11th AMS Conference on Numerical Weather Prediction
Norfolk, Virginia, 19-23 August 1996, in press
SHORT-RANGE ENSEMBLE FORECASTING PILOT PROJECT: A STATUS REPORT
Harold E. Brooks1, David J. Stensrud1, M. Steven Tracton2, Eric Rogers2,3
1NOAA/ERL/National Severe Storms Laboratory
2NOAA/NWS/National Centers for Environmental Prediction
Camp Springs, Maryland
3General Sciences Corporation
In the summer of 1994, the National Meteorological Center (NMC), now the National Centers for Environmental Prediction (NCEP), and the National Severe Storms Laboratory (NSSL) hosted a workshop on short-range ensemble forecasting (SREF) (Brooks et al. 1995). One outcome of that workshop was the design of a one-year pilot project designed to test the operational uses of ensemble forecasting on the 0-48 hour time range. The final implementation of the project differs slightly from that described in Brooks et al. (1995). The primary difference is that, in addition to the suite of 10 members from the 80-km horizontal resolution Eta model (Black 1994), 5 members from the regional spectral model (RSM) (Juang and Kanamitsu 1994) are also run. Also, the ensemble ended up being typically scheduled for initialization at 1200 UTC on Tuesdays, rather on a date selected each week. (The schedule was altered due to computer and staff availability at the appropriate times.) Here, we would like to report briefly on the status of the pilot project.
2. PROGRESS TO DATE
The ensemble has now been run more than 30 times, yielding a data set that spans most of a year (Table 1). Data from the model simulations, as well as corresponding 29-km "meso-Eta" forecasts are available for downloading from NSSL via
nssl.noaa.gov/mosaic_files/sref.html>, along with software to view the data in x-windows environments. Generally, six-hourly data files of winds, height, tem- perature, etc., at every 50 hPa, as well as forecasts of parameters such as CAPE, Lifted Index, and categorical precipitation type are available.
Case studies of individual forecasts (e.g., Brooks et. al. 1996a,b, Cohen et al. 1996) have been done and indicate that, at least in some cases, the ensemble can offer significant value to forecasters. Statistical analy- ses of the larger data set (Hamill and Colucci 1996a,b, Hamill et al. 1996, Stensrud et al. 1996) show that, although the mean and median ensemble forecasts exhibit less error than corresponding meso-Eta forecasts for many variables, the dispersion of the ensemble is less than ideal. Most likely, this is due to the lack of spread in the initial conditions. Techniques for increa-sing the spread are under investigation by groups using other models or initialization techniques for SREF (e.g., Manikin 1995, Manikin and Ramamurthy 1996, Du et al. 1996, Stensrud et al. 1996b). These efforts typically use larger ensembles, but on fewer cases.
Cohen et al. (1996) have focused on the problem of providing information to forecasters from the ensemble. Because of the number of cases involved, the pilot project provides an ideal ground for testing methodologies across a range of conditions and forecasting problems.
The pilot project has succeeded in its goals of showing potential value of operational SREF and in identifying areas where research efforts need to be focused. First, work needs to be done on the problem of developing initial perturbations that span the range of possibilities realistically. The current technique, using differing analyses prepared by NCEP and some members of the medium-range ensemble breeding modes, does not provide a wide enough range. Secondly, the issue of presenting information to forecasters in a useful way is important. Even a 10
member ensemble provides a tremendous amount of data and, presumably, it is important to present information about some of the outliers in the ensemble to help forecasters in dealing with rare, severe events. The challenge of keeping track of all of the members of an ensemble and identifying important features quickly is formidable and requires much work. We invite others to join us in this effort and to utilize this rich source of information about numerical models.
Drs. Henry Juang and Zoltan Toth (NCEP) did the development to produce the RSE ensemble generation technique. Danny Mitchell (NSSL) developed the xgribview software and archival work. We also thank Drs. Geoff DiMego and Eugenia Kalnay (NCEP) for his support during the project.
Black, T., 1994: The new NMC mesoscale Eta model: Description and forecast examples. Wea. Forecasting, 9, 265-278.
Brooks, H. E., M. S. Tracton, D. J. Stensrud, G. DiMego, and Z. Toth, 1995: Short-range ensemble forecasting: Report from a workshop (25-27 July 1994). Bull. Amer. Meteor. Soc., 76, 1617-1624.
_____, J. V. Cortinas Jr., P. R. Janish, and D. J. Stensrud, 1996a: Application of short-range numerical ensembles to the forecasting of hazardous winter weather. Preprints, 15th Conf. on Weather Analysis and Forecasting, Amer. Meteor. Soc., Norfolk, Virginia, this volume.
_____, D. J. Stensrud, and C. A. Doswell III, 1996b: Application of short-range NWP model ensembles to severe storm forecasting. Preprints, 18th Conf. on Severe Local Storms, Amer. Meteor. Soc., San Francisco, Californiaa, 372-375.
