Create contour plot from scatter plot

I want to create a contour from this matrix with x,y and z data. x is the length, y is the width and z is the deviatoric strain.

I want to create a contour plot instead. Is that possible?

figure; hold on ; box on 
scatter(x,y,[20],-z,'filled')
ylabel('Width, \ $(\mathrm{m})$','FontSize',15); xlabel('Length, \ $(\mathrm{m})$','FontSize',15,'FontName','times')

Answer 1

[Voss]

M = readmatrix('k.txt');
x = M(:,1);
y = M(:,2);
z = M(:,3);

I = scatteredInterpolant(x,y,z);

Warning: Duplicate data points have been detected and removed - corresponding values have been averaged.

N = 80;
X = linspace(min(x),max(x),N);
Y = linspace(min(y),max(y),N);
[X,Y] = meshgrid(X,Y);
Z = I(X,Y);
contour(X,Y,Z)
colorbar

[mads skibsted]

Hi, Thanks,

But it is really import to keep the hole in the middle. There is no contours at that area..

[Voss]

You're welcome!

[Voss]

M = readmatrix('k.txt');
x = M(:,1);
y = M(:,2);
z = M(:,3);

I = scatteredInterpolant(x,y,z);

Warning: Duplicate data points have been detected and removed - corresponding values have been averaged.

N = 80;
X = linspace(min(x),max(x),N);
Y = linspace(min(y),max(y),N);
[X,Y] = meshgrid(X,Y);
idx = X.^2+Y.^2 <= 4;
X(idx) = NaN;
Y(idx) = NaN;
Z = I(X,Y);
contour(X,Y,Z)
colorbar

[Mathieu NOE]

some minor improvements (??) like the log spaced v (levels) + automatic minimum radius computation for the inner hole

M = readmatrix('k.txt');
x = M(:,1);
y = M(:,2);
z = M(:,3);


%% uncomment the code below if you want to overlay the scatter and the ontour plot
% figure;
% hold on ;
% box on
% scatter(x,y,[20],z,'filled')
% colormap('jet');
% colorbar('vert');
% ylabel('Width, \ $(\mathrm{m})$','FontSize',15);
% xlabel('Length, \ $(\mathrm{m})$','FontSize',15,'FontName','times')
% axis square

%%%%%%
% v = linspace(min(z),max(z),50); % define the levels for the contour plot
v = logspace(log10(min(z)),log10(max(z)),50);

I = scatteredInterpolant(x,y,z,'linear','none');

Warning: Duplicate data points have been detected and removed - corresponding values have been averaged.

N = 100;
X = linspace(min(x),max(x),N);
Y = linspace(min(y),max(y),N);
[X,Y] = meshgrid(X,Y);
[th,r] = cart2pol(x,y);
idx = X.^2+Y.^2 <= (ceil(min(r)))^2;
X(idx) = NaN;
Y(idx) = NaN;
Z = I(X,Y);
[c,h] = contour(X,Y,Z,v);
colormap('jet');
grid on
axis square
% clabel(c,h,v,'LabelSpacing',3000,'Color','b','FontWeight','bold')
colorbar('vert')

[Voss]

Those are definitely improvements. Thank you!

[Mathieu NOE]

hello @Voss

tx for the nice comment :)

Answer 2

[Mathieu NOE]

hello

sure you can do that

notice that I needed to create lot of levels to display the outer contour line because your z data is highly concentrated in the lower portion of the range (we can see that by plotting a histogram)

data = readmatrix('k.txt');

x = data(:,1);
y = data(:,2);
z = data(:,3);

histogram(z,200)

% Interpolate the scattered data 
xd = linspace(min(x),max(x),100);
yd = linspace(min(y),max(y),100);
v = linspace(min(z),max(z),200); % define the levels for the contour plot

[xq,yq] = meshgrid(xd,yd);
zq = griddata(x,y,z,xq,yq);

Warning: Duplicate data points have been detected and removed - corresponding values have been averaged.

% replace NaN's  generated by griddate with 0 (or a value lower than min(z)
zq(isnan(zq)) = 0;
% contour plot
[c,h] = contour(xq,yq,zq,v);
grid on
axis square

[Mathieu NOE]

this could be an alternative , with less contour lines

now the levels are generated with a log scale

and I use clabel to show what are the corresponding z value

data = readmatrix('k.txt');

x = data(:,1);
y = data(:,2);
z = data(:,3);

% Interpolate the scattered data 
xd = linspace(min(x),max(x),100);
yd = linspace(min(y),max(y),100);
% v = linspace(min(z),max(z),200); % define the levels for the contour plot
v = logspace(log10(min(z)),log10(max(z)),10);

[xq,yq] = meshgrid(xd,yd);
zq = griddata(x,y,z,xq,yq);

Warning: Duplicate data points have been detected and removed - corresponding values have been averaged.

% replace NaN's  generated by griddate with 0 (or a value lower than min(z)
zq(isnan(zq)) = 0;
% contour plot
[c,h] = contour(xq,yq,zq,v);
grid on
axis square
clabel(c,h,v,'LabelSpacing',3000,'Color','b','FontWeight','bold')
colorbar('vert')

[mads skibsted]

Hi, Thanks,

But it is reallt import to keep the hole in the middle. There is no contours at that area..

Answer 3

[Walter Roberson]

See https://www.mathworks.com/matlabcentral/fileexchange/38858-contour-plot-for-scattered-data

[mads skibsted]

Hi walter,

I have looked into this funtion, but could not make it work..

Would you maybe try with the data I have?

[Mathieu NOE]

It works on my side but with tricontour I don't know how you can manage to keep the inner hole

data = readmatrix('k.txt');

x = data(:,1);
y = data(:,2);
z = data(:,3);

% v = linspace(min(z),max(z),20); % define the levels for the contour plot
v = logspace(log10(min(z)),log10(max(z)),50);

tri=delaunay(x,y);           % triangulate scattered data
[C,h]=tricontour(tri,x,y,z,v);
% clabel(C,h)
colormap('jet');
grid on
axis square
% clabel(c,h,v,'LabelSpacing',3000,'Color','b','FontWeight','bold')
colorbar('vert')

[Mathieu NOE]

so you probably should stick with the solution based on scatteredInterpolant