Reading the NOAA cone of uncertainty

What the cone actually is

The white-and-grey cone you see on every U.S. hurricane forecast image is one specific thing, and it is usefully described in one sentence: it is the geographic envelope inside which the centre of the storm is statistically likely to pass over the next five days.

The shape isn’t a guess. The National Hurricane Center (NHC) builds the cone by taking the most recent five years of its own forecasts, measuring how far off the predicted storm centre was from the actual storm centre at 12, 24, 36, 48, 72, 96, and 120 hours, and drawing a circle of that radius at each forecast point on the map. The cone is the smooth envelope joining those circles. By the time the storm is five days out, the circle is several hundred kilometres across, simply because forecasting that far ahead is hard.

The probability buried in the cone’s width is well-defined: the historical likelihood that the storm’s centre falls inside the cone is roughly 60 to 70 percent. NHC publishes both a 60% and a two-thirds variant; the cone you see is the official two-thirds version. So one in three storms goes outside the cone shown. That is not a flaw, it is the mathematical definition.

What the cone is not

Here are four things the cone explicitly is not, all of which travellers and journalists routinely treat it as. Each is documented in the NHC’s own “About the Cone” page; they bold-face most of these themselves because they get it wrong so often in coverage.

Not a damage zone

Hurricane-force winds, storm surge, tornadoes, and dangerous rain extend well beyond the cone in almost every case. A storm with a 200 km tropical-storm-force wind radius can have its centre pass right along the cone edge while inflicting severe damage 200 km outside the cone in either direction. NHC’s standard wind-field graphic is published alongside the cone for exactly this reason; the cone shows the path, a separate graphic shows the breadth of impact.

Not the area at risk of impact

A common journalist construction is “X is in the cone” or “Y is outside the cone, so it should be safe.” Both miss the point. A coastal town outside the cone can still receive direct surge, tropical-storm-force winds, and devastating rainfall. A town inside the cone can also escape worst-case impacts entirely if the storm passes 50 km to either side. The cone is about the centre’s path, not the population’s exposure.

Not the only forecast

Reading only the cone discards the other graphics on the same NHC page that carry the actually-useful information for trip-planning decisions: the wind-arrival probability map, the storm-surge inundation map, the watch/warning bar (more on this below), and the wind-field cones that show where tropical-storm-force and hurricane-force winds are forecast to extend. The cone is a navigation tool. It is not the briefing.

Not a guarantee of one specific track

Inside the cone, the storm’s centre might trace any of dozens of plausible lines. Sometimes the actual track turns out to follow the cone’s exact centreline (the “official forecast”). More often it wanders from one side to the other. NHC publishes a separate graphic with the spaghetti plot of all ensemble model tracks; reading that alongside the cone gives a much more honest sense of the uncertainty than the cone alone.

Four common misreadings, and what to do instead

1. “The cone shrinks as the storm gets closer, so it’s getting more accurate”

The cone shrinks because forecasts shorter than five days are more accurate than forecasts five days out. Yes, it is more accurate. No, that doesn’t mean the storm is “coming for you” or “turning away.” The cone shrinking is a property of the calendar, not of the storm. Don’t read trajectory information into shape changes.

2. “The cone barely overlaps my hotel, so I’m fine”

The cone shows where the centre of a storm with a roughly 200–400 km wind field might pass. If the edge of the cone touches your hotel even at 96 hours out, you should assume hurricane-force winds and surge could reach you. Read the wind-field graphic and the storm-surge inundation map (linked from every NHC advisory page), not just the cone.

3. “The cone has shifted east, so I should head west”

Cone shifts of less than 100 km between advisories are within the normal forecast jitter. NHC issues advisories every 6 hours during an active storm, and small shifts in the centreline are routine even when the storm is doing nothing unusual. Don’t make irreversible logistical decisions on individual shifts; look at the advisory text discussion (linked from every advisory page), which explains what changed and why.

4. “The cone doesn’t cover us at all”

Two scenarios where this is wrong. First, beyond 120 hours the cone simply isn’t drawn; you may be in a high-probability zone that the cone doesn’t yet show because the forecast horizon hasn’t extended that far. Second, the cone is centred on the storm’s expected centre; if you are 100 km off the cone but downwind of the storm’s wide rain shield, you’re still going to get the storm even if you don’t get the eyewall.

