How do I plot wind direction timeseries with arrows

103 visualizaciones (últimos 30 días)
giacomo labbri
giacomo labbri el 12 de Jun. de 2020
Editada: David el 21 de En. de 2022
Hi,
I would like to plot a timeseries of wind direction. I would like represent the direction at each point in time with an arrow. The direction of the wind is expressed in degrees. I have saw that quivers might do what I want with some tricks but I was wondering if there was a more specific function.
Thanks in advance,
Giacomo
  3 comentarios
Adam Danz
Adam Danz el 12 de Jun. de 2020
Editada: Adam Danz el 14 de Jun. de 2020
As my answer indicates, the quiver function is what you should use given the description of the problem. However, the quiver function does not accept datetime values. It only accepts numeric values or values that can be converted to double. That's for a good reason. Imagine a vector whose base is at x="January 1, 2000" and y = 10 and has (u,v) components of (2,1) (in other words, it extends 2 units in the x direction and 1 unit in the y direction). What does that mean with datetime values?
What is "2" in datetime? 2 days? 2 hours? Matlab wouldn't be able to figure that out. But you could figure that out since you know the context of your data.
So, you'll need to convert the datetime values in order to use quiver. Then you can set the x tick labels aftewards to datetime values using datetick.
giacomo labbri
giacomo labbri el 13 de Jun. de 2020
thanks that was really helpful! :)

Iniciar sesión para comentar.

Respuesta aceptada

Adam Danz
Adam Danz el 12 de Jun. de 2020
Editada: Adam Danz el 21 de En. de 2022
The quiver() function is what you want to use. It's desiged exactly for this purpose.
Since your vectors are in degrees and the quiver function requires the x,y component vectors, use pol2cart() to convert your polar vectors to cartesian vectors. You'll also need to convert deg-->rad for the pol2cart inputs.
% WindDirection is the direction of wind (deg)
% WindSpeed is the magnitude of the wind vector
[xWind, yWind] = pol2cart((pi/180).*WindDirection, WindSpeed);
quiver(x, y, xWind, yWind)
  3 comentarios
David
David el 20 de En. de 2022
Editada: David el 21 de En. de 2022
One issue with this is that the the angles won't necessarily line up. If you have a lot of x points, it will compress the xWind component and throw off the angles.
There was an extra parenthesis. Try this:
% WindDirection is the direction of wind (deg)
% WindSpeed is the magnitude of the wind vector
[xWind, yWind] = pol2cart((pi/180).*WindDirection, WindSpeed);
quiver(x, y, xWind, yWind)
Adam Danz
Adam Danz el 21 de En. de 2022
Thanks for pointing out the typo. Fixed.

Iniciar sesión para comentar.

Más respuestas (1)

David
David el 20 de En. de 2022
Editada: David el 21 de En. de 2022
A problem with quiver for this is that the horizontal axis is shared by both time and the x component of windspeed, so you have to make choices about how you scale things if you want northeast winds to go on the 45° angle and the y axis to measure speed
If you leave the aspect ratio free, the date axis is correct, but the 45° winds won't be proper.
x = datenum(t); y=x*0;
quiver(x,x*0, xWind, yWind,0,'.');
datetick;
If you make the aspect ratio equal (`axes equal`) then the 45° will be proper, but you may need to scale the time data to see your data, and then that screws up the automatic date tick labelling. Then getting the time tickmarks right is tricky. You can capture the automatically generated tickmarks, find the corresponding times, and then replace the labels with the corresponding times. See https://www.mathworks.com/matlabcentral/answers/391904-datetick-at-feather-plot#answer_312939
If you don't need the dateticks, this works well for maintaing angles through resizing and zooming:
sfact = 1000; % or whatever looks nice
x = datenum(t); y=x*0;
quiver(x*sfact,x*0, xWind, yWind,0,'.');
axis equal;
set(gca,'XTickLabel',[],'YTickLabel',[])
% ... tickmark replacement magic
Or if you keep the time axis, and the yWind, you could scale the xWind by a factor to match the aspect ratio between time and yWind, but then you can't resize/reshape your plot without decalibrating the angles.
I like this one best if you want to label the time axis:
x = datenum(t);
y= x*0;
quiver(x, y, xWind, yWind,0,'.')
x = datenum(t); y=x*0;
sfact = 1;
rg=1000:3200;
quiver(x(rg),x(rg)*0, xWind(rg)*sfact, yWind(rg),0,'.');
ylim([-10,10]);
xlim([min(x(rg)),max(x(rg))])
% asjust plot to desired size/shape/aspect
datetick;
% capture the aspect ratio between time and yWind
dd=daspect;ax=gca; sfact = 1/(dd(2)/dd(1)/ax.PlotBoxAspectRatio(2));
[dd(2)/dd(1) ax.PlotBoxAspectRatio dd(2)/dd(1)/ax.PlotBoxAspectRatio(2) sfact]
quiver(x(rg),x(rg)*0, xWind(rg)*sfact, yWind(rg),0,'.'); %replot scaling xWind to match time
ylim([-10,10]);xlim([min(x(rg)),max(x(rg))])
datetick; % add dateticks

Categorías

Más información sobre Vector Fields en Help Center y File Exchange.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by