Right Turn, Clyde…You Too, Erin, Gabrielle, Humberto & Melissa

How Forecasters Knew They Would

READ IN APP

The author is a postdoctoral researcher at the NOAA Hurricane Research Division, University of Colorado-Boulder. This October 29, 2025 article first appeared in The Conversation and is reprinted here with permission.

By ETHAN MURRAY

Hurricane Melissa grew into one of the most powerful Atlantic tropical cyclones in recorded history on Oct. 28, 2025, hitting western Jamaica with 185 mph sustained winds. The Category 5 hurricane blew roofs off buildings and knocked down power lines, its torrential rainfall generated mudslides and flash flooding, and its storm surge inundated coastal areas.

Melissa had been wobbling south of the island for days, quickly gaining strength over the hot Caribbean Sea, before taking a sharp turn to the northeast that morning.

As a hurricane researcher, I work with colleagues at NOAA’s Atlantic Oceanographic and Meteorological Laboratory to improve predictions of hurricanes’ tracks and strengths. Accurate forecasts of Melissa’s turn to the northeast gave many people across Jamaica, Cuba and the eastern Bahamas extra time to evacuate to safer areas before the hurricane headed their way.

Share

Throughout 2025, most hurricanes similarly veered off toward the open Atlantic, sparing the U.S. mainland. To understand the forces that shaped these storms and their paths, let’s take a closer look at Melissa and the 2025 hurricane season.

Origins

Before they evolve into powerful hurricanes, storm systems start out as jumbled clusters of clouds over the open ocean.

Many of 2025’s Atlantic tropical cyclones began life far from the U.S. coastline in the warm waters west of Africa, near the Cape Verde islands. These Cape Verde hurricanes are consistently blown toward the United States, especially during peak hurricane season.

The driving force steering these storms is a hot, semi-permanent high-pressure air mass often found spinning above the Atlantic Ocean known as the Bermuda high or Azores high.

When this high-pressure system, or subtropical ridge, is positioned farther east, closer to the Azores islands, its strong, clockwise-rotating winds typically curve tropical cyclones briskly out to sea toward their demise in the cold North Atlantic. When the high-pressure ridge is closer to the U.S. and centered over Bermuda, it can send storms crashing into the U.S. coast.

Because that high-pressure system was positioned further east in summer and fall 2025, many of the season’s strongest storms, such as hurricanes Erin, Gabrielle and Humberto, swung east of the U.S. mainland. Combined with an active jet stream above the Southeast U.S., most tropical cyclones were steered away from the Atlantic coast.

The clouds that eventually became Hurricane Melissa traveled farther to the south, avoiding the Bermuda high and making their way into the Caribbean Sea.

Balancing Act

After a tropical cyclone forms, its path is guided by the movement of air surrounding it, known as atmospheric steering currents. These steering currents direct the forward movement of storms in the Atlantic at speeds ranging from a sluggish 1 mph to a blistering 70 mph or more.

Hurricane Melissa’s meandering track was determined by these steering currents. At first, the system was caught between winds from high-pressure systems to its northwest and southeast. This setup trapped the storm over the warm Caribbean Sea for days, just to the south of Jamaica.

As a tropical cyclone is steered by outside forces, its internal makeup also constantly evolves, changing how the storm interacts with its steering currents.

When Hurricane Melissa was a weak, lopsided system, it didn’t receive much of a push from its upper-level environment. But hurricanes draw energy from warm water, and Caribbean sea surface temperatures have been rising. As Melissa gained strength from very warm ocean water below, it grew taller. Like a skyscraper reaching high into the air, major hurricanes like Melissa have towering thunderstorms and feel more of a push from upper-level winds than weaker storms do.

Melissa’s center also became aligned vertically, allowing the tropical cyclone to rapidly intensify from 70 mph to a staggering 140 mph sustained winds in 24 hours, and it would continue to strengthen.

Eventually, the precarious atmospheric balancing act holding Melissa in place collapsed. A ripple in the jet stream known as an atmospheric trough steered the hurricane to the northeast and into the Jamaican coast.

Melissa’s snail’s pace of about 2 mph was rare but not unheard of. Slow storms like Melissa are more common in October, as steering currents are often very weak or pushing in opposite directions, which can trap a tropical cyclone in place. Similar steering currents affected Hurricane Matthew in 2016.

