[1]^?

The following includes a brief description of each model (and its corresponding data assimilation system) that will be utilized in this project.More details regarding the configuration of each model are presented in Table 2.

NCEP ETA-12 Model

The NCEP Environmental Modeling Center (EMC), part of the National Weather Service (NOAA/NWS), developed the Eta-12 model (herein Eta), which is a limited-area, numerical atmospheric model. The model integrates the primitive hydrostatic equations in three dimensions. The vertical coordinate is known as the Eta, which is achieved by modifying the terrain following sigma coordinate. The advantage of the eta coordinate is that the surfaces are quasi-horizontal, thus avoiding errors associated with steep slopes of the coordinate surfaces (Mesinger, 1984) and hence improving the solution over highly variable topography such as over the Western United States. Initial values to each Eta forecast model run are provided by the fully-cycled Eta Data Assimilation System (EDAS), which incorporates the 3-dimensional variational analysis (3D-VAR) technique.

Presently, NCEP Eta surface data (from netCDF files containing Eta forecast output mapped to AWIPS Grid 215, which has a horizontal grid spacing of 20 km) is sent every six hours to TAMUCC-DNR. The forecast data are stored for a set of about 40 locations illustrated in figure 3. A significant number of locations and forecasts other than surface forecasts will complement the present database. NCEP will likely discontinue providing Eta model outputs during the project duration. The model was selected for this project to complete the ongoing database and to test how a change in numerical weather product will affect the ANN models. This transition will be tested as the WRF model outputs become available. All other three model outputs are expected to be available throughout the duration of this project and beyond.

Back to Atmostpheric Predictions for Gulf of Mexico

The Weather Research and Forecasting (WRF) Model

This model is being developed collaboratively between a number of organizations including the Mesoscale, Microscale, and Meteorology Division of the National Center for Atmospheric Research (NCAR/MMM),NOAA/NWS, the Forecast Systems Laboratory (NOAA/FSL), the Oklahoma University Center for Analysis and Prediction of Storms, and the Air Force Weather Agency. The philosophy behind the development of this model is to create an advanced mesoscale community (non-proprietary) model and data assimilation system. Also, a close relationship between the research and operational communities that will utilize the model is encouraged. The WRF is a non-hydrostatic, limited-area, numerical model in which the domain and physics parameterizations can be chosen amongst several options. The current WRF contains several options for explicit microphysics,cumulus parameterization, planetary boundary layer,radiation, and parameterization of the surface. With version 1.3 of the WRF, initial values to each model run can be provided by static initialization, with or without 3DVAR. Eventually, dynamic initialization techniques will be available.

Of importance to this project, NCEP plans to discontinue operational Eta runs by October 2004, and replace them with North American mesoscale runs of the WRF (at a resolution similar to the Eta) and make them available to NWS forecasters to support operations. WRF output to the forecasters will be available on grids currently used for the Eta, which include AWIPS Grids 215, 218, 212, and 221, with horizontal grid spacings of 20km, 12km, 40km, and 32km respectively. The project will include testing of the ANN models with both the Eta and NCEP WRF forecasts to determine potential differences. Project objectives also include running models initially trained with Eta forecasts and testing them with WRF forecasts to measure the impact of the model transition on operational forecasts.

Back to Atmostpheric Predictions for Gulf of Mexico

NCEP Global Forecast System (GFS)

This Model is a hydrostatic, global, and spectral model maintained by NCEP/EMC. The GFS runs four times a day (0, 6, 12, 18 UTC) with forecasts computed for up to 384 hours. The spectral nature of the model affects the manner in which quantities are calculated. The computation of quantities in the horizontal plane is handled by the following methodology: space derivatives are computed in spectral space and selected quantities are transformed to a Gaussian grid for the computation of nonlinear products in grid space. The advantage of this method is that derivatives are computed accurately in spectral space, yet non linear products are computed efficiently in physical space.

Initial values to each GFS model forecast are provided by the Global Data Assimilation System (GDAS). The GDAS incorporates a technique referred to as Spectral Statistical Interpolation (SSI), a type of 3DVAR, to create each analysis.

GFS output from each run, like the Eta, is sent to each NWS Weather Forecasting Office via the NWS AWIPS Satellite Broadcast Network, in the GRIB file format. The data is then written to netCDF files, containing the data mapped to several numerical grids of varying spatial resolution. For the GFS, the AWIPS grids are 202, 211, and 213 with horizontal resolutions of approximately 180 km, 80 km, and 90 km respectively.

Back to Atmostpheric Predictions for Gulf of Mexico

PSU-NCAR MM5 Model

This Model was selected as the local model. The MM5 model is the Fifth Generation of a Mesoscale Model, originally developed at Penn State University, with current development and support provided by NCAR. The MM5, like the new WRF, is a non-hydrostatic, limited-area, community numerical model in which the domain and physics parameterizations can be chosen amongst several options. The current MM5 contains several options for explicit microphysics, cumulus parameterization, planetary boundary layer, radiation,and parameterization of the surface. The model will provide increased resolution as compared to the other gridded model and the ability of the local office to adapt the physics of the model. The model will be frozen early in the project and will serve as a control for comparison with the other models for the rest of the project.

Initial values to each MM5 model forecast can be provided via static or dynamic initialization.Static initialization can be performed with or without 3DVAR.Dynamic initialization is performed via Four Dimensional Data Assimilation (FDDA.) The MM5 Modeling System will be configured by the CC-WFO participants to the project.

Back to Atmostpheric Predictions for Gulf of Mexico

For more information, questions,comments or suggestions about this project E-mail : Dr.Phillipe Tissot

Edit^?