替换蒙版的numpy数组中的值不起作用

Question

I have the foll. masked array in numpy called arr with shape (50, 360, 720):

masked_array(data =
 [[-- -- -- ..., -- -- --]
 [-- -- -- ..., -- -- --]
 [-- -- -- ..., -- -- --]
 ..., 
 [-- -- -- ..., -- -- --]
 [-- -- -- ..., -- -- --]
 [-- -- -- ..., -- -- --]],
             mask =
 [[ True  True  True ...,  True  True  True]
 [ True  True  True ...,  True  True  True]
 [ True  True  True ...,  True  True  True]
 ..., 
 [ True  True  True ...,  True  True  True]
 [ True  True  True ...,  True  True  True]
 [ True  True  True ...,  True  True  True]],
       fill_value = 1e+20)

It has the foll. data in arr[0]:

arr[0].data

array([[-999., -999., -999., ..., -999., -999., -999.],
       [-999., -999., -999., ..., -999., -999., -999.],
       [-999., -999., -999., ..., -999., -999., -999.],
       ..., 
       [-999., -999., -999., ..., -999., -999., -999.],
       [-999., -999., -999., ..., -999., -999., -999.],
       [-999., -999., -999., ..., -999., -999., -999.]])

-999. is the missing_value and I want to replace it by 0.0. I do this:

arr[arr == -999.] = 0.0

However, arr remains the same even after this operation. How to fix this?

Answer 1

Maybe you want filled . I'll illustrate:

In [702]: x=np.arange(10)    
In [703]: xm=np.ma.masked_greater(x,5)

In [704]: xm
Out[704]: 
masked_array(data = [0 1 2 3 4 5 -- -- -- --],
             mask = [False False False False False False  True  True  True  True],
       fill_value = 999999)

In [705]: xm.filled(10)
Out[705]: array([ 0,  1,  2,  3,  4,  5, 10, 10, 10, 10])

In this case filled replaces all masked values with a fill value. Without an argument it would use the fill_value .

np.ma uses this approach to perform many of its calculations. For example its sum is the same as if I filled all masked values with 0. prod would replace them with 1.

In [707]: xm.sum()
Out[707]: 15
In [709]: xm.filled(0).sum()
Out[709]: 15

The result of filled is a regular array, since all masked values have been replaced with something 'normal'.