The rest of the graphic, in plain English

A typical NHC advisory page contains four blocks. The cone is one. The other three are arguably more decision-useful for travellers.

The watch / warning bar

Coloured lines along the coast in the standard NHC graphic. The terminology is precise:

• Tropical Storm Watch (yellow): tropical-storm-force winds (39 to 73 mph / 63 to 117 km/h) possible somewhere inside the watch area within 48 hours.
• Tropical Storm Warning (blue): tropical-storm-force winds expected somewhere inside the warning area within 36 hours.
• Hurricane Watch (pink): hurricane-force winds (74 mph / 119 km/h+) possible inside the area within 48 hours.
• Hurricane Warning (red): hurricane-force winds expected inside the area within 36 hours. This is the line at which travel insurance, airline waivers, and most evacuation orders trigger.
• Storm Surge Watch / Warning (purple variants): life-threatening inundation possible / expected. Storm surge kills more people in U.S. tropical-cyclone events than any other hazard; if a surge warning covers your hotel, leave.

The Saffir-Simpson scale

Categories 1 through 5, based on sustained wind speed only:

• Cat 1: 74–95 mph. Some damage, mostly to roofs, gutters, large branches.
• Cat 2: 96–110 mph. Extensive damage to homes, trees down across power lines.
• Cat 3: 111–129 mph (“major” hurricane). Devastating damage; well-built homes lose roof decking.
• Cat 4: 130–156 mph. Catastrophic. Most trees snapped. Power and water out for weeks.
• Cat 5: 157+ mph. Total destruction of framed homes; uninhabitable for months.

The Saffir-Simpson scale describes wind only. It explicitly does not describe storm surge, rainfall, or tornado risk. Cat 1 hurricanes have killed people through surge and flooding when their winds were unimpressive. Cat 5 hurricanes have produced relatively minor flooding when they moved fast and dry. Read the wind category as one number among several.

Storm-surge inundation map

Available for any storm threatening U.S. coast at nhc.noaa.gov/nationalsurge/. It shows the height above ground that water could reach at each location. Surge is the dominant killer in tropical cyclones globally; the map is the single most important graphic for evacuation decisions in low-lying areas.

Forecast discussion (the text)

Every advisory has a written discussion linked from the page, signed by the on-duty hurricane specialist. It explains what model output the centreline came from, where the spread is widest, and what shifts the forecaster considered but rejected. It is written in plain English (with a few acronyms; GFS, ECMWF, and HWRF are the main forecast models referenced). Reading the discussion takes three minutes and gives you the texture of uncertainty the cone alone cannot convey.

Pacific and Indian Ocean cones, briefly

Outside the NHC’s area (Atlantic, Eastern and Central Pacific), other agencies issue their own tropical-cyclone forecasts and use slightly different conventions. The geometry is broadly the same; the terminology and the agency to follow change.

Western Pacific (typhoons)

Three agencies actively forecast in this basin: the U.S. Navy’s Joint Typhoon Warning Center (JTWC), the Japan Meteorological Agency (JMA), and the China Meteorological Administration (CMA). JMA’s forecasts are the official reference for international aviation in the basin. JMA’s typhoon track map looks similar to NHC’s cone but uses a wider 70% probability envelope by default. JTWC tracks are also widely cited; the two agencies often differ by 50–100 km on short-range forecasts and that disagreement is informative.

Northern Indian Ocean (cyclones)

The India Meteorological Department (IMD) is the official forecasting agency. IMD publishes “cyclone advisories” with a track-and-cone presentation. The basin produces fewer but historically more deadly cyclones (Bangladesh, Bay of Bengal coast). JTWC also publishes parallel forecasts.

South Pacific and South-West Indian Ocean

Australia’s Bureau of Meteorology covers the south-west Pacific and Australian region. Météo-France (Réunion) covers the south-west Indian Ocean (Madagascar, Mozambique). Both publish cone-style track forecasts on their public sites.

For travellers, the practical rule: pick the right agency for the basin you are in, follow that agency’s official track, and read the agency’s discussion. Don’t cross-check between basins (a JTWC forecast for an Atlantic storm doesn’t exist; a NHC forecast for a typhoon in the western Pacific doesn’t exist).

The 72-hour rule

The single most useful framework for trip decisions is the 72-hour window. Three days out, the forecast is accurate enough that you can usually tell whether your destination is in the threat zone, and the hours remaining before impact are still long enough to actually act on a leave-or-stay decision (rebooking flights, fuelling cars, securing accommodation).

