Ensemble Subsetting
Ensemble subsetting is a post-processing approach that preferrentially weights ensemble members by their relative errors at early forecast hours. For instance, ensemble members that contain small errors at forecast hour 6 could be "selected" into a subsetted ensemble that statistically contains the best performing members. This technique could be valuable in operational forecasting environments where observations are readily available to subset an ensemble before the next ensemble data assimilation cycle. Previous work has demonstrated its validity for synoptic-scale weather systems (Ancell 2016). Our research group at Texas Tech has developed this technique and applied it to convection forecasts for the spring of 2018. Testing and evaluation is ongoing at the Hazardous Weather Testbed Spring Forecast Experiment.
