accessing indexed values in a 3D array with a logical index

I have 3D array of temperature data that looks like this:

Z = rand(3,4,8);

It contains 8 slices of data on 3*4 grid. I have a logical array which is an index to a set of values on each of the 8 slices like this:

ind = logical(randi(2, [3 4]) - 1)

How to I create a new matrix which is the 3D and contains the temp data from Z for ind on each of the slices.

My Z matrix is actally 814*1294*40 and looking at other answers there should be a way to do this; however I am stumped.

Answer 1

[Walter Roberson]

ind = repmat(ind, 1, 1, size(Z,3));
Z(ind)

However, this will give you a vector of results.

How to I create a new matrix which is the 3D and contains the temp data from Z for ind on each of the slices.

That is a problem, because when you select values for each slice, you get holes, and there is no way to create arrays with holes.

You could reshape Z(ind) to [], 1, size(Z,3) which would give you a vector for each layer.

[Belinda Finlay]

Thanks for your assitance Walter. I am struggling to understand how a vector for each layer will help?

[Walter Roberson]

What would you ideally like it to look like when you had a 3D matrix which "contains the temp data from Z for ind on each of the slices", considering that not all locations in the slice are selected ?

[Belinda Finlay]

I am trying to take the data from large grid of model data and extract a portion in a polygon 250km either side of a tropical cyclone, track and 100km along the track. My aim is to visualise the temp with depth, and be able to see how it changes as the cyclones passes over.

[Walter Roberson]

Is it correct that you are intending to extract a cuboid? If so then do not use logical indexing for that, use vectors

Z(first_x : last_x, first_y : last_y, :)

The result would be a cuboid.

[Belinda Finlay]

Thanks Walter, now I need to work out how to get those values (first_x: last_x etc). I am very grateful for your assistance.

[Walter Roberson]

Let ind be the rectangular 2D mask that you had from before. Then

mask = any(ind,1);
first_col = find(mask,1);
last_col = find(mask,1,'last');
mask = any(ind,2);
first_row = find(mask,1);
last_row = find(mask,1,'last');

selected_area = Z(first_row : last_row, first_col : last_col, :);

[Belinda Finlay]

I am extremely grateful for your assistance Walter - I could hug you.