About the Texas Tech Real Time Weather Prediction System
Texas Tech University operates both a deterministic and ensemble weather prediction system to provide forecast guidance at convection-allowing scales. While we strive to provide basic deterministic and probabilistic output products in both systems that add value to forecasts of high impact events (such as a variety of convective parameters, red flag threat index for fire weather, and aviation products), our focus is to develop advanced ensemble tools that can be transitioned to operations within the National Weather Service. Ideally these tools accompany the development of an operational National Weather Service convection-allowing ensemble system. Some of these techniques involve best member approaches to pick the most likely ensemble members (Ancell 2013, Hollan and Ancell 2015), ensemble sensitivity analysis (Bednarczyk and Ancell 2014, Hill et al. 2016), sensitivity-based ensemble subsetting to improve forecast probabilities (Ancell 2016), and sensitivity-based observation targeting. Much of the research dedicated to developing these techniques has been supported by the NOAA Collaborative Science, Technology, and Applied Research (CSTAR) program. The system is designed and maintained by Brian Ancell of the Texas Tech Department of Geosciences (firstname.lastname@example.org), and the system website is designed and maintained by Mark Conder of the National Weather Service Lubbock forecast office (email@example.com) - please direct any questions to the appropriate personnel.
The Deterministic System
The deterministic system utilizes the WRF-ARW Version 3.5.1 (Skamarock et al. 2008) to create forecasts on a nested grid configuration. An outer 12-km grid encompasses the southwest U.S. and the southern and central plains, while a nested 3-km grid exists over Texas, Oklahoma, Kansas and portions of surrounding states. The 3-km grid receives its lateral boundary conditions from the 12-km grid, and the 12-km grid is forced on the lateral boundaries by the Global Forecast System (GFS). The initial conditions on both the 12-km and 3-km grids are taken from the GFS analysis. In turn, a spin up process occurs in the first hours of the WRF forecasts to achieve high-resolution structure. Both forecasts are run to 60-hr forecast time from four initializations daily at 00, 06, 12, and 18 UTC. Both domains use the Noah land surface model, Thompson microphysics, the YSU planetary boundary layer scheme, and Dudhia shortwave radiation and RRTM longwave radiation schemes. The 12-km grid uses the Tiedtke cumulus parameterization while no cumulus parameterization is used on the 3-km domain. Both grids possess 38 vertical levels. The system runs on 70 computing cores and executes the 60-hr forecast in roughly 4.5 hours.
The Ensemble System
The ensemble system utilizes the WRF-ARW Version 3.5.1 to create forecasts on a nested grid configuration. A 12-km grid exists over the entire U.S. and portions of Canada and Mexico, while a nested 4-km grid exists over a region from the Rocky to Appalachian mountains, and from Mexico to South Dakota. This 42-member ensemble is both a data assimilation and forecasting system, creating its initial conditions through an ensemble Kalman filter assimilation procedure using the National Center for Atmospheric Research Data Assimilation Research Testbed (DART; Anderson et al. 2009) system. Assimilation occurs on only the 12-km domain, and subsequent forecasts are initialized on both the 12-km and 4-km grid from that assimilation procedure. The system assimilates cloud-track winds, ACARS aircraft winds and temperatures, radiosonde winds, temperatures, and dew points, and surface winds, temperatures, pressures, and dew points on a 6-hr assimilation cycle. Extended 48-hr forecasts are produced on both the 12-km and 4-km grids twice daily from the 00 and 12 UTC initializations. Both grids possess 38 vertical levels. The ensemble members on the 4-km grid receive their lateral boundary conditions from the 12-km forecast members, while the 12-km members are forced on the lateral boundaries by the previous two cycles of the Global Ensemble Forecast System (GEFS). The physics used in all ensemble forecasts are the same as those in the deterministic system. The system runs on 698 computing cores and executes the 48-hr ensemble forecast in roughly 9 hours.
Ancell, B.C., 2013: Nonlinear Characteristics of Ensemble Perturbation Evolution and Their Application to Forecasting High-Impact Events. Weather and Forecasting, Vol. 28, No. 6, pages 1353-1365.
Ancell, B.C., 2016: Improving High-Impact Forecasts through Sensitivity-Based Ensemble Subsets: Demonstration and Initial Tests. Weather and Forecasting, Vol. 31, No. 3, pages 1019-1036.
Anderson, J., T. Hoar, K. Raeder, H. Liu, N. Collins, R. Torn, and A. Avellano, 2009: The
Data Assimilation Research Testbed: A Community Data Assimilation Facility. Bulletin of the American Meteorological Society, Vol. 90, pages 1283-1296.
Bednarczyk, C.N., and B.C. Ancell, 2015: Ensemble Sensitivity Analysis Applied to a Southern Plains Convective Event. Monthly Weather Review, Vol. 143, No. 1, pages 230-249.
Hill, A.J., C.C. Weiss, and B.C. Ancell, 2016: Ensemble Sensitivity Analysis for Mesoscale Forecasts of Dryline Convection Initiation. Monthly Weather Review, accepted, in press.
Hollan, M.A., and B.C. Ancell, 2015: Ensemble Mean Storm-Scale Performance in the Presence of Nonlinearity. Monthly Weather Review, Vol. 143, No. 12, pages 5115-5133.
Skamarock, W.C., J.B. Klemp, J. Dudhia, D.O. Gill, D.M. Barker, M. Duda, X.-Y. Huang, W. Wang and J.G. Powers, 2008: A Description of the Advanced Research WRF Version 3. NCAR Technical Note TN-475.