If you use the TSL1401 in this way, make sure the black stake is wide enough that the alignment with the photo array is not critical. However, if you get the kind of wet blowing snow that clings to everything it touches, your stake will soon disappear.
I would try using a laser line generator arrayed perpendicular to the TSL1401 array. The linescan camera would be aiming downward, with the laser line projected from an angle to the side. That way, the line will appear to move as the snow changes depth. I would also take two exposures: one with the laser on; the other, with it off. Then subtract one from the other, leaving just the laser response. Finally, use a red filter over the TSL1401's lens. That will eliminate all but the laser color and improve your optical signal-to-noise ratio.
Yesterday was the first time this winter season I had an opportunity to be home during the day when the forecast called for measurable snow (3-5"). Quite a change from last year! I had the Ping sensor rigged up and ready for a trial run since December.
I positioned the Ping about 23" from a board surface and was getting a very consistent measurement time for a full hour before the snow started to fall. I check it again when the snow depth got to about 1/2" but the measurement time (distance) remained unchanged. Fairly soon after that the measurements became erratic. By the time the snow reached a depth of about 1" the sensor was no longer responding. I cleared the snow off the board and the sensor's response returned to its original level and steady.
So what I concluded was snow is not a good acoustical reflector and the Ping sensor (at least at a distance of a couple of feet) is not likely to work. Looks like I will have to investigate some of the alternative methods of sensing snow depth.
Parallax had some liquid sensors in form of a tape strip. I was looking at them for my water tank, but they where way to short for a 8 ft. high tank.
Basic function is like a variable resistor or potentiometer, so you need to use a RCtime circuit or a ADC.
Could be worth a try, since snow is basically water...
Then there is that sharp infrared sensor instead of ping. Not sure if it will work with snow.
How about a mechanical solution. Say lowering a weight on a thin rope until it hits the surface of the snow, and measure the rotations of the wheel with a encoder?
Or some whisker like thing moved by a servo/encoder stopping when hitting the snow?
I remember some post of @Phil, using two lasers mounted at a angle and a camera to measure the distance between the two spots to calculate the water depth below some robotic water vehicle he build. Pretty sure it was @Phil.
That should work with snow, red dots on white surface. Simple laser pointer and a PropCam if there are some left still.
They're cheap, and based on surface reflectivity, not acoustics. 150cm would give you good range, but if that's not enough you could go to the 150 to 500cm version instead.
They're cheap, and based on surface reflectivity, not acoustics. 150cm would give you good range, but if that's not enough you could go to the 150 to 500cm version instead.
I tried that and had to give up. It's got the range but its accuracy doesn't cut it. They claim 10% but I was getting more like 15%. With the sensor 40" off the ground, that makes an error of 5-6". When it's just starting to snow is when you want the most accuracy, so this isn't acceptable.
Next I tried the Parallax LaserPING #28041. In testing indoors, I was getting great accuracy. It's advertised as 7% but I was getting more like 2%. But when moved outdoors, for some reason it was way off, reporting distances greater than the actual distance to the ground.
And it doesn't work at all while it's snowing, which is when you're most interested in how things are going ("Wow, it's really coming down now!").
Next I think I'll try the TLS1401 linescan sensor.
I also had an idea to have a clear pipe with IR sensors every 1/2" mounted vertically and a nearby pipe with LED's in it. By strobing the sensors, the first one that shows a different response should be the depth of the snow. That should work even while it's snowing.
This is proving to be a much harder problem than I initially thought.
So I come to this thread and find there is text in the comment box I did not post before. It goes like this:
"I think you should weigh it.
Put a one meter square platform outside.
With load sensors under it.
The load tells you the weight of snow that has landed.
"
"I think you should weigh it.
Put a one meter square platform outside.
With load sensors under it.
The load tells you the weight of snow that has landed.
"
Unfortunatey, weighing it won't give you an accurate height.
Weighing is fine if you want to know how much it weighes (and there are plenty of reasons to want to know that). But as anyone who has ever shoveled a driveway knows, snow comes in all sorts of weights, from light fluffy skiier's delight to heart attack slush.
If you want to try the IR sensors, I would put a column of LED all pointing to one IR receiver at the top of the second pipe.
Hmm. You are absolutely right. That's even simpler. For that matter, why not just put one receiver in a clear waterproof box at ground level pointing up, and one transmitter up above pointing down. Measure how much light gets through. But that might not work so well during the day in sunlight. And the weight of the snow might matter too. Sigh.
