Create an ndarray from another ndarray based on function result

Question

I have three ndarrays as follows, where x and y are my coordinate grids and z is my data:

x: [[-11.   -11.   -11.   -11.   -11.  ]
 [ -9.25  -9.25  -9.25  -9.25  -9.25]
 [ -7.5   -7.5   -7.5   -7.5   -7.5 ]
 [ -5.75  -5.75  -5.75  -5.75  -5.75]
 [ -4.    -4.    -4.    -4.    -4.  ]]

y: [[51.   52.25 53.5  54.75 56.  ]
 [51.   52.25 53.5  54.75 56.  ]
 [51.   52.25 53.5  54.75 56.  ]
 [51.   52.25 53.5  54.75 56.  ]
 [51.   52.25 53.5  54.75 56.  ]]

z: [[ 0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.]
 [ 0. 21.  0.  0.  0.]
 [21. 21.  4.  0.  0.]
 [21. 21.  4.  4.  0.]]

I wish to do something like this:

coords = np.stack((x, y)).T
for (x,y), value in np.ndenumerate(z):    
    xx,yy = coords[x][y]
    if not m.is_land(xx,yy):
        z[x][y] = 0

How do I do this correctly in numpy? Better still would be to create a new array instead of trying to change z.

UPDATE

if I do print(z) after the above code I get:

z: [[0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0.]
 [0. 0. 0. 0. 0.]
 [0. 0. 4. 0. 0.]
 [0. 0. 0. 0. 0.]]

but when I use z later in my code I get:

    plt.contour(x[:,0], x[0,:], z.T,linewidths=0.5,colors='k',z=99)
TypeError: 'int' object is not subscriptable

my code seems to be breaking z somehow.

@cᴏʟᴅsᴘᴇᴇᴅ I ask my map if this coordinate, i.e. xx,yy, is on land or not and it returns True or False. I then set the data at the corresponding position in z to 0 if it is water, otherwise I preserve the value — Baz
– Baz, Commented Feb 13, 2018 at 21:32
@cᴏʟᴅsᴘᴇᴇᴅ Its distorting z somhow. Printing z before and after doesn't show any problems (some values have become 0.), but when I use z later in my code I get "TypeError: 'int' object is not subscriptable". If I comment out the code above then I don't get this error, so I must be damaging z in a manner I dont understand. — Baz
– Baz, Commented Feb 13, 2018 at 21:37
If it is setting more values of z, then you should check what m.is_land does. — cs95
– cs95, Commented Feb 13, 2018 at 21:42

cs95 · Accepted Answer · 2018-02-13 21:42:11Z

1

To generate mask, you can use a list comprehension, then reshape the result and set the elements to z to 0 accordingly.

mask = np.reshape([
            m.is_land(i, j) for i, j in zip(x.ravel(), y.ravel())
       ], 
       z.shape
)

z[~mask] = 0

answered Feb 13, 2018 at 21:42

cs95

406k106 gold badges744 silver badges797 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

Baz Over a year ago

Thanks! Tried this with a 100*100 array and it took just a few seconds to compute, but with 500*500 it takes 3 minutes on my machine. Is this inevitable?

cs95 Over a year ago

@Baz I think so. The problem is I cannot vectorize your m.is_land function because I don't know what it is doing.

Baz Over a year ago

@COLDSPEED is_land is a method within the matplotlib.Basemap class. You give it a lat/lng coordinate and it returns if this point is on land or not.

cs95 Over a year ago

@Baz Okay. I don't think such a function is even going to be fast to begin with. What sizes are you typically working with?

Baz Over a year ago

@COLDSPEED 1000*1000 :). I think I might just have to solve this a different way by exporting the map to svg and manipulating it there.

Collectives™ on Stack Overflow

Create an ndarray from another ndarray based on function result

1 Answer 1

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related