About WPC's Probabilistic and
Percentile QPF Products

Read the official Product Description Document (PDD) for this guidance
(Note: The PDD is a pdf file and requires Adobe Acrobat Reader to view)
 

The WPC produces 6-hour quantitative precipitation forecasts (QPFs) for forecast projection days one through three at 6-hour intervals (72-hour duration). Probabilistic forecasts are computed based on a combination of WPC's 6-hour QPF and an ensemble of model forecasts. The ensemble forecasts provide uncertainty information about the QPF which is then used to construct a probability distribution about the WPC QPF. This distribution is utilized to generate probabilistic forecasts of precipitation. The 6-hour QPFs are summed to obtain 24-h QPFs, which are the basis for 24-h probabilistic QPFs generated using the same multi-model ensemble and the same method as for the 6-h probabilistic QPFs.

Ensemble Membership

Number Model (Cycle - For Day Shift) Notes
10 NCEP GEFS ensemble members (06Z) Randomly Selected
25 ECMWF ensemble members (00Z) Randomly Selected
10 Canadian ensemble members (00Z) Perturbations 1-10
1 NCEP Short-Range Ensemble Forecast (SREF) NMMB member (09Z)
2 NCEP High-Resolution Rapid Refresh (HRRR) runs (06Z and 12Z) Through Day 1.5, SREF NMMB Member and GEFS for Days 1.5-3
2 Canadian Regional Model runs (00Z and 06Z) Through Day 2.5, CMCE members for rest of Day 3
2 NCEP North American Mesoscale 3km CONUS Nest runs (06Z and 12Z) Through Day 2, Parent NAM and SREF NMMB member used for Day 3
4 NCEP Hi-res WRF ARW runs (12Z) Through Day 1.5, Time-lagged GFS, GEFS, CMCE, and SREF NMMB members used for Days 1.5-3
1 NCEP Hi-res FV3LAM (12Z) Through Day 2, Time-lagged GFS used for Days 3
1 NCEP Global Forecast System (GFS) (12Z)
1 ECMWF Deterministic (00Z)
1 NCEP Global Ensemble Forecast System (GEFS) Mean (06Z)
1 WPC Forecast
 
A binormal probability distribution (density) function (PDF), which allows skewness, is constructed such that the mode is the WPC QPF and the variance is that of the ensemble. The skewness is based on the position of the WPC QPF in the ensemble distribution. This approach to estimating the three parameters for the binormal PDF is a variation on the method of moments.
 
The probabilistic QPF forecasts provide information in two different forms:
  1. Probabilities of exceeding a threshold show filled contour levels of probability that the 6- or 24-hour accumulation of precipitation will equal or exceed the given threshold. As an example, consider the .50 inch threshold. If a point of interest falls within the 40% contour on the probability map, then the chance of precipitation exceeding .50 inch is 40% or greater. As the threshold values increase, the probabilities of exceeding them decrease.
  2. Percentile accumulations show filled contour levels of precipitation amount associated with a given percentile in the distribution. The percentile value is the percent chance of precipitation accumulating less than the depicted amount. From the opposite perspective, 100 minus the percentile is the chance of precipitation exceeding the depicted amount. For example, there is a 25% chance of precipitation accumulating less than the amounts shown on the 25th percentile accumulation map; while, there is a 75% chance that precipitation will exceed the 25th percentile accumulations. Thus, lower percentile values are associated with smaller accumulations than are higher percentile values.
Both forms of information are presented on the WPC probabilistic QPF web page. The products described under 1 above are found under the "Probability of Precipitation of at Least a Specific Amount" tab on web page. The products described under 2 above are found under the "Precipitation Amount by Percentile" tab on the web page.