Before satellites and computer models, Bermudians read the weather by older signs. One of the best known was shark oil: oil from a shark’s liver, filtered, sealed in a bottle and hung outside the house like a private barometer. In fair weather, it was expected to stay clear. Before bad weather, it was said to cloud, thicken or form a cone at the bottom.
Hurricane Emily tested our faith in shark oil. In September 1987, the storm turned towards Bermuda, strengthened quickly and crossed the island with little warning, damaging roofs, boats, trees and infrastructure. It entered local memory not only for its impact, but for the speed of its arrival.
For believers, Emily became the awkward case. The old warning system had not sounded the alarm in time. Some argued the storm moved too quickly for the oil to respond. The plainer lesson was that traditional signs could still tell Bermudians something about changing weather, but they could not track a compact, fast-moving hurricane shifting course over the Atlantic.
With modern tech, a hurricane forecast is judged mostly by one question: where will the centre go? That matters for Bermuda. For us, a shift of a few dozen miles can change the wind direction, the sea state and the level of damage. But modern forecasting has moved well beyond the line on the map. The work now involves tracking the whole storm — its structure, wind field, rainfall, waves, ocean heat and likely impact on exposed coastlines.
Over the last decade, the most obvious gain has been in track forecasting. The forecast cone is still often misunderstood. It does not show the full size of a hurricane, nor does it mean areas outside the cone are safe. It shows the probable path of the centre, based on historical forecast error. Even so, that guidance has become far more accurate than it was. Better models, faster observations and improved data assimilation have given forecasters a clearer view of the steering currents that push storms across the Atlantic.
Intensity forecasting remains more difficult. A hurricane’s path is controlled largely by broad atmospheric patterns. Its strength depends on smaller, more unstable processes inside the storm itself: eyewall formation, thunderstorm bursts near the core, dry air, wind shear and the amount of heat stored below the ocean surface. This is why a storm can be well forecast in track but still surprise forecasters by strengthening quickly.
Much of the new technology is designed to close that gap.
Satellites are the foundation of the modern system. The latest geostationary satellites can watch the Atlantic almost continuously, showing changes in cloud structure, eye formation and bursts of deep convection. Lightning mappers can detect sudden increases in lightning near the storm’s centre, sometimes a sign that the core is becoming more vigorous. Polar-orbiting satellites add microwave and infrared data, allowing forecasters to see through cloud tops and examine the storm’s internal structure and surrounding moisture. Satellites no longer simply show where a hurricane is. They help reveal what it is becoming.
The newer frontier is direct observation in places where people cannot safely go. NOAA and its partners now use uncrewed systems such as Saildrones, ocean gliders and air-launched drones to collect data from the boundary between sea and sky, where hurricanes draw much of their energy. Saildrones can remain at sea as storms pass over them, measuring wind, pressure, waves and sea-surface conditions. Gliders move through the upper ocean, showing whether a storm is travelling over deep warm water or stirring up cooler water that may weaken it. Small drones released from hurricane aircraft can sample low levels of the storm in greater detail than conventional aircraft alone.
The modelling has advanced as well. NOAA’s Hurricane Analysis and Forecast System, known as HAFS, represents a shift towards higher-resolution, coupled hurricane prediction. Rather than treating the storm, ocean and atmosphere as separate problems, newer models try to simulate the feedback between them. A storm passing over warm water may strengthen. Strong winds may churn up cooler water. Dry air may disrupt the core. Changes in the eye wall may alter the wind field. The model has to account for all of it, while the storm is moving.
Artificial intelligence is now being added to the forecast process. AI weather models can process vast historical datasets and generate forecasts with remarkable speed. Some are already showing promise in tropical cyclone track prediction. But AI is not a replacement for meteorologists, aircraft data or physics-based models. Its present value is as another form of guidance — fast, powerful and useful, but not infallible. Intensity forecasting, especially rapid intensification, remains one of the hardest problems in the field.
Storm surge mapping has also become more sophisticated. In larger jurisdictions, probabilistic surge products can show where water may rise above normally dry ground, accounting for uncertainty in track, size, speed and intensity. That kind of mapping translates meteorology into practical risk: which roads, homes, docks and low-lying areas may flood.
For Bermuda, that local interpretation is essential. The island benefits from global advances in satellites, models, aircraft reconnaissance and ocean data. But the final warning still has to be read through local knowledge — our reefs, harbours, shorelines, roads, boats, roofs and exposed infrastructure. A passing hurricane is not only a point on a map. It is wind direction, swell angle, harbour exposure, saturated ground, flying debris and the timing of the worst conditions.
The future of hurricane forecasting will be faster, sharper and more data-rich. It will not make hurricanes simple. The technology can narrow uncertainty, expose a storm’s structure and give the public more time to prepare. It cannot remove judgement from the process. That responsibility remains with forecasters: to weigh the evidence, understand the limits of the models and explain the risk before the weather arrives.
Click here to read more from A Guide to Hurricane Season 2026
