It’s once again time for the annual Spring Forecasting Experiment (SFE). As many of you may know, these experiments aim to facilitate collaboration between researchers and forecasters, encouraging researchers to better understand operational forecasting and forecasters to better understand the current capabilities of the most state-of-the-art, experimental numerical guidance systems that may eventually become operational – the number of which is increasing continually.
With so many numerical guidance systems being incorporated into last year’s SFE (there were six different ensembles in SFE 2015, along with a variety of deterministic guidance!), this year an effort was made to coordinate between the different agencies providing guidance. A big shout out goes to everyone who put their efforts into this coordination, which spans multiple agencies across the United States, including: the National Severe Storms Laboratory, the Center for the Analysis and Prediction of Storms, the University of North Dakota, NOAA’s Earth Systems Research Laboratory/Global Systems Division, and the National Center for Atmospheric Research. This ensemble of 65 members is called the Community Leveraged Unified Ensemble (CLUE), and will be a major focus of this year’s ensemble.
All members in the CLUE are initialized on weekdays at 0000 UTC (During the experiment, 0000 UTC = 1900 CST). The members have 3 km grid spacing, and cover the contiguous United States with forecast lengths ranging from 36 to 60 h, depending on the member. The subsets of the ensembles (details of which can be found in the SFE 2016 Operations Plan) are designed to test several aspects of convection-allowing model (CAM) ensemble design, particularly for an operational CAM ensemble. These tests include:
- A dynamic core comparison
- Multi-core vs. single-core ensembles
- Single physics vs. multi-physics
- Data assimilation method comparisons
- Microphysics sensitivities
- Ensemble size comparisons
While lots of verification work will remain after the experiment, participant feedback will allow for the subjective examination of some of these questions. Participants will look at the observed radar reflectivity, Local Storm Reports (LSRs), National Weather Service warnings, and Multi-Radar Multi-Sensor (MRMS) radar-estimated hail sizes to verify different aspects of the CLUE, such as simulated reflectivity, hourly maximum variables, and a variety of hail guidance. Differences between the members and subsets of members will hopefully shed some light on how different ensemble design strategies affect severe convective weather forecasting.
Clearly, we’ll have lots to discuss in the upcoming experiment. Stay tuned here for updates from SFE 2016.