Impact of tropical biomass burning emissions on the diurnal cycle of mid to upper tropospheric CO2 retrieved from NOAA-10 satellite observations
Alain Chédin(1), Soumia Serrar, Noëlle-A. Scott(1), Clémence Pierangelo(1), and Philippe Ciais(2)
(1)Laboratoire de Météorologie Dynamique/IPSL,
(2)Laboratoire des Sciences du Climat et de l'Environnement/IPSL
XIV International TOVS Study Conference, China, 25-31 May 2005
Four years (July 1987 to June 1991) of monthly mean mid to upper tropospheric CO2 mixing ratios over the tropics are retrieved from the observations, at 7.30 am (day) and 7.30 pm (night) local time, of the meteorological satellite NOAA-10. Analysis of night minus day differences (N-DD) shows large diurnal variations of CO2, of the order of 2-3 ppm, during months and over regions affected by biomass burning. The patterns of these diurnal variations are in very good agreement with the diurnal and seasonal variations of biomass burning activity. We interpret them as the signal of CO2 plumes being rapidly uplifted by fire-induced convection into the upper troposphere during the daytime peak of fire activity, and then rapidly dispersed at night by large scale atmospheric transport. The upper air CO2 diurnal cycle closely follows the seasonal distribution of burned areas from the European Space Agency's monthly Global Burnt Scar (GLOBSCAR) satellite product, which is recognized as yielding reasonable estimates of burnt areas for large and presumably intense fires. The largest N-DD values are found in 1990 over southern Africa in agreement with the reported inter-annual variability of fire activity. However, the magnitude of these extreme signatures, in the order of 5 ppm locally, is larger than what can reasonably be expected from either in situ observations or from simulations, suggesting some contamination of the N-DD retrieval by fire emission products other than CO2. It is concluded from a detailed sensitivity analysis, that the presence of high altitude and large optical depth aerosols, or of elevated tropospheric ozone concentrations, as often encountered in fire plumes, may significantly contaminate the retrieved CO2 signal (by up to 2-3 ppm for extreme events). The possible contaminating effects of undetected fire-induced thin cirrus (optical depths less than 0.05 at 14 μm), is also quantified.