How to interpolate 2 dimensions of a 4D array ?

I have a 4D array which represents velocity(ocean currents) in a region, and I have to interpolate it to a different size.

Some insights on data;

1) Variables of 4 D array are (Longitude, latitude, depth, time).

2) Size of the array is [176,239,5,248]. I have to change/interpolate the size of array to [112,152,5,248]. Can someone please help in making the above said change(s)?

This is link to data file.

https://drive.google.com/file/d/1Jcp6F1-Bv85FUMyITiGa0AJncKnLsUMF/view?usp=drive_link

Thanks in advance!

[Dyuman Joshi]

On what basis is the change to be made?

[bhakti]

Hello Dyuman Joshi,

Thanks for your response.

Preferably, any linear regridding options.

[KALYAN ACHARJYA]

Please provide the array sizes [x,y,z,t]?

[bhakti]

Hello Kalyan,

Thanks for your response.

Size of the array is [176,239,5,248] and it has to be changed/interpolated to [112,152,5,248].

Answer 1

[Torsten]

% Generate artificial data in longitude and latitude (176 points resp. 239
% points)
longdata = linspace(0,2,176);
latdata = linspace(0,5,239);
veldata = rand(numel(longdata),numel(latdata),5,248);
% Prescribe new grid in longitude and latitude direction (112 points resp.
% 152 points)
longdata_new = linspace(0,2,112);
latdata_new = linspace(0,5,152);
% Build new grid
[longq,latq] = meshgrid(longdata_new,latdata_new);
% Compute new velocity data
veldata_new = zeros(numel(longdata_new),numel(latdata_new),5,248);
for iz = 1:5
    for it = 1:248
        veldata_cut = veldata(:,:,iz,it);
        veldata_new(:,:,iz,it) = interp2(longdata,latdata,veldata_cut.',longq,latq).';
    end
end

[bhakti]

Thank you!

Answer 2

[Sulaymon Eshkabilov]

Your data size is [176 238 5 248] and not [176,239,40,248].

In this case, interpolate your multi-D data, griddatan() to be used: DOC

[bhakti]

I apologize, I must have pinned the wrong file.

[Torsten]

The dependent variable that you want to interpolate with respect to (Longitude, latitude, depth, time) is missing.

[bhakti]

The variable is velocity(ocean currents) as mentioned.

Velocity is represented on a grid ( Longitude, latitude, depth) at different instants of (time).

[Torsten]

And you want to compute velocity on a new grid for Longitude and Latitude with depth and time unchanged ?

[bhakti]

Hello Torsten,

Thanks for your response.

Yes, that's what I am looking for.

Answer 3

[Matt J]

You can use imresizen, to be downloaded from

https://www.mathworks.com/matlabcentral/fileexchange/64516-imresizen-resize-an-n-dimensional-array?s_tid=srchtitle

A= rand([176,239,5,248]);

B=imresizen(A, [112,152,5,248]./[176,239,5,248]);

whos A B

  Name        Size                       Bytes  Class     Attributes

  A         176x239x5x248            417274880  double              
  B         112x152x5x248            168878080  double              

[bhakti]

Hello Matt,

Thanks for your response.

Am I allowed to as a substandard question? I might be wrong but is the usage of 'imresize' in this case contextually correct even though technically interpolation and imresize might be based off the same principles?

[Matt J]

I don't fully understand the question, but imresizen is doing interpolation, just as you requested.

[bhakti]

Okay, I would try this.

Thank you.