How to convert y data for equal intervals of x, z to z data for equal intervals of x, y

x and z are tables of size 100x100 using meshgrid.

Values ​​outside the required range of z are treated as NaN.

The y value is the result of the calculation using x and z.

If the result of the calculation for z is a complex number, NaN is processed.

What code should I use to find z for equally spaced x, y?

An error occurs when trying to do 3D interpolation.

I attached the variable as an attachment.

I would greatly appreciate your help.

[Madheswaran]

@Yunjai Could you also post the minimal version of code you were trying to run and the error you were facing?

Answer 1

[Bruno Luong]

Rescaling is important when doing scatering interpolation. Avoid unstable extrapolation

load example.mat
XYZ = [X(:), Y(:), Z(:)];
XYZ = rmmissing(XYZ,1);

[minx, maxx] = bounds(X, "all");
[miny, maxy] = bounds(Y, "all");
XN = (XYZ(:,1)-minx)/(maxx-minx);
YN = (XYZ(:,2)-miny)/(maxy-miny);
F = scatteredInterpolant(YN, XN, XYZ(:,3)  , 'linear', 'none');
[XNG, YNG] = deal(linspace(0, 1));
ZG = F({YNG, XNG});
XG = minx + XNG*(maxx-minx);
YG = miny + YNG*(maxy-miny);
figure
surf(XG, YG, ZG);
shading interp
hold on
contourf(XG, YG, ZG, 'Zlocation',-1.3)
colorbar
plot3(X(:),Y(:),Z(:),'.b')
view(15,50)

[Bruno Luong]

If you want exptrapolation I recommend using either the legacy griddata 'v4' or John d'Erric Inpaint_NaNs

Answer 2

[Walter Roberson]

load example.mat
XYZ = [X(:), Y(:), Z(:)];
XYZ = rmmissing(XYZ,1);
F = scatteredInterpolant(XYZ(:,1), XYZ(:,2), XYZ(:,3));
[minx, maxx] = bounds(X, "all");
[miny, maxy] = bounds(Y, "all");

[XG, YG] = ndgrid(linspace(minx, maxx), linspace(miny, maxy));
ZG = F(XG, YG);
surf(XG, YG, ZG);
shading interp