Welcome to the NOAA Institutional Repository |
Stacks Logo
Advanced Search
Select up to three search categories and corresponding keywords using the fields to the right. Refer to the Help section for more detailed instructions.
Clear All Simple Search
Advanced Search
A threat score statistic for hurricane track forecasts
Filetype[PDF-397.34 KB]

  • Personal Author:
  • Description:
    "I recently had occasion to read a report in which two sets of hurricane track forecasts were compared. The statistic used in the comparison was the distance between the observed and forecast position of the storm center. While it was easy to grasp the idea that a reduction in this statistic was indicative of an improved forecast, it was less obvious that the statistic is by itself a good measure of the utility of the forecast track of the storm. Slow moving storms may be poorly forecast even though the displacement error is small compared to that obtained in a qualitatively better forecast of a fast-moving storm. This fact suggests that the displacement error should be normalized by the smaller distance traveled by either the forecast, or the observed storm center. Such a statistic would be analogous to the S1 score used widely to evaluate numerical model circulation forecasts. But since the hurricane threatens an area around the storm center, it occurred to me that another statistic, widely used in precipitation forecast verification, might find application in validating the utility of a numerical prediction of the track of a hurricane. The statistic I have in mind is the threat score which is simply defined as the ratio of the area correctly predicted to be threatened to the "union" of the area actually influenced and the area predicted to be influenced. In order to be precise one must define the extent of the threatened area around the storm-track. At the present time, the numerical guidance produced by NMC does not indicate the extent of damaging winds and seas, thus for our purposes we have arbitrarily chosen to describe the threatened area to lie with a radius of one degree of latitude from the storm center. To assess the virtues and faults of the threat score statistic, we have applied it to the 1978 Pacific storm FICO and to the NMC hurricane model forecasts made for that storm. In addition we have applied the statistic to the storms used in the comparison test mentioned in the first paragraph"--Introduction.

  • Document Type:
  • Place as Subject:
  • Main Document Checksum:
  • Supporting Files:
    No Additional Files
No Related Documents.
You May Also Like: