I want to apply a mask of 2 dimensions (an NxM array) to a 3 dimensional array (a KxNxM array). How can I do this?

2d = lat x lon

3d = time x lat x lon

import numpy as npa = np.array([[[ 0,  1,  2],[ 3,  4,  5],[ 6,  7,  8]],[[ 9, 10, 11],[12, 13, 14],[15, 16, 17]],[[18, 19, 20],[21, 22, 23],[24, 25, 26]]])b = np.array([[ 0, 1, 0],[ 1, 0, 1],[ 0, 1, 1]])c = np.ma.array(a, mask=b)  # this behavior is wanted 

There are quite a few different ways to choose from. What you want to do is align the mask (of lower dimension) to the array that has the extra dimension: the important part is that you get the number of elements in both arrays the same, as the first example shows:

np.ma.array(a, mask=np.concatenate((b,b,b)))  # shapes are (3, 3, 3) and (9, 3)
np.ma.array(a, mask=np.tile(b, (a.shape[0],1)))  # same as above, just more general as it doesn't require you to specify just how many times you need to stack b.
np.ma.array(a, mask=a*b[np.newaxis,:,:])  # used broadcasting