Applying Time-Expended Sampling to Ensemble Assimilation of Remote-Sensing Data for Short-Term Predictions of Thunderstorms

By sampling perturbed state vectors from each ensemble forecast at additional time levels shifted by ±τ (where τ is a selected interval) the analysis time, time-expanded (TES) can not only sample timing errors (or phase errors) but also triple size for covariance construction without increasing size. In this study, TES was applied to convection-allowing ensemble-based warn-on-forecast system (WoFS), four severe storm events, reduce computational costs that constrain real-time applications in assimilation of remote-sensing data radars and geostationary satellite GOES-16. For event, implemented against 36-member control experiment (E36) reducing 12 tripling × 3 = 36 with 2.5 min, 5 min 7.5 three experiments, named E12×3τ2.5, E12×3τ5 E12×3τ7.5, respectively. A 0–6-h created hourly after second hour during experiment. The statistics were evaluated event found be little affected TES, while cost. forecasts produced verified multi-sensor observed/estimated rainfall, reported tornadoes damaging winds event. verifications indicated experiments had about same capability quality as E36 predicting rainfall probabilities winds; addition, predictive sensitive τ, although they slightly enhanced selecting min. These results suggest attractive useful cost-saving WoFS generation short-term severe-weather forecasts.