June 10, 2013

Using Unmanned Aircraft for Severe Thunderstorm Prediction and Warning

After the tornado outbreaks in Moore City and Oklahoma City, Oklahoma, there was a flood of controversy about the role of storm chasers.   With the tragic loss of three experienced storm chaser/scientists, the injury of the several Weather Channel chasers (see picture below of their car), and the veritable traffic jam of chasers in the immediate vicinity of the Oklahoma City tornado (see image), a number of observers have suggested that there must be a better way of observing tornadoes.

Furthermore, when a severe thunderstorm pops up in an unexpected location (and this happens plenty of times), there are often no chasers in the neighborhood to give the National Weather Service ground-truth observations.
Position of storm chasers (with GPS and who reported their position) during the Oklahoma City tornado event.   There were certainly more chasers than this in the area.  The background image is the Doppler velocities, with the yellow/blue couplet indicated the mesocyclone of the storm

We need a way that puts fewer chasers at risk, reduces the potential for chaser-induced traffic jams, and collects more useful information.  An approach that would allow the National Weather Service to get the information they need when storms develop unexpectedly.    And a way to get large amounts of information before severe thunderstorms develop and thus enhance short term (1-12 hr) forecasts.

Could such a breakthrough be possible? 

I believe that answer is emphatically yes.

The approach?  Take advantage of small unmanned aircraft capable of providing both video and weather information while in flight.  And as I will explain below, the planes are ready for action and available from a local company here in the Pacific Northwest.

Imagine you had a small, relatively inexpensive aircraft (costing roughly 50 K$) that had the ability to fly with a high-definition video camera and the same weather instruments in radiosondes (balloon launched weather sensors).  The plane would have a large range (1000s of km) and the ability to stay aloft for extended periods (1-2 days).   A plane that had GPS so an accurate position was known at all times, satellite communication (so it was in constant communication), a sophisticated onboard computer for data collection and flight control, great strength (so severe turbulence would not damage it) and the ability to be easily deployed and recovered.

Such a plane, under the control of National Weather Service staff (or a private firm hired by the NWS) could fly out to potential tornadic storms and fly around funnel clouds or threatening structures (e.g., rotating wall clouds).   Continuous imagery of the developing tornado, with accurate position information, would be available to NWS forecasters, along with insitu observations from the aircraft sensors.   All of this could be done with complete safety and high reliability.     If a new storm developed, such small aircraft could be deployed quickly and, with speeds of 50-100 mph, could get into position quickly and safely (unlike some overly exuberant chasers that drive very fast over rain-slickened roads).    Perhaps you could get an idea of the view of such a small aircraft from the imagery sent back by some TV helicopters during some events (see below).  The imagery could be fed to local TV stations and emergency managers in real time.

 I should stress that it would not be necessary to fly such aircraft into the storm, but just to keep a respectful distance (not unlike the TV helicopters do near some major cities).  The large hail and heavy rime icing in severe convection would not be good for the health of these small aircraft.

Amazingly, an aircraft capable of such unmanned surveillance is now available for purchase from the Aerovel Company of White Salmon, Washington.  Aerovel is a company founded by Dr. Tad McGeer, an extraordinarily talented and innovative aircraft designer, who has created a series of highly successful small Unmanned Aerial Vehicles (UAVs), including many that are now used by our military (e.g., the ScanEagle).   His firm has now developed and successfully tested the Flexrotor, a small aircraft capable of taking off vertically like a helicopter, going into horizontal flight, fly a long mission, and then return to base, landing as a helicopter (see picture).   Go to this web site to see an amazing video of a flight.  A technological triumph.

The current version of this aircraft has amazing capabilities, including a 3,000 km range and an ability to stay aloft for 40 hr, while carrying the instrumentation and video cameras noted above.  It could revolutionize the monitoring of severe storms, and provision of warnings to local residents and businesses.

