parfor nested for loops

The following script works finely in for loop, but I need to implement parfor to speed up. I am looking for help how to implement parfor in the following nested for loop. Many thanks for your kind help.

 Z_anom = zeros(n,m);

 d = 0; 

 for y = 1:n
    for c_day = 1:m
        d = d + 1;
        g = 0; % counter for the grid point
        for i = 1:nlat
            for j = 1:nlon
                  g = g + 1;
                Z_anom(d,g) = (Z(c_day,y,i,j) - Z5(c_day,i,j))*sqrt(cosd(datalat(i)));

            end
        end
    end
end

[OCDER]

What are the values of these?

n
m
nlat
nlon

What are the sizes of these?

Z
Z5
datalat

Before using parfor, try to vectorize the Z-Z5*sqrt(costd(...)). It'll be much faster.

[Eb Bed]

The variables are length i.e n-15,m -361, nlat-80 and nlon-70

Z =rand(361,15,80,70)

 Z5 =rand(361,15,80,70)

datalat = rand(80)

[Walter Roberson]

Well, for one thing

sqrt(cosd(datalat(i)))

only needs to be calculated once each time i changes.

[OCDER]

Can you double check this math?

(Z(c_day,y,i,j) - Z5(c_day,i,j))*sqrt(cosd(datalat(i)));
                           ^ are you missing "y,"?

[Eb Bed]

Dear OCDER it is not missed. Y is year c_day is day of the season and z5 is averaged over the year then subtract from z to get anomoloud

[OCDER]

Hm, something isn't adding up.

 Z_anom = zero(n, m); 
 for y = 1:n
    for c_day = 1:m
       d = d+1;
          ...
             Z_anom(d, g) = ...
    end
 end

This means size of Z_anom will [n*m, nlat*nlon], and NOT [m, n] as you started out with.

Also, datalat size is [80, 80], but it seems you are accessing i from 1:nlat, where nlat = 80?

Please provide the size of all variable as per my 1st comment.