Tragically, stalled tropical cyclones often bring prolonged rainfall, winds, flash flooding and storm surge with them. The wind and downpours can be extreme for mountain communities, as their high topography enhances local rainfall that can trigger mudslides and flooding, as Jamaica saw from Melissa.

Better Forecasting

Meteorologists generally understand how atmospheric steering currents guide tropical cyclones, yet forecasting these wind patterns remains a challenge. Depending on the atmospheric setup, certain hurricanes can be harder to forecast than others, as changes to steering currents can be subtle.

New approaches to hurricane track forecasting include using machine learning models, such as Google DeepMind, which outperformed many traditional models in forecasting storm tracks this hurricane season. Rather than solving a complex set of equations to make a forecast, DeepMind looks at statistics of previous hurricane tracks to infer the path of a current storm.

NOAA Hurricane Hunter reconnaissance data can also accelerate progress in predicting tropical cyclone paths. Recent tests show how accounting for specific measurements from within a hurricane can improve forecasts. Certain flight patterns that Hurricane Hunters and drones fly through strong hurricanes can also improve predictions of a storm’s path.

Scientists and engineers aim to further improve hurricane track and intensity forecasts through research into storm behavior and improving hurricane models to better inform the public when danger is on the way.

© 2025
411 Walnut St. No. 1944, Green Cove Springs, FL 32043
Unsubscribe

Uncrewed Vessels Will Use AI To Interpret Nav Data

U.K. Researchers Teaching Control Systems How To Understand Sailing Directions

READ IN APP

The author is a regular contributor to Marine Industry News of the U.K., which published this story on October 16. It is reprinted here with permission.

By GEMMA HARRIS

A research project has been launched in Plymouth to teach autonomous vessels to read and act on official navigation data.

The eight-month initiative, led jointly by the United Kingdom Hydrographic Office in Taunton and Plymouth-based Marine AI, aims to develop AI that is capable of interpreting Admiralty sailing directions and radio navigation warnings.

The Admiralty is the British government agency historically responsible for its Navy. Now, it is also in charge of hydrography, charting, marine data and advice on maritime matters.

“This is the first time anyone has attempted to process Admiralty Sailing Directions and Radio Navigation Warnings in a way that an autonomous control system can act upon,” said Oliver Thompson, technical director at Marine AI. “By proving this capability on the water, we are closing one of the biggest gaps in (uncrewed vessel) autonomy and taking a major step toward safe, fully automated operations.”

Share

Such a project represents a world first in applying Large Language Models (LLMs) to process maritime navigation information for autonomous control systems.

The maritime autonomy software firm, Marine AI, will retrain its baseline LLM to translate unstructured, text-based navigational data into formats usable by its GuardianAI autonomy software suite. The goal is to enable Maritime Autonomous Surface Ships to make safe, real-time operational decisions using the same authoritative information relied upon by professional mariners.

Currently, uncrewed vessels depend on humans to interpret navigation warnings and sailing guidance, much of which is distributed through legacy systems and written in natural nautical language. The project will address these challenges by using AI to convert this into structured data that can be integrated into autonomous decision-making systems.

In spring 2026, there is a planned on-water demonstration, when the ZeroUSVs Oceanus12 vessel, fitted with Marine AI’s GuardianAI suite, will navigate Plymouth’s waters using the newly developed capability. The trials will run alongside advanced simulation exercises and are expected to inform the International Hydrographic Organisation’s S-100 data framework—one that is underpinning the next generation of digital navigation standards.

Mark Casey, head of Research, Design and Innovation at the Hydrographic Office, said: “Working with Marine AI allows us to push the boundaries of how autonomous systems can use official hydrographic information. The outcomes will not only support the safety of lives at sea but also feed directly into the development of the International Hydrographic Office’s S-100 framework, ensuring that Hydrographic Office data continues to set the global benchmark for safe navigation in both crewed and uncrewed vessels.”

Plymouth, on the south coast of Southwest England, has become a national hub for autonomous maritime research, and this new project presents an opportunity to further strengthen its role as a testbed for uncrewed vessel technology.

Read more stories like this one in the Marine Industry News.

You’re currently a free subscriber to LOOSE CANNON. For the full experience, upgrade your subscription.

Upgrade to paid