Study on using hard-to-obtain observations from uncrewed aircraft in Hurricane Maria to improve analyses and forecasts published in Monthly Weather Review

For the first time, near-surface observations obtained by small Uncrewed Aircraft Systems (or drones) in a major hurricane (Hurricane Maria of 2017) were utilized in a state-of-the-art computer. Such observations are impossible to obtain using regular hurricane hunter aircraft. Furthermore, a new technique to identify and eliminate observations that may not be useful in improving analyses or forecasts was devised. The end result was large improvements to the analyses of the hurricane. The results speak to the importance of parallel and consistent advancements in modeling, data assimilation, and observational capabilities to obtain a better depiction of hurricane structure to improve numerical model forecasts.

Co-author Joe Cione holding a Coyote sUAS in front of a NOAA P-3 aircraft.

Meteorologists study tropical cyclones by flying into them with hurricane hunter aircraft. However, some regions of the tropical cyclone, like near the ocean surface, are too dangerous to fly in. So, NOAA has worked with a number of organizations to develop small Uncrewed Aircraft Systems (sUASs) that can gather data in these areas. One such system is the Coyote, which measures wind velocity, temperature, moisture, and pressure, typically several times every second for up to one hour. An important test of new data types like those from the Coyote is whether they will improve computer model analyses and forecasts. We ingest the data into the models using a procedure known as data assimilation. 

A unique dataset was obtained from two Coyotes near the center of Hurricane Maria in 2017. For this study, the data from the Coyote is put into a version of NOAA’s Hurricane Weather and Research Forecasting (HWRF) model using an experimental data assimilation technique developed at the Atlantic Oceanographic and Meteorological Laboratory known as the Hurricane Ensemble Data Assimilation System (HEDAS). To help improve the impact of the data on the model forecast, a new technique that identifies data that may not be useful and removes them from the model before they can influence the forecast was devised. This was the first time ever that such high-resolution sUAS observations were utilized in an advanced hurricane model.

Comparison of best analyses and observations from Hurricane Maria on 23 September 2017 using all the data available. (a) analyzed radar reflectivity pattern, (b) observed reflectivity pattern from the NOAA P-3’s tail Doppler Radar along with wind velocity 2 km above the ocean surface, (c) analyzed wind speed 2-km above the ocean surface, and (4) observed wind speed from the NOAA P-3 tail Doppler radar.

Important Conclusions:

  • The Coyote observations improved the analyses of Maria’s position, intensity, and structure in Hurricane Maria (2017).
  • Further improvements were obtained by using the new technique to identify observations that may not improve the analyses and removing them from the model. 
  • In current forecast models, data are assimilated every 6 h. Additional improvements to the analyses were achieved by assimilating the data at a very frequent rate (every 5 min) into the model. This suggests a possible way to improve model forecasts.

You can read the manuscript at For more information, contact

Partial funding support was provided through the Cooperative Agreement NA67RJ0149 between NOAA and the University of Miami. Appreciation goes out to all NOAA Aircraft Operations Center (AOC) staff who supported P-3 sUAS Hurricane Hunter missions into Maria. Without their dedication these missions would not have been possible. The authors acknowledge the NOAA Research and Development High Performance Computing Program for providing computing and storage resources that have contributed to the research results reported within this paper.

One thought on “Study on using hard-to-obtain observations from uncrewed aircraft in Hurricane Maria to improve analyses and forecasts published in Monthly Weather Review

Comments are closed.