Inverse modeling of biomass burning emissions using Total Ozone Mapping Spectrometer aerosol index for 1997

Abstract

We present results from an inverse model study to determine biomass smoke emissions for the year 1997 by comparison of modeled aerosol index (AI) with that measured by the EP TOMS instrument. The IMPACT model with Data Assimilation Office (DAO) meteorology data in 1997 is utilized to obtain aerosol spatial and temporal distributions. Then a radiative transfer model is applied to generate the modeled AI. A Bayesian inverse technique is applied to optimize the difference between the modeled AI and the EP TOMS AI in the same period by regulating monthly a priori biomass smoke emissions in seven predefined regions. The modeled AI with a posteriori emissions is generally in better agreement with the EP TOMS AI. The a posteriori emissions from Indonesia increase by a factor of 8–10 over the a priori emissions due to the Indonesian fires in 1997. The annual total a posteriori source increases by about 13% for the year 1997 (6.31 Tg/yr black carbon and 67.27 Tg/yr smoke) in the base scenario, with a larger adjustment of monthly emissions. The sensitivity of this result to the a priori uncertainties, the height of the smoke layer, the cloud screening criteria, the inclusion of an adjustment of emissions outside the main biomass burning regions, and the inclusion of the covariances between observations in different locations is discussed in a set of sensitivity scenarios. The sensitivity scenarios suggest that the inverse model results are most sensitive to the assumed uncertainty for a priori emissions and the altitude of aerosol layer in the model and are less sensitive to other factors. In the scenario where the uncertainty of a priori emissions is increased to 100% (300% in Indonesia), the total annual black carbon emission is increased to 6.87 Tg/yr, and the smoke emission increases to 73.39 Tg/yr. The a posteriori emissions in Indonesia in the scenario with increased uncertainty are in better agreement with both the TOMS AI and with previous estimates for the Indonesian fires in 1997. In the scenario where biomass smoke from large fires are elevated by 1 km in altitude, the annual total black carbon emissions are 5.68 Tg/yr, and the smoke emissions are 60.44 Tg/yr, almost unchanged from the a priori emissions.