Forecast very nervous

[Bernd]

Hello, my name is Bernd and i am from Germany.

First i want to say thank you very much for this really great tool, i love it.
Using the Wiki i have translated my Cumulus into german.
All my friends were astonished about the nice design. (and i was proud )

I noticed that the forecast, that Cumulus shows on my website, changes
very quickly. Into one our i see several forecasts. At the same period my
wetterstation WH1080 shows only one forecast into it's display.

I have activated the option " Cumulus forecast".
If i deactivate this option, i got not forecast on my website.
Can Cumulus use the forecast that the wh1080 generates ?
How can i make the "Cumulus forecast" more constant ?

My updateintervall is 5min.


Bernd

[steve]

I don't know too much about how the forecast operates, but if one of your parameters is tending to fluctuate around the boundary between one result and another, then the forecast will change. The wind direction on Fine Offset stations is particularly bad for this. Cumulus calculates an average bearing over the last 10 minutes (and uses this in the forecast). If it is the wind direction causing the fluctuation, you can increase the period over which Cumulus calculates the average by editing cumulus.ini (with Cumulus stopped), and adding an entry to the [Station] section:

AvgBearingMinutes=x

where x is the number of minutes you want Cumulus to use.

Other parameters used in the forecast are the current pressure and pressure trend. If your pressure is currently fairly stable but fluctuating slightly, this could also cause the forecast to change. The other inputs to the calculation are the pressure "extremes" from the station settings screen.

The station doesn't make its own forecast available to software.

[nking]

I have found setting the AvgBearingMinutes=32 is a good start point. You need to add this parameter to the Cumulus.ini file located in the Cumulus programme folder. There is also another parameter FCPressureThreshold=0.2 which as I understand is setting the sensitivity to pressure change, however, I haven't had much success with altering this value.

[steve]

There is also another parameter FCPressureThreshold=0.2 which as I understand is setting the sensitivity to pressure change

I'd completely forgotten about that - thanks.

It sets the value of the pressure trend which is needed before the forecast calculation uses 'rising' or 'falling'. E.g. if you set it to 0.2, then the pressure trend value has to be greater than or equal to 0.2 before 'rising' is used, and similarly -0.2 for 'falling'. I'm surprised it wasn't any help.

It should be possible to determine whether it's the wind direction or the pressure trend which is changing the forecast by watching the display and noting the average wind bearing (it only uses the compass position) and pressure trend values, and seeing which changes when the forecast changes.

[Bernd]

Thank you Steve and Neil for your tips.
First i will try to set the bearing value because i can see that the wind direction sensor
often change it's direction, especially in low winds.
I have seen a second value were i can set the bearing to zero at windstill.
Might that in this case be helpfull too ?

Bernd

[steve]

I have seen about a second value were i can set the bearing to zero at windstill.
Might that in this case be helpfull too ?

Not as far as the forecast is concerned, as it already does a check for 'calm' and uses that instead of the direction.

[Bernd]

Does cumulus log how often it changes the forecast, into one of the log files ?
So i could look how often it changes into a given time range an i could see if
my changes take effect.

Until now it doesn't seem to make great difference that i have changed
the av bearing to 30min and also the pressure threshold to 0,2.
May be it changes more often because Cumulus makes more Differentiations as
the forecast of the station itself.

I will overserve the behavior the next time.

Bernd

[steve]

It updates the forecast every time it reads data from the station. It doesn't log anything to do with the forecast. The forecast isn't a serious facility, it's for novelty use only, I don't regard it as part of the meteorological data produced by Cumulus.

If it's changing often, it should be fairly easy to see from watching the display what is causing the change. Does your average wind speed keep dropping to zero? That would cause the forecast to change. Are you using the station average wind speed or getting Cumulus to calculate it?

[Bernd]

The forecast isn't a serious facility, it's for novelty use only, I don't regard it as part of the meteorological data produced by Cumulus.

Oh this is a great pity, because the one and only thing i do find every time interesting on the display
of my wh1080 is the forecast and it is matching pretty good.
I think a user who will visit my weather website will be interested to the forecast too, may be more than
into other weather datas.

Does your average wind speed keep dropping to zero?

The last two days it does. We haven't much wind currently.

Are you using the station average wind speed or getting Cumulus to calculate it?

I haven't configured one of them to calculate an average windspeed, exept the new line into the ini. with av 30min,
like you told me to do.

Bernd

[steve]

You could try getting Cumulus to calculate the average wind speed - and there is a corresponding setting in the ini file (which is probably documented in the wiki) to set the period over which the average is calculated. If it is indeed the wind speed dropping dropping to zero frequently that causes the forecast to change, then getting Cumulus to do the average - perhaps over a longer period - might stop it changing so much. But you really need to determine which factor it is that is causing the change.

[nking]

Here are the settings I use which are a good start point

FCPressureThreshold=0.2
AvgBearingMinutes=32
AvgSpeedMinutes=16

For the Configuration/Station setting screen I have "Calculate 10 min wind average” ticked. I have tried various pressure settings but have found 950 -1050 works best for me. In my experience the forecast has been most sensitive to the wind direction. Bear in mind that this forecast is never likely to be as good as your national weather service, they have so much more technical data to draw upon i.e. satellite, radar etc…. but it’s fun none-the-less.

[Bernd]

Thanks again Neil.

Actually i dont want to set cumulus to calculate an average wind speed, because of i'm a friend
of using windenergie and for that it is helpfull to get as many windspeed measurements as possible..
It seems that i have to think about what is more important for me.

Bernd

[Bernd]

Hmm.... i see that the entry for the pressure threshold value starts with " FC ".
I think that must mean " forecast ". ?

May be i can make out an entry for the average windspeed wich does only take effect to the
calculation of the forecast ?

For example like this : FCAvgSpeedMinutes=16
That would be very fine for me and would save my windspeed datalog interval of 1 min.


Bernd

[steve]

For that to work I would have to write all of the code to handle two separate average wind speeds, one of which was only for the forecast. And you already know what I think of the forecast. I can't add every requested change, there just isn't time - I have to prioritise. I have had a lot of complaints in the past about the forecast; I really wish I hadn't added it.

I don't really understand what exactly you're doing with the wind speed data for your wind energy project, but wouldn't the 'gust' values (the value Cumulus shows as 'latest') and the values from an average speed taken over a longer period be just as useful? The speed would still be logged at 1-minute intervals, it would just be a moving average over a longer period.

[Bernd]

I don't want to occupy you to much, i just want to find out what is possible and what isn't.
I just thought so, because of the pressure value has this " FC " in front.
If things that i ask for are not possible it's ok for me.

Steve you wrote that the speed would still be logged at 1-minute intervals.
Does this mean that into the logs (or wherever the datas are stored) the interval
is still 1 min, also i set it to (for example) 10min into station setup?
So there is the chance for me to calculate with this stored datas.

Calculations for windenergie must use as many measurements per time unit as possible.
The reason is that an average windspeed of 4m/s could be in extreme cases caused by
5 measurements with 8m/s and 5 measurements with windstill or may be caused by 10
measurements with 4m/s.
But the energetic content into both cases is highly different, although is goes in both cases
out to average 4m/s. Thats coming out to the fact that double the windspeed means 8 times
the power into the wind.

So for accurate calculations of how much power was into the wind i must have many measurements.
I was happy to notice that the wh1080 is able to deliver a snapshot every 48 seconds.

I don't know how the gust values are calculated, but i think it's not helpfull for me.

Bernd