Analysis of the Top Significant Fields in Analog-Based
Severe Probability Guidance (23)
Kyle Perez, Saint
Louis University, St. Louis, MO
Alexander Elmore, Saint
Louis University
Chad Gravelle, NOAA/NWS
Operations Proving Ground, CIMSS - University of Oklahoma, Norman, OK
Charles Graves, Saint
Louis University
Abstract:
Previous research has
demonstrated the effectiveness of using historical analogs to aidin forecasting
impactful weather in both the short- and long-term forecast period. Fornearly
the past decade, the Cooperative Institute for Precipitation Systems (CIPS)
fromSaint Louis University has been utilizing historical analogs to provide
guidance forforecasters. Today, current research is focusing on investigating
the usefulness andreliability of severe weather probabilistic guidance for the
extended forecast (i.e., 3-8days). For this new method, the Global Ensemble
Forecast System (GEFS) reforecastdata set is used to find similar environments
to GEFS mean forecasts. Once the bestmatching analogs are found in the
reforecast data set, a logistic regression model is fitusing environmental
fields from the GEFS (i.e., the predictors) and the severe weatherreports
(i.e., the predictand). The output of this model represents the probability
ofsevere weather within 110 km of a grid point when applied to the current
forecast. Thispresentation will focus on the analysis of the top significant
fields found by the logisticregression to discriminate between a severe from a
non-severe environments usingvarious past severe weather events. The events
used were based on collaborative workwith various National Weather Service
Weather Forecasting Offices. The analysis of thetop fields will hopefully
provide forecasters insight into the performance of theAnalog-Based Severe
Probability Guidance, leading to an increase in the level ofconfidence in using
the product in National Weather Service operations.