Overview of the NMC Production Suite November 24, 1993 John D. Stackpole Chief, Production Management Branch Automation Division file:winworda\handbook\overview ABSTRACT As an operational forecasting unit, the National Meteorological Center (NMC) is charged with providing a variety of objective weather analyses and forecast products to users throughout the world on a scheduled basis. These products are based, obviously, on synoptic and quasi-synoptic observations made of the world's weather conditions which find there way to the NMC by various national and international communications routes. These observations are then processed (terrible word) by sets of different computer jobs which are run in specific sequences on a specified schedule. Each of the jobs contain components needed to generate analyses, forecasts, and a variety of derived products tailored to the specific needs of sundry users. As many as six different sets of forecast products are generated in each 12 hour period. This requires a substantial amount of computer resources and careful management of the time available on these system. In the following we describe: the overall data flow through the NWS and NMC as observations get transmogrified into forecasts, the computer resources available to NMC, and the components of the suite of regularly scheduled forecasts. The last named includes thumb-nail sketches of the numerical models used in making the forecasts. The data received at NMC are collected from a myriad of sources, including domestic NWS observations transmitted via AFOS, aviation reports received from the ARINC company (an FAA arrangement), international observations received on the WMO Global Telecommunications System (GTS), and satellite soundings produced by the National Environmental Satellite, Data, and Information Service (NESDIS). The vast majority of these data pass through the National Weather Service Telecommunications Gateway (NWSTG) on their way to NMC. A total of approximately 9000 different reports are processed every twelve hours, including radiosondes, land surface and ship observations, fixed and drifting buoys, aircraft reports, satellites, cloud tracked winds, etc. The sets of computer jobs which ingest meteorological information, analyze it, predict the future state of the atmosphere, and prepare the output for world-wide distribution consist of a number of sub-suites, or "runs". These runs, although differing in purpose and content, have much in common: each includes gathering up as many current observations of the atmosphere as are available at the time of the run, analyzing the state of the atmosphere, forecasting its future state, and then making the observations, the analysis, and the forecast information available to a wide assortment of users both inside and outside of the NWS. There are five, and sometimes six, such runs; all but one are repeated twice per day, using observational data taken within 3 hours of the 0000 UTC (00Z) and 1200 UTC (12Z) synoptic observing times. These are referred to as the 00Z and 12Z "cycles", respectively. An additional "Final" run is performed using a much later data collection time which includes all delayed data and also uses 0600 UTC and 1800 UTC data in analyses made for those times. The forecasts from this run provide the guess fields for all of the NMC analysis systems, with the partial execption of the regional run which has a Regional Data Assimilation System (RDAS) all its own. The six runs are named either by their relative position within each cycle ("Early" or "Final"), or by their general purpose ("Regional", "Aviation", "Hurricane", or "Medium-Range Forecast"). The runs are referred to using the following abbreviations: ERL (Early), intended to provide a "quick-look" regional forecast update using an early data cutoff with the LFM model; now also includes the eta model with a not-quite-so-quick-look but much higher resolution and far better physical parameterization; Note: the LFM model is being phased out of NMC operations; RGL (Regional), providing additional detailed regional forecasts using more rawinsonde data and providing detailed precipitation guidance; AVN (Aviation), intended to provide longer range, global forecasts of synoptic systems and upper level winds; HCN (Hurricane), executed on demand to provide forecasts for hurricane tracking; MRF (Medium-Range Forecast), providing extended range global forecasts for up to 15 days, with ensemble forecast techniques as well; and FNL (Final Analysis), used to assimilate all the observed data reaching NMC into the best possible analyses and short-range forecasts to be used as the basis for the next cycle's operational analyses. The RGL run has a Regional Data Assimilation System (RDAS) all its own. Each of the runs comprises many jobs, some doing meteorological computations using the high-speed vector processing computer, while other jobs on other computer systems are used to collect and format observed data, generate graphics and bulletins for external transmissions, or preserve the data, analyses, and forecasts for research or other future needs. Indeed, a large fraction of the total elapsed time in the forecast runs is taken up with the generation of products for displaying and disseminating meteorological information. Output from the NMC analyses and forecasts are distributed in a variety of forms and over a variety of media to meet the needs of the parent National Weather Service and other users, both nationally and internationally. As with the incoming data, the vast majority of NMC's output products pass through the NWSTG on their way to the end users.