2D Interpolation & Extrapolation with non linear Sample grid points and nan

Hi,

I have this 2D map wich I would like to interpolate to make square but also extrapolate.

interp2 does seem to work with nan inputs as it just returns same nan as output.

V=[1.029	1.000	1.060	1.048	nan	nan	nan
1.049	1.014	1.038	1.092	1.075	nan	nan
nan	1.039	1.079	1.036	1.029	1.003	nan
nan	nan	1.042	1.072	1.000	0.998	1.028
nan	nan	nan	1.038	1.024	1.076	1.022
nan	nan	nan	nan	1.082	0.997	1.011];


X=[8 10	15 25 35 45	52];

Y=[5 10 16.25 22.5 28.75 35];

Xq=ones(numel(Y),numel(X)).*X;

Yq=ones(numel(Y),numel(X)).*Y';

Z=interp2(X,Y,V,Xq,Yq)

Z = 6×7

1.0290 1.0000 1.0600 1.0480 NaN NaN NaN 1.0490 1.0140 1.0380 1.0920 1.0750 NaN NaN NaN 1.0390 1.0790 1.0360 1.0290 1.0030 NaN NaN NaN 1.0420 1.0720 1.0000 0.9980 1.0280 NaN NaN NaN 1.0380 1.0240 1.0760 1.0220 NaN NaN NaN NaN 1.0820 0.9970 1.0110

I tried some inpaint functions from the community but none of them seems to be made to work with non linear sample grid points (0 2 7 17...).

I am open to suggestions

Thanks in Advance

Laurent

[Jeffrey Clark]

@Laurent Davenne, add ,'makima' to your interp2 which should cause extrapolation of available 2 or more points.

[Torsten]

interp2 does seem to work with nan inputs as it just returns same nan as output.

Everything else would have been a surprise. What do you expect ?

[Laurent Davenne]

@Jeffrey Clark, I tried but this does not work either.

Insufficient finite values to interpolate.

[Laurent Davenne]

@Torsten, Thanks, really helpful!

Answer 1

[Voss]

V = [ ...
    1.029   1.000   1.060   1.048   nan	    nan	    nan
    1.049   1.014   1.038   1.092   1.075   nan	    nan
    nan	    1.039   1.079   1.036   1.029   1.003   nan
    nan	    nan	    1.042   1.072   1.000   0.998   1.028
    nan	    nan	    nan	    1.038   1.024   1.076   1.022
    nan	    nan	    nan	    nan	    1.082   0.997   1.011]

V = 6×7

1.0290 1.0000 1.0600 1.0480 NaN NaN NaN 1.0490 1.0140 1.0380 1.0920 1.0750 NaN NaN NaN 1.0390 1.0790 1.0360 1.0290 1.0030 NaN NaN NaN 1.0420 1.0720 1.0000 0.9980 1.0280 NaN NaN NaN 1.0380 1.0240 1.0760 1.0220 NaN NaN NaN NaN 1.0820 0.9970 1.0110

X = [8 10 15    25   35    45   52]; % using the X and Y from your code rather than
Y = [5 10 16.25 22.5 28.75 35];      % from your image, which are 5 less

% make a scatteredInterpolant of the points where V is non-NaN
[XX,YY] = meshgrid(X,Y);
idx = ~isnan(V(:));
I = scatteredInterpolant(XX(idx),YY(idx),V(idx));

% interpolate/extrapolate to all points
Z = reshape(I(XX,YY),numel(Y),[])

Z = 6×7

1.0290 1.0000 1.0600 1.0480 1.0831 1.0621 0.9944 1.0490 1.0140 1.0380 1.0920 1.0750 1.0286 1.0002 1.0322 1.0390 1.0790 1.0360 1.0290 1.0030 1.0234 1.0147 1.0212 1.0420 1.0720 1.0000 0.9980 1.0280 0.9854 0.9911 1.0106 1.0380 1.0240 1.0760 1.0220 0.9513 0.9626 0.9872 1.0543 1.0820 0.9970 1.0110

% visualization
x_lim = [min(X) max(X)];
y_lim = [min(Y) max(Y)];
z_lim = [min(Z(:)) max(Z(:))];
temp = {V Z};
names = {'V' 'Z'};
for ii = [1 2]
    subplot(2,1,ii)
    surface(XX,YY,temp{ii},'FaceColor','interp')
    set(gca(),'YDir','reverse')
    view(2)
    xlim(x_lim);
    ylim(y_lim);
    xlabel('X');
    ylabel('Y');
    colorbar();
    caxis(z_lim);
    title(names{ii});
end

[Laurent Davenne]

@Voss, Fantastic!

Thank you so much!

[Voss]

You're welcome!