How to shift certain arrays in a matrix downward

I have a large matrix where data for each month of 100 years is stored for 12 variables. Some of the variables have data starting from later date. How can a shift those variables down. For example I have a matrix;

1 4 6 9 0 2

3 6 8 8 2 5

2 5 6 8 9 0

3 5 7 9 2 1

And I'd like the output to be (with NaNs replacing the empty places);

1 4 N N N 2

3 6 6 N N 5

2 5 8 9 0 0

3 5 6 8 2 1

Provided I have stored in a different matrix what the lag would be, in this case say lag=[0, 0, 1, 2, 2, 0].

Answer 1

[Jos (10584)]

This is a fast and easy-to-read solution:

    M = [1 4 6 9 0 2
         3 6 8 8 2 5
         2 5 6 8 9 0
         3 5 7 9 2 1]   % matrix
    S = [0 0 1 2 2 0]   % shift for each column
    M2 = nan(size(M)) ; % pre-allocation for speed
    for k=1:size(M,2)   % loop over columns
       M2(1+S(k):end, k) = M(1:end-S(k), k) ; % copy column k with a shift
    end
    disp(M2)

[SMA]

Great, I was doing something similar but messed up somewhere. It works with minor modification for my data.