17:15 〜 18:30
[AAS01-P05] LETKFデータ同化手法による温室効果ガスの地表面フラックスの推定
キーワード：炭素循環、温室効果ガス、データ同化、LETKF
We present global CO2 flux estimations using the local ensemble transform Kalman filter (LETKF) system with the GOSAT obtained XCO2 and the WDCGG compiled CO2 concentration data. In the previous study [Miyazaki et. al., 2011], a performance of the LETKF system was evaluated using GOSAT column pseudo-data in reference and the other various types of CO2 concentration data. Here, we use the GOSAT retrievals to estimate the flux with the 4-D data assimilation system.
The data assimilation system used in this study was developed by Miyazaki et al., 2011, on the basis of the LETKF scheme [Miyoshi et al.]. A basic methodology of the LETKF follows the original EnKF [Ott et al., 2004; Hunt et al., 2007].The covariance localization [Houtekamer and Mitchell, 2001] is used to remove long range spurious correlations. The state vector augmentation method [Anderson, 2001; Aksoy et al., 2006; Tong and Xue, 2008] has been applied to simultaneously estimate the atmospheric CO2 concentration as model states together with the surface CO2 flux as uncertain model parameters. The surface fluxes at every model grid points are analyzed with 4-daily assimilation window during 2012 year. The ensemble size is hundreds. The transport model is coupled with the Center for Climate System Research/National Institute for Environmental Studies/Frontier Research Center for Global Change (CCSR/NIES/FRCGC) atmospheric general circulation model (AGCM) version 5.7b [Numaguti et al., 1995]. The model spatial resolutions are horizontally T42 truncation (approximately 2.8 degree) and vertically 32 levels up to 7 hPa. The surface CO2 concentrations used in this study are obtained with the flask sampling data observed at sites in the surface network, which is archived at the WDCGG, and the XCO2 concentrations are retrieved from GOSAT soundings using the RemoTec algorithm [Butz et al., 2009]. These observational data assimilate into the transport model. The LETKF system performance is evaluated by error reduction ratio of the posterior to prior ensemble fluxes.
We show analysis results that are the error reduction ration depending on various types of the observational data and seasonal variability of the optimized fluxes over aggregated land scale.
Acknowledgements. The authors thank the RemoTeC Proxy products retrieved from GOSAT TANSO-FTS SWIR
spectra using the RemoTeC algorithm that is being jointly developed at SRON Netherlands Institute for Space Research and the Karlsruhe Institute for Technology (KIT).
The data assimilation system used in this study was developed by Miyazaki et al., 2011, on the basis of the LETKF scheme [Miyoshi et al.]. A basic methodology of the LETKF follows the original EnKF [Ott et al., 2004; Hunt et al., 2007].The covariance localization [Houtekamer and Mitchell, 2001] is used to remove long range spurious correlations. The state vector augmentation method [Anderson, 2001; Aksoy et al., 2006; Tong and Xue, 2008] has been applied to simultaneously estimate the atmospheric CO2 concentration as model states together with the surface CO2 flux as uncertain model parameters. The surface fluxes at every model grid points are analyzed with 4-daily assimilation window during 2012 year. The ensemble size is hundreds. The transport model is coupled with the Center for Climate System Research/National Institute for Environmental Studies/Frontier Research Center for Global Change (CCSR/NIES/FRCGC) atmospheric general circulation model (AGCM) version 5.7b [Numaguti et al., 1995]. The model spatial resolutions are horizontally T42 truncation (approximately 2.8 degree) and vertically 32 levels up to 7 hPa. The surface CO2 concentrations used in this study are obtained with the flask sampling data observed at sites in the surface network, which is archived at the WDCGG, and the XCO2 concentrations are retrieved from GOSAT soundings using the RemoTec algorithm [Butz et al., 2009]. These observational data assimilate into the transport model. The LETKF system performance is evaluated by error reduction ratio of the posterior to prior ensemble fluxes.
We show analysis results that are the error reduction ration depending on various types of the observational data and seasonal variability of the optimized fluxes over aggregated land scale.
Acknowledgements. The authors thank the RemoTeC Proxy products retrieved from GOSAT TANSO-FTS SWIR
spectra using the RemoTeC algorithm that is being jointly developed at SRON Netherlands Institute for Space Research and the Karlsruhe Institute for Technology (KIT).