Tropospheric carbon dioxide and methane temporal variability using atmospheric infrared sounding data: a case study of Pakistan

Abstract

Rising levels of carbon dioxide (CO₂) and methane (CH₄) in the atmosphere are significant contributors to global climate change, although regional differences and mechanisms are poorly understood, especially in South Asia. This study examines the spatial and temporal patterns, seasonal changes, and climatic effects of CO₂ and CH₄ over Pakistan through satellite measurements (AIRS, 2002–2017), weather, and vegetation indicators (NDVI). We evaluate the contribution of human-made activities, biomass burning, and natural processes (e.g., monsoon or soil respiration) to the regulation of greenhouse gas (GHG) concentrations. Moreover, we assess the contribution of long-range transportation by our neighboring areas (the Middle East and Central Asia) using HYSPLIT trajectory modeling. The results show an average yearly growth of CO₂ (2.1 ppm per year) and CH₄ (3.5 ppb per year), seasonal peaks of CO₂ (spring) and CH₄ (summer), associated with agriculture, temperature-dependent respiration, and monsoonal cycles. CO₂ and NDVI (−0.50) and CH₄ and NDVI (+0.64) depict negative and positive associations, respectively, and play the role of vegetation as a carbon sink and wetland and rice paddy emissions. Other significant findings of the study are sudden changes in GHG patterns (CO₂: 2009; and CH₄: 2007–2014) that occur with upward temperatures, indicating climate feedbacks. This study incorporates radiative forcing dynamics and air mass paths, which provide important insights into the regional GHG drivers and their climatic implications and contribute to policy interventions to reduce emission levels in South Asia. The cloud fraction had a negative correlation with both CO₂ (r = −0.36 and p < 0.04) and CH₄ (r = −0.20 and p < 0.03). The trajectories of the air mass of the rear indicate that the distant pollution of neighboring countries is a factor. Burning of crop residues, car emissions, forest burning, and others release small quantities of gases and contaminants into the air. This study compares atmospheric CO₂ and CH₄ prediction models. The dominant trend is strong linearity. In the case of CH₄, linear regression is the best and most suggested model. In the case of CO₂, ARIMA provided the most accurate forecasts by detecting minor autocorrelation. More complicated models, such as LSTM, failed to work, which proved that simpler models are effective on this kind of data.