How to display a matrix with many NaN-values?

Hello,

I have many values of a 2D field and their corresponding coordinaes in a given x-y-range. Yet the coordinates do not form a uniformly occupied grid, it is more like a sparse matrix, the x and y values are often unique and partially only conain one single value per column/line.

Nevertheless, I want to plot my data and interpolate over the NaN entries but the common functions like pcolor and colormap only return a blank figure (except from the denser occupied edges as you can see in the figure).

How do I recieve a interpolated field?

data=file(:,1); %the data points

X=unique(file(:,2));

Y=unique(file(:,3)); %the x and y-coordinates

Field=NaN(length(X),length(Y)); %creates an empty grid

for j=1:length(data)

Xj=find(X==file(j,2));

Yj=find(Y==file(j,3));

Field(Xj,Yj)=data(j); %fills in the known values

end

figure()

pcolor(Field); shading interp;

[Akira Agata]

How about using scatteredInterpolant function?

[Jan]

You want to interpolate over the rectangular grid. This is a good idea. As you can see, pcolor does not do what you want to without this interpolation.

The conversion would be faster with:

X = file(:, 2);
Y = file(:, 3);
Field = NaN(max(Y), max(X));
Field(sub2ind(size(Field), Y, X)) = file(:, 1);

Answer 1

[Timo Conrad]

Ok, here is what I got now:

Thank you @Akira Agata, scatteredInterpolant was helpful, indeed.

X=unique(file(:,2));
Y=unique(file(:,3));
Field=NaN(length(X),length(Y));
Xj=find(X==file(j,2));
Yj=find(Y==file(j,3));
for j=1:length(file(:,1))
    Xj(j)=find(X==file(j,2));
    Yj(j)=find(Y==file(j,3));
    Field(Xj(j),Yj(j))=file(j,1);
end
[Xq,Yq]=meshgrid(X,Y);
figure()
F=scatteredInterpolant(file(:,2),file(:,3),file(:,1)) ;
F.Method='natural';
F2=F(Xq,Yq);
pcolor(F2);shading interp;
end

It creates a fully colored map of the interpolated data:

Thank you for your help.

[Akira Agata]

I'm glad to hear that my previous comment was somehow helpful to plot your data.

But, looking at your code, I feel there must be more efficient way to create this plot.

If possible, could you upload your 2D matrix (= the variable 'file' in your code) so that people here, including me :-) , can try??

[Timo Conrad]

Sure, thank you for the help!

I know some variables defined here are redundant, I just wanted everybody to understand the data.

Nevertheless, I am grateful for any improvements!

The matrix 'file' is attached and consists of column 1: the data, column 2, 3 & 4: the x, y,z-coordinates.

[Akira Agata]

Ah, it's a Karman vortex, isn't it?

Thank you for sharing the data. The following is a simple example:

load('file.mat');

x = file(:,2);
y = file(:,3);
z = file(:,1);

[xg, yg] = meshgrid(...
  linspace(min(x), max(x)),...
  linspace(min(y), max(y)));

F = scatteredInterpolant(x, y, z, 'natural');
zg = F(xg, yg);

figure
surf(xg,yg,zg,...
  'EdgeColor','none')
view(2)
daspect([1 1 1])
box on
colorbar('southoutside')