I also had an idea to have a clear pipe with IR sensors every 1/2" mounted vertically and a nearby pipe with LED's in it. By strobing the sensors, the first one that shows a different response should be the depth of the snow. That should work even while it's snowing.
Curtain type sensors I would expect to be very tolerant of variations in snow type, and if you measured analog light transmission, you could get away with fewer sensor nodes.
LED strips are widely available, but strips of detectors not so easy to source.
- perhaps an LED strip and a single camera ?
The LED strip could even show the snow peak, for user feedback, when it is not actively driving the camera-capture
Addit: Not so visible, but with the advantage of a single mounting tube, could be a reflective sensor PCB design, using a sensor like this
Faces out into the snow, and relies on snow present = much more reflected light back into the tube.
One tinylogic 1G79 D-FF per sensor, allows a serial-clocked readback.
An idea from an old fisherman that usually doesn't goes fishing as often as he desires to.
Depart from the basic idea of a fishing rod with an electric-driven fishing reel, some black-colored braided line and a lead-loaded pyramidal sinker.
As for the sinker, the darker its surface color, the best. Cover with black paint, for better results, and don't forget to mount it with its base facing down, so as to snow doesn't pile over its surface.
You'll also need two extra parts (the cheapest, the best): an extra spool (almost any spool will be useful) and any rotary encoder you could find, inside some hiden and forgoten shelves. Couple them and mount at one end of the "fishing rod", the farther one, from the pivoting axis.
The line would be of any black-colored braided type. I would recomend nylon, but any kind of braided line that could be used to twist some few turns over the secondary spool, without sliping, will do de job.
Any suitable stick (wood, rigid metal/plastic) that you could drill a pitoting hole, will turn into your balancing fishing rod.
The "rod" dimensions will be based on its stiffness and the space needed to accomodate all the components.
Don't forget to include some controller board (almost any one could do the job), any end-of-course sensor you have and a suitable power suply.
The rest will be up to you to decide.
As usual, an image worth a thousand words.
Good fishing.
Henrique
P.S. My fault. I've forgot to design a pyramidal-shaped roof, at the top surface of the motor/controller box. We usually don't have snow at the tropics, at least not in Rio de Janeiro. :cool:
P.S. 2 - I soon became concerned that snow melting and subsequent ice formation could impair the whole setup.
Maybe a reversed U-shaped rain gutter could be used to cover the fixture.
Frequently pulling and releasing the line to take measurements could help in avoiding ice formation (except, possibly, during ice storms, when the weather conditions could be too severe).
The absence of snow in my city is limiting my ideas.
That's what I'm using right now. In my indoor experiments, I found it to be very accurate at the range I need (around 40"). But once I moved it outdoors, I quickly discovered it only works at night. During the daytime, it thinks the ground is 80" away, and that's not even in direct sunlight. And it also fails while snow is falling, day or night.
I've seen other people talking about using ultrasonic or laser sensors. I don't know how they do it. I haven't been able to get either the Sharp IR sensor or the LaserPING to work for me.
Here is a article that is quite comprehensive on the subject: howmuchsnow.com/snow/
It mentions Parallax Ping sensor
Yup, I've seen that too - lots of good info there. I don't know how he makes that work. My LaserPING is shrouded and pointing straight down but daylight swamps it.
Next I think I'll try the TLS1401 linescan sensor.
A camera has appeal, as their price is falling all the time.
Linescan is more of a niche, as the volumes are very low.
...
Faces out into the snow, and relies on snow present = much more reflected light back into the tube.
One tinylogic 1G79 D-FF per sensor, allows a serial-clocked readback.
The appeal of the linescan camera is its simplicity and low power draw. My weather station runs on a solar panel and when I'm interested in snow is precisely the time of year when there's the least sun.
A regular camera pointed at a yardstick and a bit of image processing might in the end be the most accurate but I'm using a Particle Photon as a controller and I don't think it has the grunt needed for that.
I do like the one-tube reflectivity sensor idea. That sounds viable.
All these things need to be tried out, and unfortunately in the past week all the snow we had has melted. Hopefully we'll get at least one more blast before spring sets in.
...
Frequently pulling and releasing the line to take measurements could help in avoiding ice formation (except, possibly, during ice storms, when the weather conditions could be too severe).
The absence of snow in my city is limiting my ideas.
Yes, it never snows when you need it :-) This idea also sound like it would work too. However I really want to avoid mechanical solutions because as you point out they're prone to freezing up. I want something that's going to be maintenance-free.
I'm glad to see this thread has attracted renewed interest.