But the great potential of this aircraft does not end there.   Before storms develop it could fly in the lower portion of the atmosphere (lower few thousand feet) in a way to get voluminous observations of the environment in which convection (thunderstorms) will develop.   Many of us believe that this is an essential ingredient for further progress in severe storm prediction.  In fact, an experiment using manned aircraft (the Mesoscale Predictability Experiment, MPEX) is going on this spring to prove this hypothesis. 

And after severe storms hit, such small aircraft could get extensive imagery of the damaged area to aid first responders and others.

So what stands in the is way of this vision?

First, the National Weather Service needs to consider the potential of such aircraft seriously and funds for the experimental use of the small planes must be found.  Perhaps a local foundation (e.g., Gates, Bezos, Allen) could fund a demonstration project.

Second, the Federal Aviation Administration (FAA) must allow these planes to fly in U.S. airspace.  To put it mildly, the FAA has not  encouraged unmanned aircraft and has severe limitations on this use.  They need to rethink their objections for  UAVs applied for storm forecasting.  Such planes could potentially save many lives.  They will do most of their flying around severe thunderstorms, where few commercial or general aviation flights dare travel.  The planes will be in constant contact with controllers and their positions will be known at all times, so other aviation will be constantly appraised of their position. 

It would be ironic if we could use unmanned aircraft to kill and damage our foreign enemies, but were unable to use them to save many American lives.

Small unmanned aircraft have the potential to revolutionize severe storm prediction and warning, and would give the National Weather Service a powerful tool for protecting American lives and property.  Such UAVs have other potential  meteorological applications, like hurricane reconnaissance.   I believe they are the future, but it will take some effort and vision to make it a reality.


  1. See also Insitu, a Washington company in the town of Bingen (on the Columbia River). The company is now owned by Boeing. It produces UAVs mostly for military use, but they could presumably be adapted for weather applications. Their website implies that they developed ScanEagle. The Aerovel website only claims that their developers previously worked on ScanEagle.

  2. Yes, but the issue is that these can become either projectiles or debris based on a given storm that can cause property damage or injuries/death to people if they crash or come apart.

    The liability issues come into play and no one would want to foot the insurance bill vs potential lawsuits.

  3. You might like to watch this: "Timelapse of a supercell near Booker, Texas" at http://www.mikeolbinski.com/theblog/2013/06/timelapse-of-a-supercell-near-booker-texas/

    (Google is being goofy at the moment - I hope I haven't sent this multiple times.)

  4. In fact, NSF has/is already supported work along these lines under the auspices of our joint NSF-NOAA 'VORTEX2' project. The following links are illustrative:




    I should also point out NSF's newly acquired (albeit conventionally piloted) storm-penetrating A-10 aircraft, which is now expected to come online for test flights in 2014. It's extended duration/far higher operational ceiling as compared to its predecessor (a single-prop T-28 aircraft) have the potential to strongly impact in-situ severe storm research:




  5. Storm spotters cost the government almost nothing - just an annual or biennial training event that lasts an hour or two where one NWS staff member (or even another spotter) teaches a number of spotters.

    Spotters are sometimes supplemented by chasers, but the roles of the two should not be confused. Most chasers (I'm a spotter and a chaser, and a former pilot), other than the few scientific teams, are there to enjoy the storms, only reporting if nobody else is around to do it.

    While UAV's have a role in storm research, it is highly unlikely that they will be suitable for routine severe thunderstorm observation, for a number of reasons.

    1) Cost - they are a lot more expensive than a spotter, who is virtually free.

    2) Communications - how do they communicate back to NWS? Spotters do it for free by ham radio or using their own cell phones. They may transmit video if its available, but if not, they are trained to interpret the storms. The aircraft would have to transmit video. Good luck finding reliable bandwidth for that - especially for free!

    3) Coverage - it is not unusual for a severe weather event to have dozens of cells that all need to be watched. That's a lot of expensive aircraft in the air at once.

    4) Turbulence and strong winds - one reason there are no aircraft around severe thunderstorms is because it isn't a good place for aircraft to go. The smaller (and slower) the aircraft, the worse its tolerance for the winds found around one of these storms. When you have 80 mph rear flank downdrafts, it's pretty hard for an aircraft to maintain position, impossible for a small drone!

