Snow Detection ideas...IR reflectance...
peeves
Posts: 6
I was thinking I could maybe detect falling snow using something like the QTI reflective sensor aimed sideways on a highly emissive black surface with the plane perpendicular to the beam. Then when it snows, occasionally the sensor will detect closer to white than black.
The problem is I'd prefer to have the sensor/black plane distance at least ~5-10"+ so that the snow can freely fall between the 2 without turbulent effects.
Can anyone recommend a sensor that would meet my needs? It would have to be an IR reflective sensor since it needs to function at night as well.
The problem is I'd prefer to have the sensor/black plane distance at least ~5-10"+ so that the snow can freely fall between the 2 without turbulent effects.
Can anyone recommend a sensor that would meet my needs? It would have to be an IR reflective sensor since it needs to function at night as well.
Comments
On thought would be to use a line scanner. You could put the line scanner opposite a black background, and use LEDs on the side the of the line scanner to illuminate the flakes. This would give a high contrast between a flake and the background. It could also work for other types of precipitation.
John R.
I wonder if the answer might be (Disclaimer:I live in Southern Alabama where snow is pretty darn rare) to establish a reference distance for a light sensor, then back it up mechanically, measuring the point it disappears. Encoder would measure the distance. At the point where your light sensor reads nothing on a clear day, you have reference. Then work from that.
Alternative two might be to create a variable light source, and a photoreceiver set apart. Then reduce the light source in steps (PwM). If perfectly clear you should get a 1-1 ratio (correcting for distance). if snowing (or raining) you would get a 1-lessthan1 as you lost light to interference. Secchi disk in reverse. Laser diode would do nicely.
Now both will suffer from one annoyance, the sensors need cleaning. Dust, smoke, (in my case small crabs, algae and slime). So you may need to wipe them down once a month. A bit of clever programming and you could get it to tell you when it's needed.
My techie-do-list includes a water clarity meter so keep us posted on how it goes.
One simple "genreally used" way to measure water clarity (turbidity) electronically is to shine and LED (or other light source) across a gap and use a photo sensor (photo transistor) to measure the brightness. Can can be calibrated to the secchi disk, and/or in many cases just the raw data used when only a comparison is needed (the water is "cloudier today than yesterday").
For both this, and the original post, it might be worth checking out Dr. Allen's web site. I'll see if I dig up the URL.
Edit: here's the url for EM Systems (Dr. Allen's web site): http://www.emesystems.com
John R.
Led in a tube - That does work, although I've discovered that in my environment, the algae REALLY like the LED and I'm pretty sure the glass shrimp thought it was love. The stuff in my tube was, in it's own right, pretty funny.
The longer these things stayed submerged the less accurate they got. My son's answer was a servo to pull it out of the water between readings. I've thought about a test-tube brush on a servo to clean every evening.
We had to quit playing while we finished the other projects. Once school starts, time becomes critical. Electric outboard finishes this weekend, weather station in a month and then it's back to the dreaded aquarium project.
Thanks
Here's a graph of reflectance vs wavelength of different snow crystal radii.
Obviously snow scatters all visible wavelengths well in all directions so they appear white to us. Rain drops tend to back scatter light more. Therefore, if my atmospherics physics isn't too rusty, if I'm emitting visible light on one side of a horizontal segment and receiving it on the other side, more light should be received through the snow than through the rain. Obviously weather radar deals with different wavelengths, but it's the same idea...rain has a higher reflectivity (back scattering) than snow and wet hail has the highest reflectivity.
With all of that in mind, I'm still having trouble figuring out how to best make this work in my head. Do you think it would be better to try to differentiate it via back scattering (amount of light received on the same side of the emitter) or by the received being on the other side (forward scattering)? I'd probably have to experiment with the differing precip types, wavelengths, and precip intensity.
As for the line scanner...I would want a digital output from this so I can manipulate the data. I'm just learning this BASIC stamp stuff so I'm working with that microcontroller for now. From what I've read, I would need to set a voltage threshold which would analyze the analog readings as a 1 (light) or a 0 (dark). If I could somehow find that threshold which best differentiates between rain drops and snowflakes, this could work really well. I could also estimate the snow intensity by comparing the amount of "1" valued pixels in the entire 128 pixel sample.
By turning one led at a time on the reflected and transmitted intensities could be measured for each wavelength, with no leds on used as the zero reference.
-Phil
I believe the angle of refraction of visible light through a rain droplet is about 40-42° depending on the specific wavelength. So you're suggesting to place the linescan imager centered at this angle to the emitted light?
-Phil
This from a guy in Los Angeles who hasn't seen snow in years!
How about a scale with a water tight pan on top? A variable resistor aligned with the movement of the weight meter could tell you the weight of the liquid, solid or when both fell, and a clock could tell you how fast it is falling or even when it fell. The only problem is you have to dump the pan but if you put it close to a window it should not be much of a problem. In Pennsylvania snow falls at around a 30 degree angle but sometimes it looks horizontal. Oh yeah, you gotta dump pretty often.
The more I thought about it I wondered if I could make a sleet/hail sensor too using acoustics.
Does anyone think the Parallax sound impact sensor has the capability of detecting "pings" from these types of precipitation? I have no idea what kind of metal I would use (to enclose the sensor) to get the loudest sound or what kind of decibels this would even put out. A quick look at the sensor specs tells me the range of detection can be adjusted.
http://www.parallax.com/StoreSearchResults/tabid/768/txtSearch/sound/List/0/SortField/4/ProductID/614/Default.aspx
http://www.howmuchsnow.com/ping/
Yes and that's another project. :smilewinkgrin:
That is simply for snow depth though and not for actual detection on if it is falling or not. I've had a few discussions with Chris on that sensor. He's quite the weather gadgets guru.
http://www.crh.noaa.gov/grr/snowsensor/
An ultrasonic sensor might work if the snow flake is big enough...This is a cool idea and it would be very useful, Especially for someone who lives in New Hampshire. We get a LOT of snow up here...