GEOCOLOR image generated from NOAA GOES-East of Category-5 Hurricane Michael on October 10, 2018, over the Gulf of America approaching the Gulf Coast. Source: NESDIS STAR
Imagery from geostationary and low Earth polar-orbiting satellites has revolutionized how hurricanes are predicted and tracked, giving communities time to prepare and evacuate.
A hurricane, a type of tropical cyclone, is a large, swirling storm that forms over tropical waters. They often begin near the Earth’s equator, sometimes hitting southern U.S. states, and can bring heavy rains and strong winds as far north as Canada. Each year, over 80 tropical cyclones form worldwide, affecting nearly two billion people. Winds ranging from 74 mph (Category 1) to 157+ mph (Category 5), along with widespread flooding from storm surges and inland heavy rain, can cause catastrophic damage that takes years to rebuild. In some cases, the damage is irreparable.
Live Hurricane Tracker
Before satellites, predicting and tracking hurricanes over the open ocean was extremely difficult. Communities often had little or no warning before a storm made landfall, leading to devastating losses. While we can't prevent hurricanes, today's early warnings give people crucial time to prepare and evacuate. These lifesaving warnings are possible largely thanks to NOAA satellite data.
NOAA’s weather satellites precisely measure air and water temperatures, moisture, and other essential information. This data helps weather forecasters to monitor vast ocean areas far beyond the reach of land-based instruments, even allowing scientists to watch storms forming and evolving in real time.
Predicting a Hurricane
Accurately predicting when a storm will form and where it will go is vitally important. When forecasting tropical cyclones, the more data used, the more accurate the prediction. To advance this effort, NESDIS scientists have developed a machine learning algorithm to help predict these devastating storms using:
- Environmental conditions, such as winds and warm ocean waters available to develop and strengthen cyclones;
- Satellite observations of current thunderstorm activity;
- Locations of current tropical cyclones since they often cannot form near each other.
This algorithm can predict a hurricane up to 48 hours before it develops. As it takes in more high-quality data, its accuracy and reliability will continue to improve. Looking ahead, forecasts could predict hurricanes up to a week in advance by combining machine learning and synthetic aperture radar. The data produced is shared between NOAA, the Department of Defense, and other partner agencies and is also displayed in real time while recorded for future analysis.