What is the significance of the snow-depth measurement? How high it piles up, regardless of how fluffy or dense? Or how much water it contains? If the latter, a heated funnel feeding into a standard rain gauge might be the simplest solution.
What is the significance of the snow-depth measurement? How high it piles up, regardless of how fluffy or dense? Or how much water it contains? If the latter, a heated funnel feeding into a standard rain gauge might be the simplest solution.
No, the former. I want to impress my friends in Florida with how much snow we got :-) Actually, water content would be interesting to know too, both as a matter of intellectual curiosity and to know if I should get out the snowblower or if I can just shovel the stuff. But my solar panel definitely won't support a heater.
What is the significance of the snow-depth measurement? How high it piles up, regardless of how fluffy or dense? Or how much water it contains? If the latter, a heated funnel feeding into a standard rain gauge might be the simplest solution.
No, the former. I want to impress my friends in Florida with how much snow we got :-) Actually, water content would be interesting to know too, both as a matter of intellectual curiosity and to know if I should get out the snowblower or if I can just shovel the stuff. But my solar panel definitely won't support a heater.
No need to melt the snow to calculate the water content if you can find a way to measure the depth. Just accumulate it on a platform of known area and weigh it.
No need to melt the snow to calculate the water content if you can find a way to measure the depth. Just accumulate it on a platform of known area and weigh it.
But then you need a way to clear the platform before each new snowfall.
No need to melt the snow to calculate the water content if you can find a way to measure the depth. Just accumulate it on a platform of known area and weigh it.
But then you need a way to clear the platform before each new snowfall.
Not necessarily. If you are constantly tracking depth and weight the numbers will tell you when when snow is being compacted and evaporating. That would also provide some additional data of interest.
Comments
I would try using a laser line generator arrayed perpendicular to the TSL1401 array. The linescan camera would be aiming downward, with the laser line projected from an angle to the side. That way, the line will appear to move as the snow changes depth. I would also take two exposures: one with the laser on; the other, with it off. Then subtract one from the other, leaving just the laser response. Finally, use a red filter over the TSL1401's lens. That will eliminate all but the laser color and improve your optical signal-to-noise ratio.
-Phil
I positioned the Ping about 23" from a board surface and was getting a very consistent measurement time for a full hour before the snow started to fall. I check it again when the snow depth got to about 1/2" but the measurement time (distance) remained unchanged. Fairly soon after that the measurements became erratic. By the time the snow reached a depth of about 1" the sensor was no longer responding. I cleared the snow off the board and the sensor's response returned to its original level and steady.
So what I concluded was snow is not a good acoustical reflector and the Ping sensor (at least at a distance of a couple of feet) is not likely to work. Looks like I will have to investigate some of the alternative methods of sensing snow depth.
Here are few images from yesterday's test...
Basic function is like a variable resistor or potentiometer, so you need to use a RCtime circuit or a ADC.
Could be worth a try, since snow is basically water...
Then there is that sharp infrared sensor instead of ping. Not sure if it will work with snow.
How about a mechanical solution. Say lowering a weight on a thin rope until it hits the surface of the snow, and measure the rotations of the wheel with a encoder?
Or some whisker like thing moved by a servo/encoder stopping when hitting the snow?
I remember some post of @Phil, using two lasers mounted at a angle and a camera to measure the distance between the two spots to calculate the water depth below some robotic water vehicle he build. Pretty sure it was @Phil.
That should work with snow, red dots on white surface. Simple laser pointer and a PropCam if there are some left still.
my 2 cents,
Mike
They're cheap, and based on surface reflectivity, not acoustics. 150cm would give you good range, but if that's not enough you could go to the 150 to 500cm version instead.
Next I tried the Parallax LaserPING #28041. In testing indoors, I was getting great accuracy. It's advertised as 7% but I was getting more like 2%. But when moved outdoors, for some reason it was way off, reporting distances greater than the actual distance to the ground.
And it doesn't work at all while it's snowing, which is when you're most interested in how things are going ("Wow, it's really coming down now!").
Next I think I'll try the TLS1401 linescan sensor.
I also had an idea to have a clear pipe with IR sensors every 1/2" mounted vertically and a nearby pipe with LED's in it. By strobing the sensors, the first one that shows a different response should be the depth of the snow. That should work even while it's snowing.
This is proving to be a much harder problem than I initially thought.
"I think you should weigh it.
Put a one meter square platform outside.
With load sensors under it.
The load tells you the weight of snow that has landed.
"
As the LEDs get covered the sensor should no longer "see" them. Assuming IR will not go through snow (which it may for some distance).
