MESHGRID and Contour

Hello,

I have 3 vectors x,y,z; with x and y representing co-ordinates and z their values, which are temperatures assigned to the 13 points. Basically what i want to do is create a sort of 'heatmap' that will display temperatures on a 2 dimensional plot that corresponds to the correct x and y coordinates. Here is what I did and where I got stuck:

[x1,y1] = meshgrid(x,y)

Now the problem is how do i get the z into a correct 13*13 matrix to use contour(x1,y1,z)?

Any help would be appreciated.

J

Answer 1

[Walter Roberson]

Do you have a z for each x / y combination? e.g., z(2,3) = f(x(2), y(3)) ? Or do you have only a vector in which x(1) relates only to y(1) and z(1), x(2) only to y(2) and z(2), etc. (scattered values)?

If you have only scattered values, and you have a sufficiently new MATLAB version, see http://www.mathworks.com/help/techdoc/ref/triscatteredinterp.html

[J]

My data creates a 4*4 grid when it is on a 2-d x,y plot

Here is my data (i am using log scales):

X Y Intensity

0.000000187 0.000000111 84.50318071

0.000000187 1.11E-08 17.93305682

0.000000187 1.11E-09 12.03762387

0.000000187 0.00000111 10.50386162

1.87E-08 0.000000111 83.57328642

1.87E-08 0.00000111 9.9743495

1.87E-09 0.000000111 75.01349657

1.87E-09 1.11E-08 103.5619297

1.87E-09 0.00000111 6.666592105

1.87E-10 0.000000111 42.93567326

1.87E-10 1.11E-08 64.46820503

1.87E-10 1.11E-09 66.86520097

1.87E-10 0.00000111 1.607438028

[J]

im sorry for the bad spacing, i copied it from excel

Basically, i have 4 different x's and four different y's and values for each combo

[J]

also i have versions R2007a and R2011a available

[Walter Roberson]

You only appear to have 13 values to put into 16 matrix locations ? What do you want to have happen for the places where the data is missing? Should NaN be put there? Did you want interpolation to fill in the NaN as a step after that and before drawing the image?

[J]

This helped me figure it out, i was repeating X points

[Walter Roberson]

If you had all of the data pairs, in a regular order, then the matrix could be filled in by taking the z vector, reshaping it, and transposing the result, with the corresponding x and y value orders extracted from the vectors. You might also need to do a flip of the matrix as your x appears to be in decreasing order.

Another approach that would work even with incomplete values would be:

[xu, xjunk, xidx] = unique(x);

[yu, yjunk, yidx] = unique(y);

ValMatrix = accumarray([xidx(:), yidx(:)], [], [], NaN);

IValMatrix = inpaint_nans(ValMatrix);

contour(xu, yu, IValMatrix);

In the above, inpaint_nans is the MATLAB File Exchange contribution of that name, written by John D'Errico.