A remote sensing technique for global monitoring of power plant CO<sub>2</sub> emissions from space and related applications

Abstract. Carbon dioxide (CO2) is the most important anthropogenic greenhouse gas (GHG) causing global warming. The atmospheric CO2 concentration increased by more than 30% since pre-industrial times – primarily due to burning of fossil fuels – and still continues to increase. Reporting of CO2 emissions is required by the Kyoto protocol. Independent verification of reported emissions, which are typially not directly measured, by methods such as inverse modeling of measured atmospheric CO2 concentrations is currently not possible globally due to lack of appropriate observations. Existing satellite instruments such as SCIAMACHY/ENVISAT and TANSO/GOSAT focus on advancing our understanding of natural CO2 sources and sinks. The obvious next step for future generation satellites is to also constrain anthropogenic CO2 emissions. Here we present a promising satellite remote sensing concept based on spectroscopic measurements of reflected solar radiation and show, using power plants as an example, that strong localized CO2 point sources can be detected and their emissions quantified. This requires mapping the atmospheric CO2 column distribution at a spatial resolution of 2×2 km2 with a precision of 0.5% (2 ppm) or better. We indicate that this can be achieved with existing technology. For a single satellite in sun-synchronous orbit with a swath width of 500 km, each power plant (PP) is overflown every 6 days or more frequent. Based on the MODIS cloud mask data product we conservatively estimate that typically 20 sufficiently cloud free overpasses per PP can be achieved every year. We found that for typical wind speeds in the range of 2–6 m/s the statistical uncertainty of the retrieved PP CO2 emission due to instrument noise is in the range 1.6–4.8 MtCO2/yr for single overpasses. This corresponds to 12–36% of the emission of a mid-size PP (13 MtCO2/yr). We have also determined the sensitivity to parameters which may result in systematic errors such as atmospheric transport and aerosol related parameters. We found that the emission error depends linearly on wind speed, i.e., a 10% wind speed error results in a 10% emission error, and that neglecting enhanced aerosol concentrations in the PP plume may result in errors in the range 0.2–2.5 MtCO2/yr, depending on PP aerosol emission. The discussed concept has the potential to contribute to an independent verification of reported anthropogenic CO2 emissions and therefore could be an important component of a future global anthropogenic GHG emission monitoring system. This is of relevance in the context of Kyoto protocol follow-on agreements but also allows detection and monitoring of a variety of other strong natural and anthropogenic CO2 and CH4 emitters. The investigated instrument is not limited to these applications as it has been specified to also deliver the data needed for global regional-scale CO2 and CH4 surface flux inverse modeling.