The 'wind speed probability' graphical product gives the current likelihood that a specific point on a map will be struck by hurricane (74 mph) and/or by tropical storm strength (39mph) winds during the next 5 days.
The product is designed for clarity and usefulness of information. Its presentation allows decisionmakers to see at a glance what the chance is that a given location will be hit by damaging winds. The product is provided for tropical cyclones worldwide. Forecast probabilities are updated every 6 hours, except for systems in the North Indian and Southern Hemisphere basins where updates occur every 12 hours. The product is developed by modelling the errors in the forecast track and intensity of historical storms.
Each 'wind speed probability' graphic displays the probabilities (in percent) that 1min sustained wind speeds of at least hurricane Cat 1 strength (64kt or 74 mph) and/or 1min sustained wind speeds of at least tropical storm strength (34kt or 39 mph) will occur at each specific point on the map during a given time period out to 120 hours.
The probabilities are shown from 1% to 100% in colourcoded 5% bands. To aid guidance a Table (below) relates the probabilities to simple descriptions that the event will happen. For example, a probability of 50% indicates a medium or evenschance that the event will occur.
Table relating the forecast probability values of being struck by hurricane (74 mph) and/or by tropical storm strength (39mph) winds to an equivalent simple description that the event will happen. 

Separate 'wind speed probability' maps are provided from the current time to each available forecast lead time for where the storm exists. For example, in the North Atlantic separate wind speed probability maps are displayed for the periods 09 hrs lead, 021 hrs lead, 033 hrs lead,...,0117 hrs lead.
The 'wind speed probability' graphical product offers distinct advantages to users compared to other tropical cyclone forecast products. Decisionmakers can now tell at a glance what the current chance is that a location will be hit by damaging tropical cyclone winds. Businesses, industry and the general public can assess better the risk specific to their own location caused by an active tropical cyclone. This quantitative information will help with better preparedness decisions. Insurers and reinsurers may obtain realtime information on the likelihood of potential loss for their portfolios. Humanitarian relief agencies may obtain more useful advance warning on the likelihood of aid being needed at a given location.
Examples of the 'wind speed probability' graphical product are shown below for hurricane Ian (2022). The probability maps are based on the National Hurricane Center advisory numbers 4(issued at 03:00 GMT on September 24th, 2022) and 19 (issued at 21:00 GMT on September 27th, 2022). Ian made landfall as a Cat 4 hurricane at 03:10 GMT on September 28th, 2022 near Punta Gorda.
At 03:00 GMT on September 24th (108 hours before Ian's Florida landfall), Tampa Bay had predicted chances of 45% and 10% of being struck respectively by tropical storm strength and hurricane strength winds (Figure 1). By 21:00 GMT on September 27th (6 hours before landfall), these chances had risen to 90% and 25% respectively (Figure 2). In the event Tampa international airport recorded a peak gust of 65 knots.
Figure 1. Probabilities (in percent) of experiencing 1min sustained wind speeds of at least tropical storm strength (34kt or 39 mph) (left image) and hurricane Cat 1 strength (64kt or 74 mph) (right image) from hurricane Ian during the 120 hours starting at 03:00 GMT on September 24th, 2022.
Figure 2. Probabilities (in percent) of experiencing 1min sustained wind speeds of at least tropical storm strength (34kt or 39 mph) (left image) and hurricane Cat 1 strength (64kt or 74 mph) (right image) from hurricane Ian during the 120 hours starting at 21:00 GMT on September 27th, 2022.
The wind speed probabilities are computed by modelling and combining the errors in the forecast track position with the errors in the forecast extent of winds in each storm quadrant. The modelling is performed separately for each forecast lead time and for each tropical cyclone basin. Errors are computed and modelled from the official forecast advisories issued from 2016 to 2020 inclusive. A latitudelongitude grid resolution of 0.1 x 0.1 degrees (~11km) is employed throughout.
The forecast track errors are obtained by computing, as a function of lead time, the distribution of zonal and meridional forecast errors. These distributions are well modelled as normal based on the mean and standard deviation of each error. Probability maps giving the likelihood that the storm centre will be located in a given grid cell are then produced for each forecast lead time.
The error distribution for the forecast extent of winds in the four storm quadrants (NE, NW, SW and SE) is modelled as a function of wind speed threshold (34 or 64 knots) and forecast lead time. These forecast error distributions are well modelled as normal. These distributions are combined with the realtime forecast values for the four quadrant wind radii to obtain  in a stormcentred framework  the gridded probability that the wind speed will exceed the threshold in question. With the gridded probabilities being a function of wind speed threshold, lead time and the four quadrant wind radii, 30 million unique probability maps are possible. Each of these probability maps is combined with the corresponding probability map for the error in the storm position to obtain the 'final' wind speed probability maps.