Coupled Data Assimilation and El Nino Prediction

The world's first application of 4-dimensional variational data assimilation method, which has an ability to most efficiently assimilate the observational information into the model, to a coupled atmosphere-ocean-land surface model has been successfully accomplished. This enables us to reproduce the atmosphere-ocean-land surface state of the 1990s as an integrated data, mainly focusing on the seasonal to interannual climate variability (Figure 1). Furthermore, the preliminary experiment illustrates that the skill of El Nino prediction is significantly improved by the initialization of coupled field using the coupled data assimilation method (Figure 2).

Figure 1. The representation of the NINO3.4 SST (red and blue lines: reproduced, black: observation). Coupled data assimilation (bottom panel) reproduces the state of the 1990's much better than simulation (top panel).		Figure 2 The prediction skill of NINO3.4 SST (red curve: initialization by CDA, blue: initialization by IAU, black: persistence). The initialization by coupled data assimilation significantly improves the prediction skill compared to other initialization methods.