Vertical localization strategy for radiance data within KIAPS-LETKF system

We have attempted to estimate forecast sensitivity to observations (FSO) within KIAPS-LETKF system using the method introduced by Kalnay et al. (2012) which has been applied to NCEP GFS-EnSRF data assimilation system by Ota et al. (2013). Ota et al. (2013) has shown promising results of 24-hour Ensemble Forecast Sensitivity to Observation (EFSO) with real conventional data. Within KIAPS-LETKF data assimilation system that has been successfully implemented to NCAR CAM-SE model, we have examined a performance of estimating EFSO for several forecast times. In addition, we have investigated better way to define the localization in the case of forecast time from 24 hours up, using OSSEs with Lorenz 40-variable model and NCAR CAM-SE model. New localization method can change the localization scale depending on a moving distance of the center of localization function. We will discuss how the flow-dependent localization function can improve the results.