P6 Preliminary results of Local ETKF based ensemble prediction system using different inflation factor methods

Kay, Jun Kyung, and Hyun Mee Kim, Yonsei University

Ensemble prediction plays an important role in quantifying the distribution of forecast uncertainties. To conduct regional short range ensemble forecasts for meso-scale events in Asia, Weather and Research Forecast Model Advanced Research (WRF-ARW) modeling system is used with 24 ensemble members. The initial ensemble perturbations generated by Ensemble Transform Kalman Filter (ETKF) with localization are added to the National Center for Environmental Prediction (NCEP) final analysis, then run for 72 h. ETKF is a family of ensemble square root filters, and initial ensemble perturbations are updated by solving Kalman Filter equations considering the error estimates of observation and forecast. The random perturbations from background error statistics are added in the lateral boundary condition to consider the uncertainty of the regional ensemble forecast system in a limited domain.

Because the ensemble spread underestimates the forecast error uncertainties, two different kinds of inflation method that include multiplicative inflation factor and additive inflation factor are implemented in the ensemble prediction system, and the effect of these inflation methods on the quality of ensemble prediction are investigated in terms of probabilistic forecasts.