Experimental Graphical Tropical Weather Outlook
Updated 17 March 2008
Contents
- Overview
- Changes for 2008
- Explanation of the Graphical TWO
- Probabilistic Genesis Forecasts
- Methodology
- Availability
- Examples
- Potential Benefits of a Graphical TWO
- Feedback and Suggestions from 2007
- References
4. Probabilistic Genesis Forecasts
While not a part of the formal experimental GTWO product, NHC also tested in-house probabilistic TC genesis forecasts during
the 2007 hurricane season. The purpose of this exercise was to determine if there was sufficient accuracy in forecasting tropical
cyclogenesis and whether such a forecast could be used as the basis for categorical genesis forecasts in later versions of the
GTWO.
NHC forecasters subjectively assigned a probability of genesis (0 to 100%, in 10% increments) to each area of disturbed weather
mentioned in the text TWO. The assigned probabilities represented the forecaster's assessment of the chance that TC formation
would occur during the ensuing 48-h period. These experimental probabilistic forecasts remained "in-house" and were not released
to the public. The genesis forecasts were verified based on TC development from the final NHC best-track data. A more detailed
description of the probabilistic genesis forecast and its verification is presented in Brown et al. (2008)*.
The results of the probabilistic genesis forecast verification suggest that NHC has skill at categorizing disturbances into three
bins based on their potential for genesis (low, medium, and high (Tables 1 and 2). That is, the verification indicates that it
is possible to distinguish between three broad categories of genesis likelihood in both the Atlantic and eastern North Pacific
basins. For example, disturbances with a low probability underwent genesis on average less than 10% of the time in both basins,
while disturbances in the high category underwent genesis greater than 60% of the time on average in both basins. These results
were used as the basis for development of a three-tiered color-coded genesis probability forecast that will be included in the GTWO
starting in 2008.
|
Expected % |
Verifying % |
Number of Forecasts |
0-10% (Low) |
5 |
3 |
389 |
20-50% (Med) |
28 |
18 |
263 |
60-100% (High) |
71 |
66 |
53 |
Table 1. Binned Forecast Reliability Table for the
Atlantic basin. The expected percentage is an average of all the forecast
probabilities within the bin.
|
Expected % |
Verifying % |
Number of Forecasts |
0-10% (Low) |
6 |
6 |
179 |
20-50% (Med) |
26 |
47 |
162 |
60-100% (High) |
70 |
90 |
29 |
Table 2. Binned Forecast Reliability Table for the
eastern North Pacific basin. The expected percentage is an average of all the forecast probabilities
within the bin.
Next: Methodology
|