How to calculate mean wind direction

Hello!

I need help figuring out how to calculate mean wind direction when my data is in degrees (0-360). I just realized my current program does not take into account that the data is circular, and the mean of 355 and 5 will be 180, instead of 0. Any help is greatly appreciated! I am a beginner when it comes to MATLAB programming

[José-Luis]

Instead of 0, you mean?

[Jenna Marie]

Ah yes thank you for catching that! Accidentally mixed up 360 degrees in a circle with 365 days in a year. It's been a long day....

Answer 1

[José-Luis]

 average = mod(sum(data),360)./numel(data)

Please accept an answer if it helped you.

[?]

180 and 180 mean is 0?

[Ana Soph]

i can stablished for 1 minute to 10 minutes data, how can i do that?

[Ana Soph]

How can convert to 1 minute to 10 minutes, a wind directio data ? please, i have a matrix fo 1440x31

[giacomo labbri]

I tried to use the proposed function but for me it didn't work. Here is the function I wrote instead (based on https://en.wikipedia.org/wiki/Mean_of_circular_quantities

function [windir_avged] = windir_avg(windir) 
  windir_avged=180/pi*angle(sum(exp(1i.*windir.*pi/180))/numel(windir));
  if windir_avged<=-0.001 %it should be 0 but the it would inclde some slightly negative numbers that are just due to numerical precision
      windir_avged=abs(windir_avged)+180;
  end
 end

The idea is to look at angles as vector of unit lenght and treat these as complex numbers

[Ana Soph]

THANK YOU SO MUCH! :d

[Ana Soph]

how can i calculate for rows ?

[Ana Soph]

windir are my data ?

[Walter Roberson]

The below version handles arrays.

function [windir_avged] = windir_avg(windir) 
  windir_avged=180/pi*angle(sum(exp(1i.*windir.*pi/180))/numel(windir));
  mask = windir_avged<=-0.001;  %it should be 0 but the it would inclde some slightly negative numbers that are just due to numerical precision
  windir_avged(mask) = abs(windir_avged(mask))+180;
 end

[Naseef Muhammed]

According to your script, average of 345 and 355 degrees are 190 !

[Walter Roberson]

@Naseef Muhammed

It is not clear which script you are referring to?

[Naseef Muhammed]

@Walter Roberson I'm refering to the function 'windir_avg' you provided above. In that function, if we given 2 anlges from third or fourth, it will give a wrong answer. for example, if we put 345 and 355 the answer will be 190. similarly, 240 and 250 will result in 295. I hope it can be resolved if we edit your script as below:

function [windir_avged] = windir_avg(windir)
windir_avged=180/pi*angle(sum(exp(1i.*windir.*pi/180))/numel(windir));
if windir_avged < 0
    windir_avged=windir_avged+360;
end

[Walter Roberson]

I assumed that @giacomo labbri provided the correct formula in https://www.mathworks.com/matlabcentral/answers/128381-how-to-calculate-mean-wind-direction#comment_896547

Answer 2

[Jenna Marie]

Thank you so much for your response!

What does numel mean?

[José-Luis]

No worries.

 doc numel

Counts the number of elements in the matrix.

Answer 3

[Walter Roberson]

See unwrap() but you will need to convert to radians

Answer 4

[Robert Daly]

I needed a solution that would ignore NAN values in the data.

Converts the direction data into X & Y vector components, averages those, then converts back to direction.

function [windir_avged] = windir_avg(windir)
    [x,y] = pol2cart(deg2rad(windir),ones(size(windir)));
    x=mean(x,'omitnan');
    y=mean(y,'omitnan');
    [windir_avged,~]=cart2pol(x,y);
    windir_avged = rad2deg(windir_avged);
end

[Soeren Bilges]

Preferred and robust solution, thanks.