"stacked" polar plot

I have a time series of velocities (speed and direction). I would like to do a polar plot where:

• The radial distance from the center is the speed of a given time point

• The angle is the vector angle of that time point

That part is easy. But I would like to do histograms, with the data binned by radial distance (i.e., binned by speed), and by angle, with the color of the bin proportional to the number of points in that bin.

So the plot would potentially have multiple sectors at different radial bins within an angle range, and each sector would have a color given by the number of points in that sector/radial bin.

Basically, it would be doing a histogram of data in a regular polarplot of the data (individual points), but the histogram would be in radial/angular bins, with each bin colored by the number of points in that bin.

Note: this is NOT the same as the radial distance giving the frequency of occurrence of that angle (i.e. polarhistogram). I need the radial distance to be the speed (in bins, stacked with greater speeds being at larger radii).

Thanks for your help!

[Peter Franks]

I guess I'm not explaining myself clearly. Here's an example of the kind of thing I'm after (but not exactly the same):

The difference is that I want the color to be the number of data points in a particular bin (or the frequency of that bin), and the radial distance to be the speed. But I don't know how to stack the different-colored "sectors" (bins in radius and angle - the blue and green sectors) on top of each other, from the center, moving outward.

So again: in my plot I want the radius to be speed, with the data binned by speed, and the color to be proportional to the number of points in a speed bin at a particular angle range.

[Cris LaPierre]

Can you share your data as well as the code you have written? It is much easier if you provide something to work with.

[Peter Franks]

The data are pretty simple: speed varies from 0 to something, and angle goes between 0 and 360 degrees (2pi radians). Here's some code to generate reasonable values.

speed = rand(1,1000);

vangle=rand(1,1000)*pi*2;

I have not written any code because I'm stumped as to how to make stacked plots (like the figure I included above, which I downloaded from the web).

I'll point out that wind roses (like the one in my previous post) are standard plots in many fields, and there are MATLAB programs to make them. I'm trying to do something different: basically switching the speed and the frequency. The speed will be binned radially, and the frequency will be the color of the bin. I don't know how to do this!

Existing code for wind roses:

https://www.mathworks.com/matlabcentral/fileexchange/47248-wind-rose

https://www.mathworks.com/matlabcentral/fileexchange/65174-wind_rose-wind_direction-wind_speed

I can't figure out how to hack the code to get what I want...

Answer 1

[Adeline]

I understand that you are trying to plot a wind rose diagram with your data of speed and angle. You want the data to be binned by speed but color coded by frequency. Using the windrose function referenced in the comments, you can bin the data by speed. Since the data is binned by speed in this function, you can color code them only based on speed.

The following example illustrates the above description:

speed = randi(25,1,1000); 
vangle=randi(360,1,1000);
wind_rose(vangle,speed);

Note: If radius is speed and the bin is color coded by frequency (i.e., the number of points in a each speed bin) then the chart would look like a multi-level pie chart as shown below. This cannot be achieved using a polar histogram. As the histogram plot would require one axis to be defined by the number of points in the bins.

[Peter Franks]

Thanks, Adeline! Given how difficult it was to get the solution I was after, I've moved on to an entrirely new approach to the data (which seems to be working well). I appreciate your time and expertise.

[Saurav]

Hi, I am also stuck in the situation same as Peter. Could you tell me which function/ code you used in order to get the plot as this?

And if possible, could you please tell me if you could plot the probability of the data instead of the count?

Thank you.

[Walter Roberson]

I believe they are using https://www.mathworks.com/matlabcentral/fileexchange/65174-wind_rose-wind_direction-wind_speed