• 120+ hours out (5+ days): Don’t cancel. The storm could go anywhere within a wide area. Watch advisories, but don’t rebook yet (most airlines won’t waive fees this far out).
• 72 to 120 hours (3 to 5 days): Plan B. Identify alternative routings, check airline waiver policies, look at refundable accommodation. Don’t commit irreversibly.
• 72 to 36 hours (3 days to 1.5 days): Decision window. By now the cone shows your destination as inside or outside the threat zone; the watch / warning bar will be set. If you’re in a watch or warning area and you can leave, leave. Airline waivers normally activate around 72 hours out.
• Inside 36 hours: If you haven’t left, shelter is now your better option than evacuation. Roads degrade fast as evacuation volume rises; a building inland, away from windows and surge zones, is safer than a vehicle on a buckling highway.

Country brief: where travellers go

Florida and the U.S. Gulf Coast

Atlantic hurricane season runs 1 June to 30 November, with peak activity August through October. NHC is the authoritative source. Florida coastal counties have well-established evacuation zone systems (Zone A through E in most counties); your hotel will know its zone. Always know yours before you check in.

The Caribbean

Same Atlantic hurricane season as Florida. Smaller islands have less infrastructure for evacuation; a storm that the Florida coast shrugs off can be a major event in the Bahamas, Turks and Caicos, or the eastern Caribbean. Insurance trip-cancellation cover sometimes triggers earlier for Caribbean-bound travel than for U.S. mainland.

Mexico (both coasts)

Pacific season runs 15 May to 30 November, Atlantic season 1 June to 30 November. NHC forecasts cover both. Mexican civil protection (Protección Civil) issues evacuation orders for coastal municipalities; resort properties usually run their own brief alongside.

Philippines and Vietnam (typhoon season)

The Philippines averages 20 named storms per year, more than any other country. Season is May through October. Follow the Philippine Atmospheric, Geophysical and Astronomical Services Administration (PAGASA) for local forecasts; JTWC for the broader Pacific picture. Vietnam’s central coast (Da Nang to Nha Trang) is the most exposed.

Japan

Typhoon season runs May through October, peaking August and September. JMA is the authoritative source. Japan’s infrastructure is the most typhoon-resilient in Asia; trains and aviation suspend hours ahead of landfall but resume quickly. Hotels in Okinawa and Kyushu run drill protocols and are very used to it.

Bangladesh and the Bay of Bengal

Two short cyclone seasons, April–May and October–November. The Bay of Bengal is the deadliest tropical-cyclone basin in the world historically, due to low-lying coastlines and population density. IMD is the regional authority. Coastal travel during the seasons is best confined to inland Dhaka and the highlands.

Australia

Cyclone season runs November to April in the northern half of the country. Bureau of Meteorology (BOM) issues forecasts and warnings; Queensland and Northern Territory have well-established coastal warning systems.

Trip planner checklist

A short, actionable checklist for a coastal trip booked into a tropical-cyclone season:

• Travel insurance with cyclone cover. Read the policy wording. Many policies define “named storm” or “hurricane warning” as the trigger; some require it to be in place 48 hours before your departure date. A few exclude any trip booked when a storm was already named.
• Refundable accommodation. Adds 10–20 percent to the cost; pays for itself the first time you need it.
• The official agency app for your destination’s basin (NHC, JMA, BOM, IMD, PAGASA). Free.
• Your hotel’s evacuation zone if Florida, Texas, or the Caribbean. Ask before you check in, not after.
• A hard-copy print of your travel documents in your passport pouch. Phone networks often degrade to text-only for hours after landfall.
• Cash in small notes. Card terminals fail when power does.
• The 72-hour rule mentally pre-loaded. Decide in advance which forecast tier triggers which action.

One more time

The cone is the storm centre’s likely path, drawn from five years of NHC’s own forecast accuracy. It is roughly two-thirds reliable. It tells you nothing about damage zone, wind field, surge, rainfall, or tornado risk. Read the wind-field graphic and the surge map alongside it. Decide on the 72-hour rule. When the warning goes up, leave on the warning rung, not the watch rung, not the order rung. The cone is a beautiful piece of statistical communication; treat it as one input, not the answer.