Study on data assimilation of vapor isotope with the ensemble Kalman filter

Since the first isotopic AGCM came out in 1980s, it has been dreamed of "Isotope Reanalysis" for better analyses of global vapor/precipitation fields in high spatial and temporal resolution, but it has not been realized in the same way as the common Reanalysis products. There were three main reasons for this incapability: 1) Too few observation data available 2) Low accuracy of the model and 3) No appropriate assimilation algorithm available. However, the situation has been changed: 1) vapor isotopes have recently become observable in a spatially and temporally high resolution by a remote sensing technique (e.g., Worden et al., 2007); 2) the recent isotopic AGCM also has shown nice reproducibility of precipitation isotopes when the large scale circulation fields are constrained (Yoshimura et al., 2008); and 3) the Ensemble Kalman Filter approach, which is practically the most appropriate for data assimilation of the isotopic AGCM, became feasible in terms of computational cost (e.g., Miyoshi and Yamane, 2007). Furthermore, the nudged simulation, which was carried out prior to the realization of the dream, showed a significant correlation (R=0.55) in vapor dD compared with the TES observations, implying large possibility of improvement in modeled vapor/precipitation isotopes by having isotopic information assimilated. This study therefore aims to make the first "Isotope Reanalysis" by applying LETKF and TES data to IsoGSM, and some experimental results are going to be presented.