Open AccessNational Blend of Models: A Statistically Post-Processed Multi-Model Ensemble
Author(s) -
Jeffrey P. Craven,
David E. Rudack,
Phillip E. Shafer
Publication year - 2020
Publication title -
journal of operational meteorology
Language(s) - English
Resource type - Journals
ISSN - 2325-6184
DOI - 10.15191/nwajom.2020.0801
Subject(s) - dew point , meteorology , probabilistic logic , computer science , environmental science , ensemble forecasting , global forecast system , service (business) , process (computing) , operations research , weather research and forecasting model , geography , engineering , business , artificial intelligence , marketing , operating system
The National Blend of Models (NBM) is the culmination of an effort to develop a nationally consistent set of foundational gridded guidance products based on well-calibrated National Weather Service (NWS) and non-NWS model information. These guidance products are made available to the National Centers for Environmental Prediction centers and NWS Weather Forecast Offices for use in their forecast process. As the NWS continues to shift emphasis from production of forecast products to impact-based decision support services for core partners, the deterministic and probabilistic output from the NBM will become increasingly important as a starting point to the forecast process. The purpose of this manuscript is to document the progress of NBM versions 3.1 and 3.2 and what techniques are used to blend roughly 30 individual models and ensembles for a number of forecast elements and regions. Focus will be on the core elements such as (1) temperature and dew point temperature, (2) winter weather, fire weather, thunderstorm probabilities, and (3) wind speed and gusts.