A Preliminary Survey of Forecaster Decision Making in the Blended Model Era

As the National Weather Service (NWS) implements the National Blend of Global Models (NBM), forecast methodology continues to evolve away from creating “grids-from-scratch” toward an “initialize-collaborate-modify” approach for population of the National Digital Forecast Database.  This evolution requires forecasters to identify windows of opportunity for manual deviation from model blends, where human intervention continues to improve forecast and warning services for high-impact weather relative to blended numerical model initializations.  Recognition of such windows of opportunity requires use of external data and local, regional, and national forecaster expertise.  

Previous studies, such as Environment Canada’s Project Phoenix, have shown that an over reliance on model data can result in a degraded forecast. Model blends add another layer with the potential to compound this problem. An integral challenge in training NWS forecasters for their role in interaction with the NBM is establishing tools and techniques to properly identify windows of opportunity and bases for deviations from blends.  Surveys and forecaster interviews were conducted to gain insight into forecaster decision making processes using NBM predecessor blends such as Consensus of All models and Southern Region SuperBlend.  Preliminary results show that most windows of opportunity are found in short-term (days 1-2) forecasts of precipitation.  High-resolution Convection-Allowing Models currently not incorporated into operational model blends, observations, and conceptual models are often used as inputs for modifying short-term forecasts critical to impact-based decision support services.  Other reasons for deviation include: adding precision beyond the default blend resolution, correcting for known model and parameterization errors, and subjective evaluation of model consistency and recent model performance. This presentation will discuss preliminary results of the surveys and forecaster interviews as well as implications for the future of the forecaster training and decision making tools.