SPATIO-TEMPORAL ASSESSMENT OF CARBON MONOXIDE CONCENTRATIONS OVER PAKISTAN IN RELATION TO CLIMATE CHANGE

Abstract

Air pollution has emerged as a key environmental issue of Pakistan as its major cities are ranked among top polluted cities in the South Asian region. Due to lack of adequate ground-based monitoring facilities, satellite observations are the only viable source for long-term assessment of air quality over the region. The current study analyses the spatio-temporal distribution of CO over Pakistan from year 2000 to 2017 using CO total columns derived from MOPITT observations. An overall absolute decrease of 9.00 ± 1.1 x 1016 molecules/cm2 in columnar CO is observed with temporal decrease of 4.81% at the rate of 0.26% per year. The study showed a distinct yearly decrease in the levels of columnar CO over different provinces of Pakistan irrespective of higher anthropogenic activities. Therefore, contributions of CO emissions from various sectors using ECCAD (Emissions of Atmospheric Compounds and Compilation on Ancillary Data) emission inventories datasets were explored to understand anthropogenic and biomass burning emissions of CO over Pakistan. The trends in emission inventories are calculated and compared to the corresponding columnar CO. The datasets from biomass burning emissions inventories showed decreasing trends similar to columnar CO whereas, all other anthropogenic emissions inventories datasets showed increasing trends. This increase is mainly attributed to increased CO emissions from residential, industrial and transportation sectors. However, it’s difficult to relate decrease in CO emissions from biomass burning sectors to decrease in CO columns during the study period over Pakistan. Overall contributions of biomass burning emissions from agricultural waste burning and fires are about ≤1% (for GFED4 dataset) around ≤2% (for EDGAR dataset) and 8% from agricultural waste burning (for EDGARv4 dataset). It is observed that an anthropogenic activity is increased over Pakistan during the last decade. Therefore, it is presumed that decrease in anthropogenic emissions may not be the only cause of decreasing trend in CO over Pakistan. Some other dominating factors like changes in tropospheric chemistry (sources and sinks) might be responsible for CO decrease x over the decade that needs to be further explored. Hence, CO mixing ratios (surface VMRs) from MOPITT are used to explore association of CO with ozone (O3) and its precursor gases (NO2 and HCHO) by employing OMI data over Pakistan. High levels of all these trace gases were observed over Punjab region, which may be attributed to its metropolitan importance. Seasonal Mann-Kendall (SMK) test is used to calculate the statistical significance of temporal trend by taking into account seasonality during the study period. Surface CO showed a significant decrease of about 13 ± 1% over the decade. Tropospheric ozone (TO3) exhibited a significant increasing trend with a temporal increase of 10.4 ± 1.2% per decade whereas NO2 showed a significant temporal increase of about 28 ± 2% per decade. However, an insignificant increase of about 8.3 ± 0.3% per decade is estimated for HCHO over the decade. A marked seasonal behavior is observed among the trace gases. The long-term relationship among trace gases is further explored by statistical tests. It is found that O3, NO2 and HCHO have a significant impact on the long-term decrease of CO whereas; the trend in O3 critically depends on NO2 levels along with CO over the region. A statistical model is developed to estimate the long-term association of trace gases over the region. The model indicates that O3 production may have major influence on decrease of CO over Pakistan in a (volatile organic compound) VOC-limited regime. An impulse response function (IRF) analysis using vector auto regression (VAR) approach is used to investigate the relationship between tropospheric trace gases under study. Furthermore, the study is designed to develop forecast model for carbon monoxide (CO) and tropospheric ozone (TO3) based on VAR. Results revealed that VAR model can be successfully used for forecasting the monthly concentrations of atmospheric pollutants. It is found efficient at all scales and worked well for more polluted regions. However, the performance of VAR model is quantitatively underestimated in case of relatively cleaner cities. The observed differences might be attributed to O3 precursors (NO2 and HCHO); as significant role of precursor gases is observed in forecasting CO and O3 over highly polluted regions.