Flying a wind sensor

I ended up with a couple of digital wind sensors because logging the wind speed at a KAP rig seemed like a really appropriate thing to do. I connected a Modern Device Wind Sensor (Rev. P) to my SkyPod GPS logger and lofted it with a new seven foot Rokkaku. The wind data are very noisy, so I am trying to figure out why, but it looks like the sensor might be capable of returning good data. I posted some details about the flight at Public Lab.

Above: The sensor has a somewhat delicate loop extension and it's hard to protect that without interfering with the wind readings.

Above: A 10-observation running mean (60 second averages) makes the result look really reasonable. Some more testing is needed to figure out what causes all the noise.

Above: While the SkyPod was logging data, the Canon EOS M was taking 900 photos, 33 of which are stitched into this spherical panorama of the Bread Loaf Campus of Middlebury College, Vermont.


  • edited August 2017
    Great Chris!
    I will add a link to the Kite Aerial Remote Sensing page.
  • Like the thinking Chris.

    I too have an interest in recording wind speed.

    I have been planning on recording wind speed at the ground, the KAP rig and the kite and add in line pull to compare kites and kite line.

    I purchased a couple of small used hand held wind speed Digital Anemometer on Ebay that can be placed on the kite line. Limited time has slowed my pace with these experiments.

    We found the direction anemometers are very sensitive to angle to the wind.

    I certainly would be interested in hearing more about your experiments, wind sensors and data loggers used.

    Ken Conrad and I did a bit of testing back at the AKA Convention in Oregon last October. See Ken's wind plots here.

  • That is very convincing data about wind speed and kite pull. It would be revealing to have that kind of data about different kites.

    I didn't know there were digital anemometers which logged data. I should have one of those for calibrating the hot wire wind speed sensors. It looks like they cost >$200, (here is one used for $130). Maybe Ken figured out how to collect the data from the cheaper anemometers which don't have integrated data logging.

    I have the two Modern Device wind sensors ready for the next tests. I separated the Rev. P wind sensor from the GPS receiver and temperature sensor to see if simplifying the circuit will reduce noise. I will mount it on the Picavet cross so it does not rotate when the autoKAP rig pans. That's not very good science because I am altering two variables, but it seems that these days nobody believes in science anyway. I also changed the battery pack for the Rev. C wind sensor to add a couple of volts and learn if that makes a difference. Now I need a chance to deploy everything.

  • The heated wire trick isn't a good way to read wind speed in such an open system... that's why you're getting such wild readings.

    Changes in ambient temperature, air pressure (elevation), and even sunlight will all invalidate the data.
  • Air temperature and barometric pressure vary predictably in this situation and are only a problem if you ignore them. Both are being measured continuously on the rig. It would be good to make better use of those data to improve the results.
  • I sent the wind speed sensor up again and got somewhat better results. At least there seems to be less noise. The general pattern of the result (winds aloft were 6-9 mph) agrees with my sense of how windy it was.


    This time, the wind speed sensor was riding on the Picavet cross so it did not rotate with the autoKAP rig and stayed in unobstructed air. That is probably the primary reason for the reduction in noise.


    A brief increase in wind speed beginning at 14:14 was the only reason the rig was able to climb to 240 m above the ground. It didn't stay that high for long.


    There was not enough wind to lift the rig with the camera on it, so I removed the camera. So this flight was KA, not KAP. There is more about the flight at Public Lab:

  • Public Lab has a project with NASA and I was able to use a NASA Aeropod Air Column Profiler to compare with the wind speed measurements made by the Modern Device Wind Sensor Rev P. The Aeropod has a Kestrel 5500 data logging anemometer and a vane to keep it pointed into the wind. I have made two kite flights with both the Kestrel and the Rev P sensor hanging near each other on the kite line.


    During both flights the two devices responded in synchrony to variations in wind speed, but the $300 Kestrel 5500 recorded speeds which were consistently faster than the hot wire Rev P sensor. During the first flight the Kestrel result was 5 to 7 mph faster, and during the second flight it was 2 to 3 mph faster. Note different units below.



