Page 1 of 1

Wind hourly mean

Posted: Sun 02 Jun 2013 10:52 am
by ws2080jo
one of my objectives in graduation project to calcluate hourly mean of wind (speed and dominating direction) /hour , I found that cumulus has 10min-average , how could I calcuate the hourly mean speed/direction from the daily log file , could anyone help will be appreciated?


Regards

Re: Wind hourly mean

Posted: Sun 02 Jun 2013 11:08 am
by steve
is that a rolling 60-minute average, or 24 average figures for each hour of the day?

If the latter, you would have to process the data log files, extracting the entries for each hour, and taking an average of all of the wind speed values in each set of entries.

If the former, and you were willing to 'lose' the 10-minute average, you could get Cumulus to do a 60-minute average by editing cumulus.ini and adding a line to the [Station] section:

AvgSpeedMinutes=60

Re: Wind hourly mean

Posted: Sun 02 Jun 2013 11:15 am
by ws2080jo
much appreciated AvgSpeedMinutes=60 is what I want

thanks alot

Re: Wind hourly mean

Posted: Sun 02 Jun 2013 4:23 pm
by ws2080jo
Is it correct for dominant wind direction / hour to average wind degrees. From log?

Re: Wind hourly mean

Posted: Sun 02 Jun 2013 4:56 pm
by mcrossley
The sound of a can of worms being opened! :lol:

A number of options as I see it.

1. Mean vector: You treat the wind as a vector and take into account the speed as well as direction. So if it was blowing North 50% of the time at 5mph and East the other 50% at 10mph, then the dominate direction would not be NE, it would be skewed to the East. This way you end up with a mean of the vectors, and if the directions do not vary much it is meaningful. But if say it was blowing NNE 49% and the time, and SSE the other 51% then the result will be almost East, but the wind never really blew from the east!

2. Time based bins: You sum the time the wind spends blowing in each direction and the dominant wind is the largest bin value - SSE in the second example above.

3. Binned vectors. You the sum vectors in each direction bin and the largest value wins. So like the time based, but you weight each value by the speed.

4. Time based average: You take a simple average all the directions.

5. There are probably other ways of doing as well!

Re: Wind hourly mean

Posted: Sun 02 Jun 2013 5:02 pm
by steve
mcrossley wrote:1. Mean vector: You treat the wind as a vector and take into account the speed as well as direction.
This one is how Cumulus does the average bearing.

You can use the setting AvgBearingMinutes=60, and then average bearing is also worked out over 60 minutes (using the above method).

Re: Wind hourly mean

Posted: Sun 02 Jun 2013 7:17 pm
by ws2080jo
Very useful helped me a lot thanks all for support and cumulus help us a lot in our grad project

Thanks

Re: Wind hourly mean

Posted: Wed 05 Jun 2013 4:28 pm
by ws2080jo
I perform this to day log file for 1 hours

Ve = Σ [wspd* sin(wdir)]/N
Vn = Σ [wspd*cos(wdir)]/N

UV = (Ve2 + Vn2) 1/2
AV = ArcTan(Ve/Vn)

but I'm getting values like -0.954 , should I do other operation to find wind degree

Thanks for cooperation

Re: Wind hourly mean

Posted: Wed 05 Jun 2013 4:44 pm
by mcrossley
Without checking anything, I think you will need to use an ArcTan2() function to find the correct quadrant

Re: Wind hourly mean

Posted: Wed 05 Jun 2013 4:51 pm
by ws2080jo
one of my colleagues telling me that I should convert wdir degree to radian then final result in rad convert it to degree and if less than 0 add 180? it that correct?

Re: Wind hourly mean

Posted: Wed 05 Jun 2013 4:59 pm
by mcrossley
I don't know for sure because I don't know what language you are using, but normally maths functions do take and return values in radians. I would still use an arctan2 function though.