|
|
Post by offgrid on Sept 7, 2021 15:05:29 GMT -8
A problem with wind gust readings
The value of the wind gust is bouncing all over the place, from a realistic 60km/h to 534km/h from the mysql database it looks like typically 4 in 10 are in the crazy range (>100km/h with average wind speed around 35km/h) My instinct is to damp the sensitivity by increasing the sample period. Not sure where is best to do it; in SDL_Pi_WeatherRack.py line: 280 SDL_Pi_WeatherRack._shortestWindTime = 0xffffffff and/or 347 SDL_Pi_WeatherRack._shortestWindTime = 0xffffffff ? why duplicated? or 348 time=latestTime/1000000.0 Your guidance would be appreciated ... & yes this is a windy location
Nga mihi (thanks in Maori) Charles
|
|
|
|
Post by offgrid on Sept 8, 2021 21:12:37 GMT -8
ok, I have worked more on this. to be realistic the SDL_Pi_WeatherRack._shortestWindTime should be >0.015sec (that's 4,000 rpm !!! wow for 160km/h  am I right???) ..... but to simply limit the value to >0.015 would simply mean that instead of getting gusts of 500km/h the max would be 160km/h, but still wrong. So .... do I have to take the mode of say the last 10 minimum intervals? Best place to do that would be SDL_Pi_WeatherRack??? Even if I get a reasonable number am I just masking the actual problem of around 40% of gust readings being way too high?? Guidance would still be appreciated. The above does not really make much sense to me. Nga mihi, Charles |
|
|
|
Post by offgrid on Sept 9, 2021 19:08:30 GMT -8
Attached is graph from weatherunderground. While some of the crazy gusts are reflected in the MySql data not all are - my mistake in previous post; I only checked that the crazy values existed - in the MySql file the crazy values are just under 5%.
What is also noticeable in the graph - and this is definitely present in the mysql data is that the predominance of gust reading between 55 and 65 km/h (46% of all). As per the chart attached there is the expected rough correlation between gust and wind speed, BUT here only in the lower speed range. At some point the gust speed seems to max out around 60km/h
So I still have problems with the anemometer readings ... some crazy high gusts (not as many as I first thought) and the peaking of the gust speed; except for the crazy values.
Nga mihi Charles
|
|
|
|
Post by SDL on Sept 10, 2021 9:44:46 GMT -8
offered,
Are you using a WeatherRack? And on what computer and system? I suspect that interrupts are occurring in the system that is screwing up the timing values.
BP
|
|
|
|
Post by offgrid on Sept 10, 2021 13:02:05 GMT -8
Thanks for the reply. I have only recently got round to commissioning the system - covid lockdown has it's benefits  Using the SkyWeather system with the WeatherRack supplied. Computer is a Raspberry Pi with the software is as supplied on the SD card, and I did run 'update' Nga mihi |
|
|
|
Post by offgrid on Sept 10, 2021 17:21:51 GMT -8
Not much help with the random crazy high values, except that I have not seen any when running the test. What I have observed is the apparent limiting of the gust speed to around 60km/h (37.3mph) and have observed the wind speed to exceed the gust speed (unfortunately I did not record that)
nga mihi Charles
|
|
|
|
Post by SDL on Sept 13, 2021 14:38:30 GMT -8
Charles,
What kind of a raspberry pi are you running?
BP
|
|
|
|
Post by offgrid on Sept 13, 2021 18:26:27 GMT -8
Raspberry Pi 2
|
|
|
|
Post by SDL on Sept 15, 2021 12:05:29 GMT -8
Do you have a faster one available to try?
BP
|
|
|
|
Post by offgrid on Sept 15, 2021 13:18:22 GMT -8
I have a 3B+ and I will try it - weather permitting, the station is exposed and its raining & blowing for a few days.
I bought the system to learn and get back into programming as much as the need for weather info (we have 2 w/underground stations with 5km) and I am learning! I think I have the anemometer driver more or less sussed. I will run a routine which dumps the raw data and output (speed & gust) and see if anything emerges.
I have observed that my wind speed readings are consistently around 1/2 that of the adjacent stations. As we are more exposed I expect to have higher wind speeds, but even in very low winds the reading may be high,so when the weather permits I will also check the pulses/rev as I noted in one post someone had an anemometer that produced 2/rev.
noho haumaru, kia kaha (maori for stay safe, be strong)
|
|
|
|
Post by offgrid on Sept 15, 2021 14:48:50 GMT -8
correction my wind speed 2x nearby stations, theirs is 1/2 mine not other way round
|
|
|
|
Post by SDL on Sept 17, 2021 7:05:20 GMT -8
Sounds like you are getting two interrupts when you should be just getting one. I'm not sure why this is, but you might put a divide by 2 in the code to fix this.
BP
|
|
|
|
Post by offgrid on Sept 22, 2021 15:23:41 GMT -8
Update:
Good news; checked pulses per revolution and get only one, so it looks like we really do have 2x the wind speed of our neighbours, but (now confirmed) only in some wind directions.
Still left with wind gust speed problem of most gusts capped at 60km/h plus crazy readings such as 600km/hr.
Still plan to log pulse and interrupts and try with Pi Zero.
|
|
|
|
Post by SDL on Sept 22, 2021 18:26:01 GMT -8
|
|
|
|
Post by offgrid on Sept 26, 2021 19:45:42 GMT -8
I have the 1 click/sec = 2.4m/h. ... but it does not relate to the apparent capping of gusts at 60km/h, nor to the frequent crazy readings of up to 600km/h (which I see as the the de-bounce limit).
Something is clearly creating the crazy readings - I can understand bounce or a stray interrupt, but in some ways more of a challenge
is what is clearly limiting the 'normal' gust readings to around 60km/h as can be seen in the image. I can't see anythiing in the software to account for the clipping
Kia pai tō rā (Maori for 'have a nice day') Charles
|
|
am I right???) ..... but to simply limit the value to >0.015 would simply mean that instead of getting gusts of 500km/h the max would be 160km/h, but still wrong.
