Ultrasonic anemometer for an absurdly accurate weather station

With his meteorological interests, [Carl] builds weather stations. Temperature and humidity sensors are a dime a dozen, but with his DIY ingenuity, [Carl] has built some very interesting and complicated devices. The latest of which is an ultrasonic wind sensor that uses the time of flight of ultrasonic pulses to detect how fast the wind is blowing.

[Carl]’s sensor uses four ultrasonic transducers aligned to North, South, East, and West to detect the wind speed. By measuring the time it takes an ultrasonic pulse to travel between the sensors indoors, Subtracting the in-situ measurement gives him the time of flight for each axis, and thus the wind speed.

It’s an impressive display of engineering that comes with an amazingly detailed design report. After three months of operation, [Carl] has found his ultrasonic anemometer is better than the traditional mechanical ‘egg-cup’ anemometer at measuring low wind speeds. The only real problem with the build is the fact the design makes a great bird perch, but some fine steel wire quickly corrected that problem.

32 thoughts on “Ultrasonic anemometer for an absurdly accurate weather station

  1. this is cool. There is a ton of literature out there on these, including optimal mounting positions to reduce shadowing by the transducers and the matrix transforms needed to correct for the position offset.

    Ultrasonics do do a better job than cheap cup anemometers at low wind speed, but they really shine in being able to resolve small scale fluctuations in wind velocity, which can be used to compute structure functions for analyzing turbulence and mixing. For basic meteorology, where you just want general speed and direction, you would filter most of this out, which would give you the same time response as the cup anemometer.

    1. If you read TFR, he has a temperature sensor and does correct for the effects of temperature. In fact, he has a fairly detailed analysis of how temperature affects his measurements. He also states that atmospheric pressure and humidity don’t affect his calculations.

    2. Absolutely. He also completely forgot about the Coriolis effect. By the time the sensors have picked up on the signal, the emitter has changed position! Nevermind he is completely ignoring the issue of sidereal correction. Honestly, it is like amateur hour in here.

      1. The sensors are fixed and aligned with the N-S and E-W axis as far as I can gather. Therefore the Coriolis effect does not come into play, and what about the sidereal correction? Where does it come in?

        1. Twas a joke. Anyway, unless you are revolving around on the ecliptic plane (something you can only do in the tropics twice a year) then the angle of your rotation is not 0. Even then there will be some effect, just takes one less math step to figure.

          For sideral time, if it is not being taken into account then the speed will be off up to .3%.

          The joke is in how negligible change in air density is to the overall apparatus. I simply stated a couple more corrections that are equally preposterous.

      2. You kidding right? It is measuring relative wind direction and velocity which can be corrected with a true wind vector equation. Furthermore, Local to Universal time corrections have nothing to do with this as suggest.

  2. The physics and electronics of the project are indeed amazing. But it begs a question in my (admittedly non-meteorological) mind. When you are measuring wind speed, aren’t you by definition concerned with macro measurements – sustained speeds over long(ish) periods of time that can lead to some change in weather? When measuring with such precision, are you measuring the movement of air that has anything to do with weather (or with your quadcopter flight path etc.) or are you measuring the turbulence from the four sensor arms, the post, a possible adjacent building, a hill, a tree etc. ?

      1. Three axis units exist, but the geometry is not what you would imaging. The problem is that adding more supports really shrouds the sensing volume and kills high frequency response. So typically you just tilt the sensor at an angle, then use a rotation matrix to reorient the recorded velocities into (x,y,z) space. The true 3 axis (http://www.directindustry.com/prod/campbell-scientific-europe/3-axis-3d-ultrasonic-anemometers-25532-654011.html) units use the same trick..they just use 3 paths forming an hourglass shape focused on a central region.

        Also microbursts aren’t that micro, and by the time they hit the ground they have mostly started to spread out (the horizontal wind…actually the wind shear is the problem), so z axis resolution wouldn’t gain you much for that application…not to mention the timescale for a microburst is well within the resolution of a normal anemometer and by the time you sense it at the ground, the advanced warning period is pretty much over.

    1. For weather forecast macro measurements are usually enough, but for traffic warnings bursts are very important. For example you can have wind that blows around 60km/h, but at some moments you get 100km/h bursts that last few seconds. While 60km/h is safe for example for plane landing, 100km/h burst can cause serious problems. Same thing with cars and especially trucks on bridges.

    2. Depends on what he wants the data for. He can always downfilter to more “normal” measurements. I’d kill for a good high-frequency data corpus though for modeling wind turbine structural dynamics.

      1. AS; do you build DIY wind turbines? can i see any work you may have on show?, cos im real interested and would like to start messing around with them whenever time permits me, if you dont have anything personal, a few really good links will be great thank you. :)

    3. yes and that is why standard cup anemometers (which are generally cheaper, lower power and more rugged) are still around. Even things like microbursts are low enough resolution that a standard anemometer is capable or resolving it….although once you measure it at the ground, it’s too late to do anything about it anyway.

      The main thing the high frequency wind measurements are used for is computing the turbulent dissipation rate, which tells you how strong the local mixing is. This parameter is used for things like estimating the boundary layer height and heat/moisture/chemical fluxes to/from the surface.

  3. I saw this and immediately thought “I could make that and add to my weather station!” Then I realized I have no idea how I’d integrate this with my weather station and its software. Phooey

  4. I saw this great project a while back when I thought an ultrasonic wind speed detector would be a cool build. I was daunted by just how complex they are.

    Maybe it can be streamlined/simplified.

  5. I’ve been talking about doing exactly this for years and I’ve been to busy to develop it. WAY TO GO!! :-) Now I can just go out and build one according to this plan with only a few personal tweaks if I’m so inclined. We get a lot of hail where I live and not having a vulnerable whirling anemometer or wind direction sensor would be a real durability plus. :-)

  6. Interesting. The design would be self-calibrating if it would be possible to swap transmitter/receiver (or have 2 of each, operating constantly) quickly relative to changes in the wind.

    1. Ultrasonic anemometers are by and large crap and unreliable compared to cup and vane. An advanced solution to a problem that doesn’t exist, in my opinion. I work in air monitoring and we constantly have problems with the readings from ultrasonics vs cup and vane, and these are professional grade instruments installed and operated by professionals.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s