Use of a historical analog-based winter storm guidance package for forecasting a central New York snow event Mike Evans
Motivation Models are often better at forecasting large scale patterns than quantitative precipitation. Forecasters often make forecasts by examining model forecasts of large-scale patterns, then evaluating qpf based on experience. Based on my experience, what type of precipitation pattern does this type of storm produce?. Does the model QPF look reasonable?. This type of methodology should best be employed by forecasters with the most experience.
Motivation Forecast skill is largely determined by experience. The relative advantage of highly experienced forecasters is secured by virtue of the larger set of cases from which they many draw upon. - Roebber and Bosart, 1996
Motivation continued .There were major variations in the experience and perceived abilities of the Phoenix teams, generally producing the expected variations in performance. Teams composed of more senior and capable forecasters produced the greatest improvements (over guidance) - McCarthy, Purcell and Ball, 2007
The problem. I cant remember what happened yesterday, let alone what happened with some storm from 5 years ago - An unidentified lead forecaster with many years of experience
The Saint Louis University Historical Analog- Based Winter Weather Guidance System Aids forecasters by providing easy access to a data set of winter storm events that are similar to upcoming, forecast events. Instant Experience
The Saint Louis University Historical Analog- Based Winter Weather Guidance System Ingests current GFS forecasts of large-scale forecast patterns Searches the NARR data base to find the best historical analogs to the current forecast Returns information on the analogs Mean fields, and probabilistic information is included Detailed information on the top 15 analogs is included
Parameters that determination the top 15 analogs Parameters are determined within a user-selectable domain (2 east-coast domains, 2 mid-west domains). 300 and 500 and 850 hPa heights. 850 hPa temperature and theta-e advection. 850 and 700 hPa frontogenesis. Surface temperature and sea-level pressure. Precipitable water. Storm track is included in the calculation. For details, see: http://intra.crh.noaa.gov/metdat/ssd/2008-12-16 13.01_SLU_Analog_Seminar.wmv. http://intra.crh.noaa.gov/metdat/ssd/2008-12-16 13.01_SLU_Analog_Seminar.wmv.
Select a model run for comparison
15 analogs are returned with information on the analogs available
Example December 31, 2008
Sea-level pressure and 1000-500 mb thickness December 31, 2008
700 hPa heights and frontogenesis
Vertical motion, temperature and NAM QPF December 31, 2008
How much snow have similar systems historically produced for our area?
Mean 500 hPa heights and sea-level pressure from the 15 analogs
Probabilistic snowfall information from the 15 analogs
Best analog February 26, 1994
Best analog observed snowfall
2nd best analog January 13, 2000
2 nd best analog observed snowfall
Observed snowfall December 31, 2008
Conclusion Probabilistic information from the analogs indicated a high probability for greater than 2 inches of snow across central NY, with lower probabilities of 4 and 6 inches. The best analogs indicated that a west-east band of 6 to 10 inches would be possible across central New York.
In this case, the guidance system Provided instant experience to the forecaster for these types of systems. Allowed forecasters to assess the reasonability of the model QPF, and corresponding snowfall amounts. Other cases showed similar results. Still waiting to assess the system for a major noreaster.