About the DTC Projects

The Developmental Testbed Center conducts forecasting experiments, regularly. The current experiment is highlighted, here. For links to previous experiments, please see links in the righthand, "Related Links" column.

NMM5-conus Experiment (begins April 1, 2005)

The NMM5-conus Project has two primary motivations:

1) Offer high-resolution WRF model forecasts to National Weather Service (NWS) forecasters during the convective season. NMM5-CONUS employs a 5-km spacing grid over the entire CONUS and is therefore able to represent mesoscale weather phenomena that are not captured by coarser models. The model will be run without a convective parameterization, in a configuration identical to what was run recently for the DTC Winter Forecast Experiment (15Jan05 - 31Mar05), the only exception being upgrades to the NMM model radiation physics and fixes to the initial conditions produced by the WRF Standard Initialization (SI).

2) Support the NOAA Coastal Storms Initiative (CSI) Project in a study to address the impact of model domain size in local forecasting. The NMM5-CONUS forecasts will be compared to those made with an identical version of the NMM model, except over a much smaller domain centered over Florida. The purpose of this study is to examine the importance of boundary condition influence on short-range numerical weather prediction, in order to determine whether the same degree of accuracy can be obtained with a much smaller domain as one over CONUS.

NMM5-conus Design

NOAA Forecast Systems Laboratory is conducting a high resolution NWP forecast experiment during the spring season:

Horizontal grid spacing: 5km
Vertical levels: 38
Domain Size: CONUS
Forecast period: 48 hours
Cycle: 00 UTC
Initial Conditions: ETA 212 (40 km)
Model: WRF-NMM
Parameterizations:
- Microphysics: Ferrier
- Radiation: MYJ
- Planetary Boundary Layer: Eta
- Land Surface Model: NOAH
- Cumulus: none

Objectives

1) Evaluate the performance of high-resolution NWP forecasts over a large domain during spring.

2) Contrast and compare forecasts for a local area (Florida) computed by the same model using two vastly different domains sizes.

Verification

Evaluation of the NMM5-CONUS forecasts will be done objectively through the computation of verification statistics using the WRF Verification System developed by NCEP and FSL.

Computer Resources

NMM5- CONUS is run on NOAA FSL's massively parallel supercomputers.

The Standard Initialization

(SI) is run on a single 2.2 GHz Pentium IV Xeon I/O node with 2GB RAM and a GigE connection to disk storage. The processing time is approximately 5 minutes for gribprep, 17 minutes for hinterp and 90 minutes for vinterp, adding up to a SI total of approximately112 minutes.

The WRF-NMM real is run on a single 3.2Ghz EM64T (64-bit Intel Xeon)processor with 16GB RAM.The processing time is around 45 minutes.

The WRF-NNM model is run on a total of 529 Pentium-4 xeon 2.2 GHz processors with 1 GByte memory each. No overlap of I/O and computations is used. The processing time is typically 5h 40 minutes.