    5) Hail - virtually every tornado is accompanied by large hail. Spotters can take shelter at gas stations, car washes, or in their own homes. However, a single hit from a 3" or greater hail stone would destroy a drone.

    6) Staff time - trained UAV pilots would be needed for much of the operation of the drones. Meteorologists would be needed to watch the video and make determinations (unless it could be fed to a spotter, but this wouldn't work as well as having the spotter "on the spot.").

  6. I agree that more observational data is better when trying to nowcast severe storms. Just look at the immense contribution the aircraft dropsonde data has done for storm prediction and research. Quite amazing. The NCAR G-V experiment currently underway will be very beneficial. So previously has the NOAA G-IV dropsonde missions around hurricanes and winter storms.

    I've got a hunch that drones, low level in severe storms and hurricanes won't prove so valuable - as compared to larger, expensive P-3s and C-130s. Mainly because they won't see the trouble coming their way that will kill them. So you better buy at least 20 of them if you want one to come back...

    Flying low level around a supercell thunderstorm is a very tricky dance. One wrong turn and you're in that thing and it will eat you. Nothing like your own eyes and a flying weather radar platform like the NOAA P-3s to let you know where the jaws of that thing are... and you've got a squad of our most respected weather researchers right behind you worrying about exactly that too...

    Low level drones in hurricanes? I don't like that one either. You get below two thousand feet in one of those storms and the turbulence is worse that sitting in a stucko hotub. How's your little six foot drone going to handle that? Below a hundred feet and a wave will reach up and swallow it. And what protection will small drones (and their instrumentation) have from the salt? There's a salt haze lifting off the water's surface up to a couple thousand feet. Drone motors won't run long sucking down salt, and who would trust the readings of salt encrusted instrumentation?

    Bottom line, the Global Hawk high altitude platform will probably work out well. But before we toss money at little low level drones, let's first get all the ground based doppler radars upgraded. And don't ever give up on the manned platforms for research when they're staffed with experienced operators and dedicated researchers.

  7. For better sensor data on severe local storms, I think that ground-based platforms that are much closer together than NEXRAD stations offer some possibilities.

    The dramatic price decreases in multi-sensor, communicating devices - cell phones and tables - suggests that low cost sensors could be

    spread around at high densities. A custom device based on cell technology could provide video on demand (with multiple cameras - they're dirt cheap in this form).

    It could provide barometric pressure and temperature, and even dew point if one accepts reduced accuracy (which might be somewhat correctable using on-device smarts and correlation with other sensors).

    Such a sensor could be produced for maybe a thousand dollars, and could be located on volunteer sites, and maintained (simple maintenance like cleaning, and device swap-out) by volunteers.

    Beyond that, what we really need is a much denser grid of radars. If one avoids normal government practices of overspecifying, one could do this relatively inexpensively also. They might not be as precise as NEXRAD, but they would have a different, complementary mission: see the parts of the storm NEXRAD cannot (near the ground) and see those parts in reasonable resolution.

    As I posted above, I really doubt that drones have much operational utility outside of tropical cyclone high altitude (and dropsonde) work.

  8. The process of integrating met sensors into UAVs has been underway for some time and is continuing at both universities and government entities. There are plenty of technical issues but nothing insurmountable.

    As pointed out in the comments, the real question is what is the appropriate mission profile. Forest fires, toxic plume releases and other localized situations may be suitable where trying to nowcast tornadoes may not.

    You might also want to check out what the EPA is doing with its NGAM initiative (Next Generation Air Monitoring). This isn't meteorology but it's looking at the same concept of how to use cheap, widely distributed sensors (e.g. cell phones) to create higher density networks.


  9. The problem with unmanned drones is that for all the benefits they might provide, they also represent a profound threat to privacy. The potential for abuse is almost limitless. For this reason I'm not in favor of the use of this technology.


Please make sure your comments are civil. Name calling and personal attacks are not appropriate.

The Weather Regimes of Summer

 Weather patterns tend to get "stuck" for extended periods and we have certainly seen such persistent conditions this summer.    W...