Hurricane hunters flying for the National Hurricane Center, and based out of Keesler Air Force Base in Georgia, typically use dropsondes which are tiny tracking devices attached to balloons to track wind speeds. However, these devices are expensive so the planes only drop a few per storm, which means only intermittent monitoring. Dropsondes were first used in 2007, and about 1000 are used each year for hurricane tracking, at a cost of about $750USD per unit.
Instead the GPS system on-board the weather reconnaissance aircraft works almost continuously as the plane flies through the storm, giving researchers a list of readings. The planes are listening for GPS signals that are being reflected from the oceans surface. When the oceans are choppy due to high wind speeds, the waves reflect the GPS radio waves back in various directions distorting the signal.
Winds blowing over a body of water generate heaving waves. "Imagine you blow on a hot bowl of soup," said Stephen Katzberg, a researcher at the NASA Langley Research Center and author of the study, in a press release. "The harder you blow, the bigger the 'waves' are in the bowl." Scientists have figured out how to measure these reflected 'distorted' radio waves and use the measurements to decode the speed of the wind.
The first comprehensive test run was in 2000, and National Oceanic and Atmospheric Administration (NOAA) scientists have been refining the system every year since then. They system today is almost always accurate for wind speeds between zero and 40 meters per second. While it's not yet as accurate as dropsondes, the analysis is lower cost, lower weight, and requires less power.
Similar systems could be installed in weather drones in the future, and with further technology advances it may be utilized directly in satellites. In 2016, NASA plans to launch a system of 8 small satellites called the Cyclone Global Navigation Satellite System (CYGNSS) to measure reflected GPS satellite signals from low earth orbit in order to monitor storm wind speeds from space for multiple storms.
The data collected helps to predict storm severity, and where the storm may be headed.