When calling contourf with syntax contourf(Z) (i.e., no X and Y), then the x- and y-coordinates of the contour plot are taken to be the column and row indices, respectively, of the matrix Z. That is to say, x is 1:size(Z,2) and y is 1:size(Z,1). (Your Z2 is of size 23-by-23, so your contour goes from x = 1 to x = 23 and y = 1 to y = 23.)
The matrices LS1 and LS2 contain mostly NaNs, and the non-NaN elements are identical to the elements in the corresponding positions in Z2. I figure you want to plot the points where LS1 and LS2 are non-NaN with their respective markers on top of the contour. To do that, you can use find with two outputs (rows and columns), as that will give you the locations consistent with your contour plot.
load Test
contourf(Z2,20,'edgecolor','none');
colormap(jet())
colorbar
hold on
[r1,c1] = find(~isnan(LS1));
[r2,c2] = find(~isnan(LS2));
plot(c1,r1,'b+','MarkerSize',5);
plot(c2,r2,'rx','MarkerSize',5);
