how to add contourf(X, Y, Z) in the polaraxes

Hi, I've tried to overlay a polaraxes on top of the axes, but it's easy to deform in the subplot or tiledlayout and it looks more incongruous, is there a better way to fix it, or implement some other way to implement contourf in polaraxes. Here is my code. Just like shown, the polaraxes will move as other things get added, dealing with this situation can be pretty boring. And pax.RLim doesn't work as expected.

load("S.mat","S");
figure;
t=tiledlayout(1,2);

nexttile(t);

ax=nexttile(t);
paxctf(ax,S.f,S.th,S.efth);
colorbar(ax);

function [CS,CH,pax]=paxctf(ax,f,th,Efth)
assert(isvector(f) & isvector(th),'Error: f or th is not a vector!');

[ddir,df]=meshgrid(deg2rad(th),f);
[X,Y]=pol2cart(ddir,df);
[CS,CH]=contourf(Y,X,Efth,'LineStyle','none');
axis equal;
axis off;
POS=tightPosition(ax);
pax=polaraxes(gcf,'Position',POS);

pax.Color='none';
pax.ThetaZeroLocation="top";
pax.ThetaDir="clockwise";
pax.RAxisLocation=30;
pax.RTick=[0 1/20 1/10 1/6 1/4];
pax.RTickLabel={'','20 s','10 s','6 s','4 s'};
end

[dpb]

You forgot to attach some data so somebody here can try to duplicate your efforts...and it would undoubtedly be a big help if you could show an example of what it is you're trying to create from a published paper of a hand drawn sketch or ... Otherwise, it's a guess as to what the final objective really is.

[lang]

Ok, I have updated the code.

Answer 1

[Cris LaPierre]

I see what you mean - the colorbar added to the contourf axes has shifted the plot so that it no longer overlaps correctly the polar axes.

I did some quick googling and came across this answer, which proposed adding an event listener to one of the axes. You can read more about events here.

Doesn't hurt to try it out. Here is the result.

load("S.mat","S");
figure;
t=tiledlayout(1,2);

nexttile(t);

ax=nexttile(t);
paxctf(ax,S.f,S.th,S.efth);
colorbar(ax);

function [CS,CH,pax]=paxctf(ax,f,th,Efth)
assert(isvector(f) & isvector(th),'Error: f or th is not a vector!');

[ddir,df]=meshgrid(deg2rad(th),f);
[X,Y]=pol2cart(ddir,df);
[CS,CH]=contourf(Y,X,Efth,'LineStyle','none');
axis equal;
axis off;

pax=polaraxes(gcf);
pax.Color='none';
pax.ThetaZeroLocation="top";
pax.ThetaDir="clockwise";
pax.RAxisLocation=30;
pax.RTick=[0 1/20 1/10 1/6 1/4];
pax.RTickLabel={'','20 s','10 s','6 s','4 s'};

addlistener(ax,'MarkedClean',@(varargin)set(pax,'Position',get(ax,'Position')));
end

[Cris LaPierre]

Addressing your comment about pax.RLim, I'm not sure what behavior you want, It behaves as I'd expect. Here, I've set it to match the min and max RTick values. I set them to be white here so you can see them better.

Unfortunately, lines and labels share the same color, so I didn't change theta's color. I figured it was more important to be able to read the label than see the line. Of course, you could use a differenct color, or use a different colormap for the contourf plot.

load("S.mat","S");
figure;
t=tiledlayout(1,2);

nexttile(t);

ax=nexttile(t);
paxctf(ax,S.f,S.th,S.efth);
colorbar(ax);

function [CS,CH,pax]=paxctf(ax,f,th,Efth)
assert(isvector(f) & isvector(th),'Error: f or th is not a vector!');

[ddir,df]=meshgrid(deg2rad(th),f);
[X,Y]=pol2cart(ddir,df);
[CS,CH]=contourf(Y,X,Efth,'LineStyle','none');
axis equal;
axis off;

pax=polaraxes(gcf);
pax.Color='none';
pax.ThetaZeroLocation="top";
pax.ThetaDir="clockwise";
pax.RAxisLocation=30;
pax.RTick=[0 1/20 1/10 1/6 1/4];
pax.RTickLabel={'','20 s','10 s','6 s','4 s'};
pax.RLim = [0 1/4];
pax.RColor = 'w';

addlistener(ax,'MarkedClean',@(varargin)set(pax,'Position',get(ax,'Position')));
end

[lang]

Thanks, that helps a lot! As for the Rtick, I mean, the frequency range in the contourf axes is not influenced by the pax.rlim, it may need to be adjusted manually.

[Cris LaPierre]

The polar and contour plots are on different axes, so yes, you'd have to adjust it manually.

[dpb]

Could not a callback on one fixup the other, @Cris? Each would need to futz with the other and it might get messy....

There isn't a generalized linkaxes that could specify comparative parameters between two dissimilar axes...and not likely to make the list of approved enhancement requests I'd wager? :)