    The two flights were 10 days apart and the second day was windier and warmer (by 4°C). I had repaired the Modern Device sensor when I broke it after the first flight. The Modern Device sensor seems to have great promise, but needs some calibration work.

    There was no camera on the line during the first flight, but there was plenty of wind for the second flight and the Canon EOS M took 1860 photos. The camera battery died 10 minutes before I walked the kite down after a two hour flight.



    More details and photos are at two notes at Public Lab:

  • Interesting, would you conclude that the hot wire device gives a variabe reading proportional to air temerature?
  • With n=2, I can't conclude much based on my comparisons with another anemometer. The hot wire device is all about temperature (air cools the wire proportional to wind speed), and air temperature is measured and included in the computation of wind speed. There is apparently hardware correction in the Rev P device and there is also a calibration equation in the Arduino sketch I used. But those adjustments don't do as good a job as they could. Paul Badger, the developer, said he has plans to build a little temperature-controlled wind tunnel and develop a good calibration equation, so I think I have to be patient and wait for that result.

    I also broke and then repaired the Rev P sensor between the two flights and I don't have any idea whether the repair changed the sensitivity. I was sort of surprised that it even worked after my repair.


  • I took a different route... instead of measure wind speed (which is questionably useful anyway outside of getting weather readings for science), I measure string tension just above the rig (maybe 10-15' below the kite). When I finish my whole controller rework I'll end up posting pics, but it's basically a physical spring based portable luggage scale with a linear POT put into it. It's reasonably accurate in the 5-15lb range with a max reading of about 60lb.

    This will tell me how much head room I have when lifting my rig. If the kite thrust is too low I know I could be in the danger zone of burying my camera.
  • @ChrisFastie

    you are very very skilled in all concerning electronics and I'm always surprised - upset of what you are able to do

    about the choice of sensor for wind speed a hot-wire is too small, too fast, too sensitive to direction, affected by too many variables (temperature humidity density....)

    much better probably to approach with a simple sturdy 3 cups rotating, vertical axis.... just a relationship between speed of rotation and wind speed, averaging in a larger 3D space and less affected by stream direction

    a good sensor not affected by wind direction (at least in directions horizontal) with a clear OUT signal 0; 1 V or 4; 20 mA to be connected with a logger would be a more trustable start point

    SMAC from Italy
  • Yes, the rotating cups are not sensitive to direction. The Kaindl Windmaster 2 configuration with the protective cage seems ideal. I have not seen a handheld version with rotating cups that logs data. It might be possible to reverse engineer this replacement part ($14.00) and connect it to a data logger. But that proprietary part has no spec sheet available, so success is not guaranteed. More expensive (and heavier) versions are also available.

    I am rather impressed with the results from the hot wire sensor. It responds to variation in wind speed in close agreement with the Kestrel 550, so it seems to be returning reliable data. If the absolute calibration can be achieved, it might be hard to beat the Modern Device Rev P for cost, weight, and size. The data logger I use with it has live data about temperature, humidity, and barometric pressure. These data are recorded at every time interval, and they could also be used to adjust the wind speed computation in real time. All we need is a calibration data set.

    It would be very fun to have a data logging tension sensor on the kite line and also a wind speed data logger. Flying different kites with those two sensors could produce some useful results.

  • @ChrisFastie

    OK, taking into account that you already have hot wire sensor + electronics maybe it's worth a trial...

    in your pictures the reference anemometer is hanged in clear sky with good provision for being directed along wind direction while your hot wire is always quite near obstacles, the picavet or the camera cradle,... maybe that with a proper separate mounting in clear conditions also the hot wire could record something more realistic

    hot wire are used to detect changes of wind speed within millimeters of distance point to point, they feel every stream or turbulence...

    wish you all the best possible

    SMAC from Italy
Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

In this Discussion