of earth system science

Gufran Beig^2¹  Susheel K Mittal^1²  Madhvi Rana^1³ 

[1] Department of Computer Science and Engineering, Thapar University, Patiala 147 004, India.^3;Indian Institute of Tropical Meteorology, Pashan, Pune 411 008, India.^2;School of Chemistry and Biochemistry, Thapar University, Patiala 147 004, India.^1

关键词: Indo-Gangetic plain;    ozone;    PM$_{10}$ and PM$_{2.5}$;    meteorology;    correlation analysis;    linear and nonlinear regression analysis;   

DOI  :  

学科分类：天文学(综合)

来源: Indian Academy of Sciences

【 摘 要 】

Ozone and particulate matter (PM), PM$_{10}$ and PM$_{2.5}$, were monitored along with meteorological parameters at a semi-urban location, Patiala, in the north-western Indo-Gangetic plain from December 2013 to November 2014. The annual mean concentration levels of PM$_{10}$, PM$_{2.5}$ and ozone were recorded as 178 $\mu$g m$^{-3}$, 88 $\mu$gm$^{-3}$ and 39 ppb, which also exceeded the national standards on a 24-h average basis, by 77.3% (335 days), 53.2% (338 days) and 20 days, respectively. High levels of PM (PM$_{10}$ and PM$_{2.5}$) were observed in winter (58–381 and 42–270 $\mu$g m$^{-3}$) and in the post-monsoon (71–320 and 39–320 $\mu$g m$^{-3}$) season and a rise in the level of ozone was observed in summer (22–72 ppb) and in the post-monsoon season (23–73 ppb), respectively. The rate of ozone production was the highest during the post-harvest fire period (3.94 ppb O$_{3}$/h in May and 4.23 ppb O$_{3}$/h in November). A Pearson correlation study showed the strong dependency of PM and ozone on meteorological variables. Relative humidity has the highest ranking for ozone and PM$_{10}$, while wind speed has the lowest rank for ozone, PM$_{10}$ and PM$_{2.5}$ in the order of factors affecting the level of pollutants. The generalised linear model and the neural network model (for ozone) and the random forest model (for PM) outperformed on the basis of performance indices and further cross-validation was done.

【 授权许可】

CC BY