Abstract. The Gain-form of Local Ensemble Transform Kalman Filter (LGETKF) has been implemented in the Joint Effort for Data assimilation Integration (JEDI) with the Model for Prediction Across Scales – Atmosphere (MPAS-A) (i.e., MPAS-JEDI). LGETKF applies vertical localization in model space and is particularly convenient for assimilating satellite radiances that do not have an explicit vertical height assigned to each channel. Additional efforts are made to optimize the ensemble analysis procedure and improve the computational efficiency of MPAS-JEDI's LGETKF. This is the first application of JEDI-based LGETKF for assimilating radiance data in all-weather situations with a global MPAS configuration. The system is firstly tuned for covariance inflation and horizontal localization settings. It is found that using a combination of relaxation to prior perturbation (RTPP) and relaxation to prior spread (RTPS) outperforms using RTPP or RTPS alone, and using a smaller horizontal localization scale for all-sky radiances is preferable. With the optimized inflation and localization settings, assimilating all-sky radiances of the Advanced Microwave Sounding Unit – A (AMSU-A) window channels with an 80-member LGETKF improved the forecasts of moisture, wind, clouds, and precipitation fields, when compared to the benchmark experiment without assimilation of all-sky AMSU-A radiances. The positive forecast impact of all-sky AMSU-A radiances is the largest over the tropical regions up to 7-day. Some degradation on the temperature forecasts is seen over certain regions, where the model forecast is likely biased, causing deficiencies for assimilating all-sky data. The LGETKF capability is available in the recent public release of MPAS-JEDI and ready for research and operational explorations.