TL;DR
This paper introduces a novel "top-down" method to estimate CO2 emissions from Siberian wildfires by combining satellite observations of co-emitted species (CO and aerosols) with a chemistry-transport model. The approach yields significantly higher emission estimates than existing "bottom-up" inventories, highlighting substantial uncertainties in our understanding of biomass burning impacts on the carbon cycle.
What Problem Does This Paper Address?
Accurately quantifying CO2 emissions from wildfires is crucial for understanding their impact on the global carbon cycle and climate change. Existing "bottom-up" emission inventories, which rely on satellite data like burnt area or fire radiative power (FRP) and emission factors, often have large uncertainties and can disagree significantly with each other. Furthermore, directly estimating CO2 emissions from biomass burning using traditional inverse modeling techniques is challenging due to the long atmospheric lifetime of CO2 and its complex exchange with other carbon sinks and sources. This paper aims to address this by developing a more robust method to constrain CO2 emissions from wildfires, specifically focusing on the Siberian region, which is a major contributor to global biomass burning.
Methods
The core of the method involves using satellite-observed co-emitted species (carbon monoxide - CO, and aerosol optical depth - AOD) as tracers to constrain the biomass burning rate (BBR). The BBR is then related to the fire radiative power (FRP) measured by MODIS satellites through empirically derived conversion factors (FRP-to-BBR). The key innovations are:
- Estimating FRP-to-BBR conversion factors (α): Instead of relying solely on local experimental fire data, the authors assimilate satellite observations of CO and AOD into the CHIMERE chemistry-transport model (CTM) to optimize these conversion factors for different land cover types (forest and grassland). This allows for regional-scale estimation and accounts for real-world fire conditions.
- Optimal Combination of Tracers: Estimates of α derived independently from CO and AOD are optimally combined using a probabilistic approach, reducing overall uncertainty.
- Diurnal Cycle Estimation: The diurnal cycle of fire emissions is directly estimated from FRP measurements, improving the temporal representation of emissions.
The method uses a "twin experiment" approach within the CTM to iteratively refine the α values by minimizing the difference between simulated and observed CO and AOD. Bias correction is also incorporated to account for non-fire-related atmospheric composition variations.
Datasets
- MODIS Infrared Measurements: Used to derive Fire Radiative Power (FRP) data.
- Infrared Atmospheric Sounding Interferometer (IASI): Used for carbon monoxide (CO) total column retrievals.
- MODIS Aerosol Optical Depth (AOD): Used as a proxy for aerosol emissions.
- Ground-based CO measurements: Data from the Zotino Tall Tower Observatory (ZOTTO) were used for validation.
- Global Emission Inventories: GFED3.1 and GFASv1.0 were used for comparison with the study's estimates.
Key Insights and Ideas
- CO and Aerosols as Powerful Tracers: The study demonstrates the effectiveness of using both CO and aerosols as co-emitted tracers to constrain biomass burning emissions, offering a more robust approach than relying on a single species.
- Regional Calibration is Crucial: Local experimental fire data may not accurately represent the complex conditions of large-scale wildfires. Regional calibration using atmospheric observations within a CTM is essential for accurate emission estimates.
- Significant Underestimation by Bottom-Up Inventories: The "top-down" estimates from this study are substantially higher than established "bottom-up" inventories, suggesting that current global estimates of Siberian wildfire emissions may be significantly underestimated.
- Robustness of the Method: The developed method shows robustness to variations in assumptions regarding emission injection height and diurnal emission cycles, increasing confidence in the results.
Key Results
- CO2 Emissions Estimates for Siberia (2012): The optimal estimate for total CO2 emissions from Siberian wildfires in 2012 was 392 Tg C, with an uncertainty range of 280–550 Tg C.
- Comparison with Inventories (2007-2011): The study's "top-down" estimates for annual biomass burning CO2 emissions in Siberia were 1.8 to 2.5 times larger than those from GFED3.1 and GFASv1.0, respectively.
- Validation with Ground Data: Simulations using the optimized emission factors showed good agreement with independent ground-based CO measurements at the ZOTTO site, supporting the validity of the method.
- FRP-to-BBR Conversion Factors: The derived conversion factors for forest and grassland fires were found to be consistent with previous local experimental estimates.
- Sensitivity Analysis: Estimates of CO2 emissions were found to be robust, varying by less than 15% under different assumptions for injection height and diurnal variations.
Limitations and Future Work
- Uncertainties in Emission Factors: While the method aims to reduce uncertainties, the study acknowledges that uncertainties in the underlying emission factors for CO2, CO, and aerosols still contribute to the overall error.
- Spatial Distribution Discrepancies: Significant differences in the spatial distribution of emissions between this study and existing inventories suggest a need for further analysis to understand these systematic variations.
- Potential Underestimation of Aerosol Emissions: The study notes that aerosol-derived conversion factors were larger and more uncertain than CO-derived ones, potentially indicating an underestimation of aerosol emission factors in the literature.
- Future Work: The authors suggest that incorporating satellite CO2 measurements of major fire plumes could further improve the estimation of CO2 emission ratios and reduce uncertainties. Further dedicated studies are recommended to understand the large discrepancies observed between different emission estimates and to improve the knowledge of biomass burning processes.
Summary updated