Kunii, M., 2014: The 1000-member ensemble Kalman filtering with the JMA nonhydrostatic mesoscale model on the K computer. J. Meteor. Soc. Japan, 92, 623-633.
Special Edition on AICS International Workshop on Data Assimilation
- The influence of sampling noise on background error covariance was estimated by running the 1000-member EnKF with the JMA nonhydrostatic mesoscale model on the K computer.
- The background error covariance estimated from 1000 ensemble members contained negligible sampling noise even at distant locations without covariance localization (Fig. 1).
- For the case in the current study, an ensemble size of 500 would be large enough to approximate the error covariance under the configuration with a horizontal resolution of 15 km (Fig.2).