Strike Probabilities: Instructions

[ Table Entries | Example | Simplifying Assumptions | Direct Access ]


Warning: The tropical cyclone data presented here are intended to convey only general information on current storms and should not be used to make life or death decisions: the data may not be accurate. If you are in the path of a storm you should be listening to official information sources.


Strike Probabilities: Explanation of Table Entries

Name:
The name of the storm.

Date:
The date of the forecast (Universal Time = Greenwich Mean TIme).

Time:
The time of the forecast (Universal Time).

Relative Time:
The Relative time helps with the conversion frum UT to local time.
Now column:
Time (in hours) from "now" (the time stamp at the top of the table) to the forecast time in column 3. This is the time from when the table was generated to the forecast time. This may not be from the current time since the tables are archived until a new forecast is received. The time interval should be within a few hours of the interval from the current time, however.
Fcst column:
Time (in hours) from the time the forecast was made to the time of the forecast in column 3. Longer extrapolations are less accurate

Probabilities:
T-column
Probability (%) at the time shown (the probability at an instant of Time). This number is not very useful: the probability during some interval of time is more important.
I-column
Cumulative probability (%) during the Interval from the previous time shown to the time shown. For the first row, the cumulative probability during the interval from the time of the last known storm position to the forecast time shown.
C-column
Cumulative probability (%) from the time of the last known storm position to the time shown. The final, highlighted entry for each distance gives the total strike probability during the entire forecast period.

Wind:
The forecast maximum sustained wind speed, in knots, at the time shown for the row. In most cases, the maximum wind will be just outside the eye of the storm.

Note:
nmi = nautical miles.
60 nmi = 1 degree of latitude and 1 nmi is 1.151 statute miles or 1.852 kilometers.
1 knot is 1.151 mph or 1.852 kph.

Strike Probabilities: Example

Tropical cyclones (hurricanes, typhoons, etc.) are violent storms capable of inflicting great destruction. Meteorologists have learned to forecast these storms with superb accuracy. However, no forecast is perfect, and one way to aid interpretation of the forecasts is to compute the probability that a storm will strike a particular location.

The following table shows an example of the strike probability calculation for Hamilton, Bermuda during hurricane Erika on September 8, 1997.

Tropical Cyclone Strike Probabilities for Hamilton, Bermuda: 32°22´N 64°40´W

Probability that the storm center will be within 60 nmi, 120 nmi, and 240 nmi of Hamilton, Bermuda: 32°22´N 64°40´W. Probabilities at intermediate times may be higher than those shown. The probabilities are given in percent ("<1" means "less than one percent probability") and are updated whenever a new forecast is received.

NameDateTimeRelative Time60 nmi120 nmi240 nmiWind
NowFcstTICTICTIC
ERIKA 1997-09-09 18:00 UT +33 h +33 h <1 <1 <1 <1 <1 <1 6 6 6 090
ERIKA 1997-09-10 00:00 UT +39 h +39 h <1 <1 <1 1 1 1 26 28 28 097
ERIKA 1997-09-10 06:00 UT +45 h +45 h 1 1 1 8 9 9 46 51 53 105
ERIKA 1997-09-10 18:00 UT +57 h +57 h 3 5 5 15 21 21 53 69 76 097
ERIKA 1997-09-11 06:00 UT +69 h +69 h 2 6 8 10 20 27 38 56 80 090

The above Erika probabilities were based on forecast data from 1997-09-08 09:00 UT.

The first number to examine is the highlighted number in the last row of the "C" columns. These numbers give the cumulative probability of a strike over the entire forecat period (in this case, 1997-09-08 09:00 UT through 1997-09-11 06:00 UT). The table of probabilities tells us that there is a very good chance that Erika will come within 240 nautical miles of Hamilton (80 percent) during the forecast period, but a relatively small chance that Erika will come within 60 nautical miles (8 percent) at some time during the forecast period. It is therefore likely, but not certain, that Hamilton will be spared a direct hit.

Note that an 80 percent probability means that a strike is about 4 time more likely to happen than to not happen [80/(100-80)=4]. Similarly an 8 percent probability means that a stike is about 12 times more likely not to happen than to happen.

Looking at the "I" columns which show the cumulative probabilities from one row to the next, the table also tells us that the storm's closest approach to Hamilton is most likely to occur sometime on September 10 (universal Time), when the probabilities in the "I" columns peak.

The zero hour difference between the "Now" column and the "fcst" column indicates that the forecast was 0 hours old when the strike probability calculation was run. Had the calculation been run 3 hours after the forecast, for example, the numbers in the "now" column would have been about 3 hours less than the numbers in the "fcst" column. In the example above, the now column shows that the time of closest approach is likely to be 40 to 60 hours away. This helps with the conversion from UT to local time.

The forecast winds at the time of closest approach are in the 90 to 100 knot range. So this is a strong storm which should be monitored carefully.

The following table shows the forecast data used to calculate the above probabilities:

Name Date Time (UT) Lat (degrees) Lon (degrees) Course (true) Speed (knots) Pressure (mb) Wind (knots) Gusts (knots) Type Actual / Forecast WMO
ERIKA 1997-09-08 09:00 22.3N 63.2W 5 6 955 100 ??? HUR ACT WTNT21
ERIKA 1997-09-08 18:00 23.2N 63.1W ??? ??? ??? 105 ??? ??? FOR WTNT21
ERIKA 1997-09-09 06:00 25.0N 63.0W ??? ??? ??? 105 ??? ??? FOR WTNT21
ERIKA 1997-09-09 18:00 27.0N 62.7W ??? ??? ??? 090 ??? ??? FOR WTNT21
ERIKA 1997-09-10 06:00 29.3N 62.1W ??? ??? ??? 105 ??? ??? FOR WTNT41
ERIKA 1997-09-11 06:00 33.5N 60.5W ??? ??? ??? 090 ??? ??? FOR WTNT41

The rows at 1997-09-10 00:00 UT and 1997-09-10 18:00 UT in the probability table were interpolated from the above forecast data. Hopefully, this example has made it clear how the strike probability calculation can help interpret the above forecast data.

In fact, Erika's closest approach to Hamilton was about 300 nmi around 1997-09-10 03:00 UT: Erika sped up and veered away from the forcast track shown here.


Simplifying Assumptions

The following simplifying assumptions are made in the strike probability calculations.
  1. The errors in the forecast positions are assumed to be normally distributed.
  2. The errors in the forecast positions are assumed to be isotropic.
  3. The errors in the forecast positions are assumed to increase with time as the storm position is extrapolated further into the future.
  4. The forecasts are assumed to be closely spaced in time (though the code does interpolate between forecasts).
  5. Plane geometry is assumed.
  6. No account is made for the land masses on the Earth's surface.


Direct Access

If you know the latitude and longitude of your location, you can make your own list of locations using a URL like http://www.solar.ifa.hawaii.edu/cgi-bin/StrikeProb?latitude=xx.xx&longitude=xx.xx&location=City+Name where you replace the xx.xx with the actual latitude and longitude in the format dd.mm (dd is degrees, mm is minutes; use a minus sign for south latitude and east longitude). The location is any text with a plus sign to represent a space. For example,
http://www.solar.ifa.hawaii.edu/cgi-bin/StrikeProb?latitude=+31.33&longitude=-130.33&location=Kagoshima,+Japan gives the probabilities for Kagoshima, Japan.


Important information: these data may not be entirely accurate.
Description of the strike probabilty calculation (168 kB PostScript file)

Home Back


Comments to: metcalf@akala.ifa.hawaii.edu