Extended Abstract (400K)JP1.5
Using skill scores to assist assessment of the "Man-MOS-Met Mix" in probability of precipitation (PoP) forecasting
Preston W. Leftwich Jr., NOAA/NWS, Kansas City, MO
Probability of Precipitation (PoP) has been a routine part of public forecasts of the National Weather Service (NWS), and its predecessors, since 1966. Evolving operational numerical models have continued to provide improved guidance for forecasters. Since 1972, Model Output Statistics (MOS) have provided primary objective guidance for PoP forecasts. These tools are utilized along with meteorological diagnosis/analysis to support decisions made by forecasters. As such, the overall PoP forecast process is a mix of forecaster decision, MOS guidance, and other meteorology derived from various sources (e.g., direct model output). That is, the forecast process is a "Man-MOS-Met Mix." Analysis of each of the components of this mix is important in efforts to improve PoP forecasts provided to the public. Additionally, clarification of roles of these components will be valuable in the establishment of a modernized forecast process executed via the Interactive Forecast Preparation System (IFPS).
Performance measures are used to track trends in forecast accuracy. For PoPs, a Brier Score is the commonly used statistic. Also, a MOS-based Skill Score serves as a performance measure by comparison of local office PoPs to those provided by operational MOS guidance, currently that from the operational Nested Grid Model (NGM). Further, a difference between MOS guidance and local PoPs of 20%, or more, is considered notable. In an effort (1) to assist forecasters in improving PoP forecasts issued by local NWS field offices within the Central Region, and (2) to provide direction for localization of the IFPS for future operational use, a procedure has been developed to utilize routine PoP Skill Scores to identify cases for review of both the meteorological situation and the related local forecast process.
Stratification of Skill Scores when MOS and local PoPs differed by 20% or more identified various periods when local PoPs indicated either more or less skill than MOS PoPs. Additional finer stratification led to identification of individual forecasts. For example, 24h PoP forecasts having PoP values differing by 20% or more were identified for each Central Region forecast office during January-March 2000. Reviews of the "Man-MOS-Met Mix" were conducted for each selected case. Results and suggestions for their use by local offices are presented.
Joint Poster Session 1, Ensemble Forecasting and Other Topics in Probability and Statistics (Joint with the 16th Conference on Probability and Statistics in the Atmospheric Sciences and the Symposium onObservations, Data Assimilation,and Probabilistic Prediction)
Wednesday, 16 January 2002, 1:30 PM-3:00 PM
Previous paper Next paper