Scipy Griddata Output Dimensions

Question

I'm not sure what I'm doing wrong. I'm attempting to use scipy griddata to interpolate data in an irregular grid.

from scipy.interpolate import griddata

I have two lists, "x" and "y", that represent the axes of my original, uninterpolated grid. They are both lists of length 8.

Then, I make the arrays that represent the axes of the intended final, filled-in grid.

ny = np.linspace(0.0, max(y), y[len(y)-1]/min_interval+1)
nx = np.linspace(0.0, max(x), len(ny))

I've checked and both "ny" and "nx" are of shape (61,). Then, I create an 8 x 8 list "z". Finally, I attempt to make my final grid.

Z = griddata((np.array(x), np.array(y)), np.array(z), (nx, ny), method='nearest', fill_value=0)
print Z.shape

The resulting 2D array has dimensions (61,8). I tried using "x" and "y" as lists and arrays - no change. Why is it only interpolating in one direction? I was expecting a (61,61) array output. I would have included actual numbers if I felt it would have been helpful, but I don't see how it would make a difference. Do I not understand how griddata works?

Answer 1

Here is the full code:

import numpy as np
from scipy.interpolate import griddata

# random data to interpolate
x = np.array([0, 10, 13, 17, 20, 50, 55, 60.0])
y = np.array([10, 20, 40, 80, 90, 95, 100, 120.0])
zg = np.random.randn(8, 8)

#select one of the following two line, it depends on the order in z
#xg, yg = np.broadcast_arrays(x[:, None], y[None, :])  
xg, yg = np.broadcast_arrays(x[None, :], y[:, None]) 

yg2, xg2 = np.mgrid[y.min()-10:y.max()+10:100j, x.min()-10:x.max()+10:100j]

zg2 = griddata((xg.ravel(), yg.ravel()), zg.ravel(), (xg2.ravel(), yg2.ravel()), method="nearest")
zg2.shape = yg2.shape

import pylab as pl

pl.pcolormesh(xg2, yg2, zg2)
pl.scatter(xg.ravel(), yg.ravel(), c=zg.ravel())

the output is: