Interpolating one dimension in 4-th dimensional matrix?

Hi Matlab users.

I have a matrix which is like that (132,238,35,6) this is for longitude, latitude, depth and time. And I want to interpolate the depth so I will have 70 instead of 35. The problem is that I don't know how to properly do that because if I try to call the depth like that: (1,1,:,1) I will have only the one corresponding first longitude and first latitude, not the entire depth. Also If I call like that: (:,:,35,:) the interpolation function don't work. Is there any way to interpolate a 4-th dimensional matrix so I will obtain 70 depths instead of 35?

Thank you,

Robert.

Answer 1

[Walter Roberson]

interpn(Longs,Lats,Depths,Times,TheMatrix,Longs,Lats,NewDepths,Times)

Note: if you do not want to extrapolate the most obvious new depths would include the ones half-way between the old depths, which would lead to 69 output depths, not 70.

[Argho Datta]

Hi, sorry to dig this up, but what does the argmuent NewDepths refer to? Thanks!

Answer 2

[Jan]

Do you want a linear interpolation? Then you can do it manually also:

data = rand(132,238,35,6);

% 70 values between 1 and 35:
v = linspace(1, 35, 70)

% Indices of left and right slice of the array:
t = floor(v);
p = v - t;

% Consider edge at the end:
t(end) = t(end) - 1;
p(end) = 1;
p = reshape(p, 1, 1, []);

R = bsxfun(@times, data(:, :, t, :), (1 - p)) + ...
    bsxfun(@times, data(:, :, t+1, :), p);

Answer 3

[Robert]

Thank you both guys, both variants are working just fine.

Thanks again.