Spatiotemporal analysis of tropospheric NO2 in Kathmandu Valley using TROPOMI satellite data

Abstract

Kathmandu Valley faces severe air quality degradation driven by rapid urbanization, geographic constraints, and seasonal meteorological patterns. This study presents a systematic spatiotemporal characterization of tropospheric nitrogen dioxide (NO₂) across Kathmandu Valley, Nepal, using Sentinel-5 Precursor TROPOspheric Monitoring Instrument (TROPOMI) satellite observations resampled to a 0.01° output grid spanning July 2018 to December 2024 (n = 78 monthly composites). Non-parametric trend detection methods — the Theil-Sen slope estimator and the Mann-Kendall test (with seasonal and pre-whitened variants as sensitivity checks) were applied to quantify temporal trends, seasonal decomposition was used to characterize intra-annual variability, and spatial analysis was conducted across five analytical zones to assess intra-valley concentration differences. The analysis identifies a marginally increasing trend of +5.3 × 10^-7 mol m^-2 year^-1 across the full study period; however, this trend is not statistically significant at a = 0.05 (p ≥ 0.05; seasonal-MK p = 0.41, pre-whitened-MK p = 0.36). COVID-19 lockdown periods produced a modest reduction of approximately 4.5% with a 95% bootstrap CI of (–1.4, +9.9)%, indicating the reduction is not statistically significant at a = 0.05 in NO₂ concentrations relative to equivalent months in non-lockdown years. Pronounced seasonal variability was observed, with pre-monsoon (March–May) concentrations approximately 14% higher than monsoon (June–September) levels, a pattern coincident with regional biomass burning due to forest fire and local brick kiln emissions but not by itself causally attributable to any single source. Significant intra-valley spatial heterogeneity was detected (Kruskal-Wallis H = 23.4, p<0.001), with Kathmandu Metropolitan City recording the highest mean NO₂ concentration (7.8 × 10^-5 mol m^-2, Z-score = +1.8), indicating a disproportionate pollution burden relative to other analytical zones. These findings establish a baseline for air quality monitoring in the valley and highlight the need for targeted emission reduction interventions in identified pollution hotspots.