How to show surface plot of 2D data?

Hi.

I have a set of data points, C at each x and y (for many cases). In 2D, the plot looks simply like this:

So this is a sample plot for one x. There are similar plots for other x values as well.

What I want is some kind of surface plot. How can I get it? I tried to use surf and contour, but they need their z to be a matrix, not an array.

For instance, the data looks like this:

% set1

x = [1 1 1 1];

y = [1 3 5 16];

C = [100 400 200 500];

% set2

x = [2 2 2 2];

y = [4 7 8 13];

C = [200 500 700 100];

Thank you

Answer 1

[Star Strider]

Try this (with your own vectors):

x = rand(10,1);                                     % Create Data
y = rand(10,1);                                     % Create Data
z = rand(10,1);                                     % Create Data

[X,Y] = ndgrid(sort(x), sort(y));                   % Create Interpolation Grids
Z = griddata(x, y, z, X, Y);                        % Interpolate ‘z’

figure
surf(X, Y, Z)
hold on
stem3(x, y, z, 'filled')
hold off
grid on

Experiment to get different results.

[Steven]

Thanks.

The problem is x values are fixed each time, so for instance for your example:

x = ones(10,1);                                     % Create Data
y = rand(10,1);                                     % Create Data
z = rand(10,1);                                     % Create Data

For this case, there would be an error of points being collinear. Right?

[Star Strider]

My pleasure.

Please go into a bit of detail with respect to ‘x values are fixed each time’, since I do not understand what this means. There was no mention of that in the original Question.

The rand values would likely not be collinear, especially for such short vectors as in my example. My code threw no errors when I ran it, so griddata did not detect any non-unique values.

[Steven]

I meant my x values are something like this:

x = [1 1 1 1 1 1];

Yes, your code throws no error with random numbers, but throws error for the above x values. Any idea how to resolve that?

[Star Strider]

Your data might be gridded. (I would have to see it to be certain.) If that is true, you would only need to reshape your vectors to form matrices from them, then plot the matrices. You could later use griddata to interpolate the results to a denser grid, if you wanted to.

[Steven]

For instance, these are the data:

% set1

x = [1 1 1 1];

y = [1 3 5 16];

C = [100 400 200 500];

% set2

x = [2 2 2 2];

y = [4 7 8 13];

C = [200 500 700 100];

How to either reshape or grid them then?

Thanks

[Star Strider]

Yes.

Use the same reshape call with each vector, changing only the vector argument.

[Star Strider]

I was not expecting that possibility.

Try this:

% set1
x1 = [1  1 1 1];
y1 = [1 3 5 16];
C1 = [100 400 200 500];
% set2
x2 = [2 2 2 2];
y2 = [4 7 8 13];
C2 = [200 500 700 100];

xm = [x1; x2];
ym = [y1; y2];
Cm = [C2; C2];

figure
surf(xm, ym, Cm)
grid on

Extend it for more vectors. Note that they all must have the same column sizes for this to work (that is, they all need to be row vectors with the same number of columns).

[Steven]

Unfortuantely some of them are not of the same size, but some of them are! But it works for those that are though :)

P.S., I edited the original post to include the data.

Thanks again

[Star Strider]

As always, my pleasure!

You can make them all the same size with the interp1 or interp2 functions. It is slightly more work, however you can then use all your data.

For example, to extend ‘x1’, ‘y1’, and ‘C1’ to each have a length of 7:

x1i = x1(1)*ones(1,7);
y1i = interp1((1:numel(x1)), y1, linspace(1, numel(x1), numel(x1i)));
C1i = interp1((1:numel(x1)), C1, linspace(1, numel(x1), numel(x1i)));

You could probably create a function to do this.