Trend Analysis of Nitrogen dioxide (NO2) in Northern Part of India During Paddy Residue Burning Using a Contextual Approach

Mann–Kendall (MK) test is a non-parametric technique widely used for trend analysis in time series datasets. However, the datasets tend to be noisy which increases data variance and often results in false rejection of null hypothesis. The present study investigates use of spatial autocorrelation (i.e., contextual information) to address influence of noise in the MK test. By incorporating spatial autocorrelation, the false trend can be identified, while at the same time spatial autocorrelation provides support for strengthening the results. The contextual MK test (CMK) was used for analysing NO2 trend in the Northern Indian states of Punjab, Haryana, Delhi, Uttar Pradesh, Madhya Pradesh, and Rajasthan during paddy stubble burning, using TROPOspheric Monitoring Instrument total vertical column density data. The serial correlation in the datasets was removed using pre-whitening before running the CMK and conventional MK test. In year 2021, MK test identified 12.9% of the grid cells with monotonous increasing trend of NO2, which increased to 14.1% when CMK test was used. Similarly, cells with monotonous increasing NO2 trend in year 2020, were 8.7% and 9.5% using MK and CMK tests respectively. Thus, CMK test was able to identify more cells having a monotonous increasing trend of NO2 compared to the MK test, while at the same time the spurious trend could also be efficiently handled. Subsequently, using CMK test state-wise analysis of NO2 trend was also carried out.