Error in interp2 (interpolation) command.

I woyld like to intepolate my data. I have an ascii file with 3 columns (x,y,z). Each column has 700 lines (number). I am using the following commands, in order to interpolate (smooth) my data:

clc
clear

filename1= 'mydata.csv'; 
[d1,tex]= xlsread(filename1);

y=d1(:,3);

x=d1(:,4);


z=d1(:,5);

[xq, yq] = meshgrid(...
    linspace(min(x),max(x)),...
    linspace(min(y),max(y)));

zq = interp2(x,y,z,xq,yq,'cubic');
[c,h]= contourf(xq,yq,zq);

but command window shows me:

Error using griddedInterpolant

The grid vectors must be strictly monotonically increasing.

Error in interp2>makegriddedinterp (line 228)

F = griddedInterpolant(varargin{:});

Error in interp2 (line 128)

F = makegriddedinterp({X, Y}, V, method,extrap);

could you please help me?

[Torsten]

"If X and Y are matrices representing a full grid (in meshgrid format), then V must be the same size as X and Y."

x and y are vectors.

[Ivan Mich]

I am uploading a sample of data file

[Torsten]

No, the two inputs being provided to the INTERP2 are not vectors, but are in fact the (matrix) outputs from MESHGRID.

The inputs x,y and z to interp2 the OP provides are vectors (which is wrong). The query values xq and yq are matrices obtained from "meshgrid".

Answer 1

[Torsten]

Read about the requirements for input arguments x, y and z for interp2:

https://de.mathworks.com/help/matlab/ref/interp2.html

Paragraph "Input Arguments".

[Ivan Mich]

I have already read it, but I do not understand which command should I use about xq and yq in my script? Could you please help/explain me?

[Torsten]

So your x and y are strictly monotonic vectors and your z is a matrix with length(y) rows and length(x) columns ? If I look at your code, this does not seem to be the case.

[Torsten]

Your data are not gridded. Thus you will have to use "scatteredInterpolant":

F = scatteredInterpolant(x,y,z);
[xq,yq] = meshgrid(linspace(min(x),max(x)),linspace(min(y),max(y)));
zq = F(xq,yq)

[Steven Lord]

Confirm that your x and y vectors are sorted.

issorted(x, 'strictascend')
issorted(y, 'strictascend')

One or both of those will return false. Once you know which one(s) are not strictly ascending, you need to determine why. Plotting that vector may help you identify where the flat, decreasing, or missing segments are.

Answer 2

[Torsten]

d1 = readmatrix("https://de.mathworks.com/matlabcentral/answers/uploaded_files/1165673/mydata.csv");
y=d1(:,3);
x=d1(:,4);
z=d1(:,5);
F = scatteredInterpolant(x,y,z,'nearest');

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

[xq,yq] = meshgrid(linspace(min(x),max(x)),linspace(min(y),max(y)));
zq = F(xq,yq);
contourf(xq,yq,zq)
colorbar

[Ivan Mich]

ok thank you, but the point is that i would like to "smooth" these contours that you show me in the result image as it produced. So, I think that interpolation would be suitable, but the resultedf contours are "iritable" (not smoothed).

Is there a way to make it?

[Stephen23]

@Ivan Mich: you could try a thin-plate spline: https://www.mathworks.com/help/curvefit/tpaps.html

[Torsten]

@Ivan Mich

Your z-data are not continuous - they only take values 1 1.5 2 2.5 3 ... 9. So would continuous values as 3.1865 even make sense for your application ?

And if you look at the contour plot from your raw data above: the only thing you can learn is that the values for the south-east part are lower than those for the north-west part.

Everything else like further smoothing would not be appropriate in my opinion.