Assessing JEDI GETKF Performance During Periods of Incresed Medium-Range GEFS Forecast Variability

An effort is ongoing at the National Centers for Environmental Prediction (NCEP) to replace the data assimilation framework from the currently operational Gridpoint Statistical Interpolation (GSI) to the new Joint Effort for Data assimilation Integration (JEDI) based software. One aspect needed for this transition is evaluating the atmospheric performance of the Gain form of the Ensemble Transform Kalman Filter (GETKF) solver which updates the analysis perturbations for the 80-member ensemble competent of the Global Data Assimilation System (GDAS). Early result bulk statistics have suggested that the JEDI system ensemble solver produces similar results as a comparative GSI system. Given this promising result, this study focuses on evaluating the ensemble solver for a select set of cases, periods when initial state uncertainty was linked with increased Global Ensemble Forecast System (GEFS) medium-range 500-hPa height anomaly correlation variability. Through the use of cycling systems, one using the JEDI prototype and one with a comparative version of GSI, this study will compare short-term (0–6-h) forecasts with in-situ observations, increment values, and the measurement of ensemble spread occurring in key fields linked to these uncertain events. The goal of this evaluation is to understand if JEDI can provide similar or better information than the GSI during these highly variable situations.