Cohen, A. S., T. M. Hamill, and S. J. Colucci, 1996: Evaluating Eta/RSM ensemble forecasts and display methods during selected heavy precipitation events of 1995-96. Preprints, 15th Conf. on Weather Analysis and Forecasting, Amer. Meteor. Soc., Norfolk, Virginia, this volume.
Hamill, T. M., and S. J. Colucci, 1996a: Verificaiton of Eta/RSM short-range ensemble forecasts. Mon. Wea. Rev., submitted.
_____, and S. J. Colucci, 1996b: Random and systemation error in NMC's short-range Eta ensembles. Preprints, 13th Conf. on Probability and Statistics, Amer. Meteor. Soc., San Francisco, California, 51-56.
_____, S. J. Colucci, and A. S. Cohen, 1996: Eta/RSM ensemble usefulness for short-range forecasts: Dependence on weather regime, model, and perturbation methodology. Preprints, 15th Conf. on Weather Analysis and Forecasting, Amer. Meteor. Soc., Norfolk, Virginia, this volume.
Juang, H.-M., and M. Kanamitsu, 1994: The NMC nested regional spectral model. Mon. Wea. Rev., 122, 3-26.
Manikin, G. S., 1995: Short range ensemble forecasting. M. S. Thesis, Dept. of Atmospheric Sciences, University of Illinois at Urbana-Champaign, 100 pp.
_____, and M. K. Ramamurthy, 1996: Short-range Eta model ensemble forecasting. Preprints, 11th Conf. on Numerical Weather Prediction, Amer. Meteor. Soc., Norfolk, Virginia, this volume.
Du, J., S. L. Mullen, and F. Sanders, 1996: Short-range ensemble forecasting (SREF) of quantitative precipitation. Mon. Wea. Rev., submitted. (Short version in this volume.)
Stensrud, D.J., H. E. Brooks, and D. V. Mitchell, 1996a: Validation of the short-range ensemble forecasts in upper levels. Preprints, 11th Conf. on Numerical Weather Prediction, Amer. Meteor. Soc., Norfolk, Virginia, this volume.
_____, D. J., J. W. Bao, Y. Zheng, and T. T. Warner, 1996b: Ensemble forecasting of convective weather events. Preprints, 11th Conf. on Numerical Weather Prediction, Amer. Meteor. Soc., Norfolk, Virginia, this volume.
|5 Jan. 1995||No||No||Test case|
|9 May 1995||Yes||Yes||Severe weather in eastern US (Brooks et al. 1996b)|
|20 May 1995||Yes||Yes||Flooding at end of forecast in North Dakota|
|30 May 1995||Yes||Yes||Severe weather in Texas|
|6 June 1995||Yes||Yes|
|13 June 1995||Yes||Yes|
|20 June 1995||Yes||Yes||Rapidan River, Virginia flash flood|
|27 June 1995||Yes||Yes|
|3 July 1995||Yes||Yes|
|18 July 1995||Yes||Yes|
|25 July 1995||Yes||Yes|
|1 Aug. 1995||Yes||No||Hurricanes Dean and Erin (Cohen et al. 1996)|
|7 Aug.1995||Yes||No||Flooding in Ohio|
|14 Aug. 1995||Yes||Yes|
|21 Aug. 1995||Yes||Yes|
|29 Aug. 1995||Yes||Yes||Tropical Storm Jerry in Florida|
|12 Sep. 1995||Yes||No|
|18 Sep. 1995||Yes||No|
|25 Sep. 1995||Yes||Yes||Hurricane Ishmael flooding in New Mexico|
|2 Oct. 1995||Yes||No||Hurricane Opal in Gulf of Mexico|
|23 Oct. 1995||Yes||Yes||Heavy snow in northern Plains|
|8 Nov. 1995||Yes||Yes||Lake effect snows|
|13 Nov. 1995||Yes||No||Heavy snow in Northeast|
|20 Nov. 1995||Yes||No|
|27 Nov. 1995||Yes||No||Heavy snow in several areas|
|11 Dec. 1995||Yes||Yes||Snows in Midwest, lake effect snows|
|18 Dec. 1995||Yes||Yes||Heavy snow NE US, rain southern US (Cohen et al. 1996)|
|26 Dec. 1995||Yes||Yes|
|11 Jan. 1996||Yes||No||Mid-Atlantic snowstorm|
|23 Jan. 1996||Yes||Yes|
|31 Jan. 1996||Yes||Yes||Arctic outbreak, major winter storm in South (Brooks et al. 1996a)|
|7 Feb. 1996||Yes||Yes||Northwestern US heavy rains|
|8 Apr. 1996||Yes||No|
Table 1: List of dates for which short-range ensemble was run , beginning at 1200 UTC, and archived at NSSL as of date of writing of manuscript. Yes or no indicates whether RSM or meso-Eta model data are available in the archive. Selected events or particular interest and relevant references included. Note that lack of availaility of meso
Eta data in the NSSL archive does not mean that the meso-Eta was not run on that day.