Extended Abstract (224K)2.10
The Critical Success Index and warning strategy
Harry Gerapetritis, NOAA/NWS, Greer, SC; and J. M. Pelissier
The National Oceanic and Atmospheric Administration’s (NOAA) National Weather Service (NWS) has established performance goals for various NWS warning programs expressed in terms of False Alarm Ratio (FAR), Probability of Detection (POD), and Lead Time statistics. Improvements in these statistical measures can be achieved in at least two ways: 1) increasing forecast skill by the infusion of scientific advancements, and 2) adopting warning strategies that specifically target deficiencies in either FAR or POD. The latter involves examining FAR and POD statistics to assess “over-warning” or “under-warning” of winter storms, flash floods, and other phenomena. If, for example, the FAR for winter storm warnings tends to be high, it may suggest that the guidelines that govern the issuance of such warnings should dictate a greater level of forecaster confidence that the event will occur before a decision to issue a warning is made. A difficulty inherent in such targeted warning strategies is that attempts to improve FAR may worsen POD, and vice-versa. Although the Critical Success Index (CSI) provides no unique verification information, since it is a function of both FAR and POD, understanding its behavior can help identify the component to target in order to maximize CSI. This study establishes that changes in FAR have a larger (smaller) impact on CSI than equivalent changes in POD when POD is greater (less) than 1 - FAR, and that equivalent changes in FAR and POD have the same impact on CSI when POD=1 - FAR.
Session 2, Forecast Evaluation (Room 3A)
Tuesday, 13 January 2004, 8:30 AM-2:30 PM, Room 3A
Previous paper Next paper