Satellite based analysis of fire–carbon monoxide relationships from forest and agricultural residue burning (2003–2011)

Carbon monoxide (CO) is an important greenhouse gas that is emitted during the incomplete combustion of biomass burning. In this study, we assessed the Measurements Of Pollution In the Troposphere (MOPITT) CO retrievals from two different biomass burning regions, fires in the evergreen forests of Northeast India and agriculture residue fires, Punjab, India. We analyzed long-term trends (2003–2011) in CO retrievals and fire–CO relationships including CO profiles at nine different atmospheric levels. Over a ten year period, the mean monthly CO for Northeast India ranged from 140.86 ppmv (−1σ) to 348.85 ppbv (+1σ) with a mean CO of 244.85 ppbv. We observed a clear increase in CO signal from February to March followed by a decrease in May coinciding with the fire signal. In Punjab, the mean monthly CO ranged from 158.21 ppbv (−1σ) to 286.40 ppbv (+1σ) with a mean CO of 222.30 ppbv. Comparison of mean CO during the peak fire months suggested relatively higher CO (439.06 ppbv) during March (evergreen forest burning) than October (194.83 ppbv) agricultural residue burning. We found MODIS fire radiative power (FRP) as a stronger predictor of surface CO signal than the fire counts in the evergreen forest fires. The segmented regression model fitted using nine years of FRP–CO data was useful in finding the FRP threshold impact on CO concentrations in the evergreen forests. To explain the low correlation between fires and MOPITT CO signal from the agricultural residue fires, we used the CALIPSO data to infer the smoke plume heights. Results suggested an average smoke plume height of 2.2 km during the peak biomass burning month from agricultural fires, compared to 4.61 km from evergreen forest fires. Overall, the MODIS FRP and CALIPSO data were useful in understanding the MOPITT CO sensitivity to fires.

► Satellite data revealed significant CO pollution from forest burning. ► Fire radiative power (FRP) was found to be a better indicator of CO emissions than fire counts. ► Segmented regression was useful in identifying the FRP threshold impacting forest CO emissions. ► CALIPSO data revealed low smoke plume heights for agriculture fires than forest fires.