Weather station - wind speed/rain fall

[gregmcc]

Im busy building a arduino based weather station and have got myself a rain gauge and a wind anemometer.

The rain gauge is a bucket system which triggers a reed switch and according to the data sheet 1 bucket = 0.2794mm of rain.

The anemometer works similarly where a wind speed of 2.4km/h causes once switch closure per second.

Its all wired up and I can detect the triggering of the reed switches but now comes actually part of calculating the rainfall/wind speed which is where I'm getting lost.

For both would the best be to count the number of pulses in 60 secs and then display wind speed for that 60 sec period, and also display mm of rain in 60sec? I think the more samples takes the better - should I maybe count them for 3 mins?

How do these commercial units achieve this? Do they count the number of pulses? Take the time between pulses?

 

[3v0]

I know you already have the sensors but if you ever need to replace them take a look at the stuff here
**broken link removed**

Most weather sensing is very low speed.

For rain fall you simply count the number of times the bucket tipped for total rainfall. If you want to know the cm/minute or hour you would count the tips per minute or hour.

For wind speed you can do it either way or both. We are often interested in "winds gusting to " and period measurement would be best for that. Average wind speed would be based on counts per unit time.

Humidity sensors are easy to add. Honeywell makes several that are popular.

 

[gregmcc]

Interesting site.

Thanks - I think I'll just sample for 60 secs then. Phase 2 can be the wind gusts

 

[dr pepper]

60 secs is a long time, you'll certainly miss gusts at that.
Bicycle computers measure the amount of time between switch contact makes, you could do something similar, or maybe measure the time between pulses from the anemometer up to a certain wind speed, then over that measure the frequency, you know for instance that 10kmh gives about 4.2hz.

To make sure you dont miss pulses from the rain guage either use interrupts or poll the input at a high enough rate, which is easy for most micro's, as the input is very slow.

I built a weather station way back, the 6502 was the processor of choice then.

 

[Vizier87]

Hi, this may or may not help, but you can take a look at my blog about disdrometers.

Cheers.

 

[dr pepper]

Cool.
Never heard of a disdrometer, simple and effective, so it is possible to make a completely no moving parts weather station.

 

[Vizier87]

Cool.
Never heard of a disdrometer, simple and effective, so it is possible to make a completely no moving parts weather station.

Yup. A WXT510 has a disdrometer but it cost a bomb though if I were to purchase it through my country.

 

[dr pepper]

I helped a mate build a rain detector for his remote observatory, it uses stripe contacts and has a heater to dry of rain, overall reasonably complictated and requires maintenance, if it ever needs replacing I'll lokk into this idea, sounds as though with a microcontroller a rain detector based on this idea would run for years without any maintenance.

 

[Vizier87]

I helped a mate build a rain detector for his remote observatory, it uses stripe contacts and has a heater to dry of rain, overall reasonably complictated and requires maintenance, if it ever needs replacing I'll lokk into this idea, sounds as though with a microcontroller a rain detector based on this idea would run for years without any maintenance.

Cool. But heaters? Yuck.

My project is discontinued for now, but I am extremely interested in continuing it based on the feedback on my article. In Malaysia we don't need to do weather monitoring since it's all the same throughout the years though, but we have plenty of rain for data collecting. In fact some the commercial rain gauges were tested primarily here.

An equivalent yet cheap alternative WXT510 would be acting like a beacon for electronics hobbyists... I've yet seen people construct a proper solid-state anemometer with hobbyist materials excluding the ones done by Prof Liljencrants.

If you plan on doing one, please update in a blog. I'd love seeing more development.

Cheers.

 

[dr pepper]

I did experiment with an ultrasonic anemometer, I wasnt happy with the performance so didnt use it, there are issues with ambient temperature and humidity affecting the speed of sound, mixed in with transducer variances makes it difficult to make something accurate enough to be worth using, also omnidirectional anemometers using 4 transducers requires some fairly fancy algorithms.
A unidirectional one wouldnt be so hard to build but not much use.

 

[Vizier87]

I did experiment with an ultrasonic anemometer, I wasnt happy with the performance so didnt use it, there are issues with ambient temperature and humidity affecting the speed of sound, mixed in with transducer variances makes it difficult to make something accurate enough to be worth using, also omnidirectional anemometers using 4 transducers requires some fairly fancy algorithms.
A unidirectional one wouldnt be so hard to build but not much use.

I see. I've built a thermal one but I haven't written anything about it though. The results are not that impressive. Have you tried Doppler Laser type? I read a paper about it but can't understand it though.

 

[dr pepper]

Now thats something I havent heard of, I'll research that when I'm next working nights.

 

[Vizier87]

Now thats something I havent heard of, I'll research that when I'm next working nights.

There's plenty of the mentioning part online but as for documented DIY... zilch. Here's the earliest one I know of:
https://www.sciencedirect.com/science/article/pii/0030399274900061

But whether it can be applied for weather monitoring is another question I suppose.