You may actually be able to interpolate by varying the power and/or dutycycle to the LEDs.
Bean
Weighing is fine if you want to know how much it weighes (and there are plenty of reasons to want to know that). But as anyone who has ever shoveled a driveway knows, snow comes in all sorts of weights, from light fluffy skiier's delight to heart attack slush.
There is a good paper on the optical properties of snow here: https://atmos.washington.edu/~sgw/PAPERS/1982_RGSP.pdf
It drifts in the wind, it compacts to ice. I can even estimate the temperature from the grindy or squeaky sound it makes as you walk on it.
Question might be then, what actually are you trying to measure?
-Phil
It mentions Parallax Ping sensor
A camera has appeal, as their price is falling all the time.
Linescan is more of a niche, as the volumes are very low.
Curtain type sensors I would expect to be very tolerant of variations in snow type, and if you measured analog light transmission, you could get away with fewer sensor nodes.
LED strips are widely available, but strips of detectors not so easy to source.
- perhaps an LED strip and a single camera ?
The LED strip could even show the snow peak, for user feedback, when it is not actively driving the camera-capture
Addit: Not so visible, but with the advantage of a single mounting tube, could be a reflective sensor PCB design, using a sensor like this
Faces out into the snow, and relies on snow present = much more reflected light back into the tube.
One tinylogic 1G79 D-FF per sensor, allows a serial-clocked readback.
An idea from an old fisherman that usually doesn't goes fishing as often as he desires to.
Depart from the basic idea of a fishing rod with an electric-driven fishing reel, some black-colored braided line and a lead-loaded pyramidal sinker.
As for the sinker, the darker its surface color, the best. Cover with black paint, for better results, and don't forget to mount it with its base facing down, so as to snow doesn't pile over its surface.
You'll also need two extra parts (the cheapest, the best): an extra spool (almost any spool will be useful) and any rotary encoder you could find, inside some hiden and forgoten shelves. Couple them and mount at one end of the "fishing rod", the farther one, from the pivoting axis.
The line would be of any black-colored braided type. I would recomend nylon, but any kind of braided line that could be used to twist some few turns over the secondary spool, without sliping, will do de job.
Any suitable stick (wood, rigid metal/plastic) that you could drill a pitoting hole, will turn into your balancing fishing rod.
The "rod" dimensions will be based on its stiffness and the space needed to accomodate all the components.
Don't forget to include some controller board (almost any one could do the job), any end-of-course sensor you have and a suitable power suply.
The rest will be up to you to decide.
As usual, an image worth a thousand words.
Good fishing.
Henrique
P.S. My fault. I've forgot to design a pyramidal-shaped roof, at the top surface of the motor/controller box. We usually don't have snow at the tropics, at least not in Rio de Janeiro. :cool:
P.S. 2 - I soon became concerned that snow melting and subsequent ice formation could impair the whole setup.
Maybe a reversed U-shaped rain gutter could be used to cover the fixture.
Frequently pulling and releasing the line to take measurements could help in avoiding ice formation (except, possibly, during ice storms, when the weather conditions could be too severe).
The absence of snow in my city is limiting my ideas.
That's what I'm using right now. In my indoor experiments, I found it to be very accurate at the range I need (around 40"). But once I moved it outdoors, I quickly discovered it only works at night. During the daytime, it thinks the ground is 80" away, and that's not even in direct sunlight. And it also fails while snow is falling, day or night.
I've seen other people talking about using ultrasonic or laser sensors. I don't know how they do it. I haven't been able to get either the Sharp IR sensor or the LaserPING to work for me.
Yup, I've seen that too - lots of good info there. I don't know how he makes that work. My LaserPING is shrouded and pointing straight down but daylight swamps it.
A regular camera pointed at a yardstick and a bit of image processing might in the end be the most accurate but I'm using a Particle Photon as a controller and I don't think it has the grunt needed for that.
I do like the one-tube reflectivity sensor idea. That sounds viable.
All these things need to be tried out, and unfortunately in the past week all the snow we had has melted. Hopefully we'll get at least one more blast before spring sets in.
Yes, it never snows when you need it :-) This idea also sound like it would work too. However I really want to avoid mechanical solutions because as you point out they're prone to freezing up. I want something that's going to be maintenance-free.
I'm glad to see this thread has attracted renewed interest.
-Phil
No need to melt the snow to calculate the water content if you can find a way to measure the depth. Just accumulate it on a platform of known area and weigh it.
Not necessarily. If you are constantly tracking depth and weight the numbers will tell you when when snow is being compacted and evaporating. That would also provide some additional data of interest.