Surface plot with 3 vectors Matlab

Question

I need to be able to do a surface plot using data from 3 vectors. I found similar information, but no method seems to work with my data. My X and Y columns are evenly spaced, but not in increasing order. I tried different methods, but none of them seem to give me what I want, which is a simple surface linking close points together. I tried the following:

[X Y]=meshgrid(x,y);
Z=griddata(x,y,z, X,Y);
surf(X,Y,Z);

This is not exactly what I want, because it creates a surface at z=0 and makes it look more like a volume plot than just a surface. It also runs very slowly on my computer (probably from creating all the gridpoints). If I could get something that doesn't require as much memory it would be ideal (my vectors have about 20k values each), but this is not a necessity.

***Edit: I also tried using the scatteredInterpolant method found here ,but the function doesn't seem to be recognized by MATLAB and I get this error:

Undefined function 'scatteredInterpolant' for input arguments of type 'double'.

Also here is an image of my problem:

You can see that we can't see under the surface, there is some z=0 plane blocking it.

If you have anything for me, any help is appreciated.

Thanks in advance.

**Edit 2: I added sample vectors , they're my x,y and z values from left to right.

***Edit 3: Here's an image of the triangulation I get. As you can see some points are being ignored for some reason, which gives those long and weird looking blue triangles.

Mike

Answer 1

If you have points which are described by vectors, and you want to plot them you could always use a Delauny triangulation. The function in matlab is called Tri=delauny(X,Y,Z) . The data generated by this function can be shown with either trimesh(Tri,X,Y,Z) or trisurf(Tri,X,Y,Z) . Keep in mind trisurf is only for 3D data. If you want to adjust the transparancy of plots in your graph use the alpha setting.

I hope this helps

Answer 2

As conventional methods seem to fail, I would suggest you to do it manually.

1. Create a Z matrix full of NaN values. The size of the matrix should be dependant on your x and y values.
2. Loop over all occuring x , y , pairs and put their (average?) z value in the right position of your Z matrix.
3. Loop over all NaN values and interpolate their value. Perhaps using filter2 .
4. Use surf to plot the resulting surface

Answer 3

To me it looks like you just need to sort your data before plotting.

Here is an example which I believe is similar to your case (since I could not download your data).

x = [2 1 4 3 -1 -3 -4 -2];
y = [1 2 3 4 -1 -2 -3 -4];
z = 32 - x.*x - y.*y;

[X1 Y1] = meshgrid(x,y);
Z1 = 32 - X1.*X1 -Y1.*Y1; 
surf(X1,Y1,Z1)

aux = sort([x;y],2);
x = aux(1,:);
y = aux(2,:);

[X2 Y2] = meshgrid(x,y);

Z2 = 32 - X.*X - Y.*Y;
figure()
surf(X2,Y2,Z2)

The first figure results in a very problematic surface:

The second figure contains the desired surface: