Application of a Neighborhood POP Forecasting Approach to a County Warning Area

A neighborhood approach for forecasting probabilities of precipitation (POPs) was tested at four sites in the High Plains Region. The neighborhood approach used ensemble data from the 2007 and 2008 Hazardous Weather Testbed Spring Experiments, and created static POP tables by post-processing quantitative precipitation forecasts (QPFs). Considering Brier scores for a one year period from April 2012 to April 2013, the neighborhood approach's POPs were competitive with those of Model Output Statistics (MOS) for 12 hour forecast periods through 60 hours, after which MOS performed better than the neighborhood approach. The results showed that a relatively simple neighborhood approach based on static POP tables can be competitive with the dynamic and more sophisticated MOS, and that training the QPF-POP relationship past 60 hours may make the neighborhood approach perform better in the later forecast periods.

This presentation will include a description of the neighborhood approach, a summary of the results for April 2012 through April 2013, and some ideas concerning how the neighborhood approach might be modified to improve forecasts beyond 60 hours.