Tuesday, 25 January 2011: 12:00 AM
613/614 (Washington State Convention Center)
David E. Rudack, NOAA/NWS, Silver Spring, MD; and D. P. Ruth

Over the past two decades, there have been numerous publications addressing the need to quantify the underlying uncertainty associated with weather forecasts. One approach discussed in the literature – so-called kernel density estimation (KDE) - involves fitting ensemble data to a probability density function (PDF) which can then be used to quantify the likelihood of a deterministic event in probabilistic terms. Glahn et al. (2009) show that applying a model output statistics (MOS) approach to the PDF will, on average, yield a well-calibrated representation of the true probability space. This method has been applied with good success to the sensible weather elements such as spot temperature and dewpoint as well as maximum daytime and minimum nighttime temperatures. While this approach quantifies the uncertainty of MOS forecasts, the MOS forecast uncertainty is associated with the low-resolution Global Ensemble Forecasting System (GEFS) forecasts and does not quantify the uncertainty associated with the operational Global Forecasting Model (GFS) MOS forecasts derived from the high-resolution GFS. To address the latter, the Meteorological Development Laboratory (MDL) has developed a prototype product that conveys uncertainty information associated with 3-hour GFS MOS spot temperature forecasts for the likelihood that the MOS operational temperature forecasts will (1) not depart by more than +5 degrees F, (2) depart by more than +5 degrees F (too warm), and (3) depart by more than -5 degrees F (too cool). Best categorical forecasts are also generated from these discrete categories.

This paper discusses the methodology used to develop this uncertainty forecast product. Verification results of uncertainty forecasts are also presented for both probabilistic and best category forecasts. Finally, a graphical approach using GFS MOS forecast climatologies in concert with GFS MOS uncertainty forecasts is introduced to demonstrate how this product can be used as a tool to alert forecasters where and when guidance forecasts may need to be modified.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner