Python griddata meshgrid

in Python I want to interpolate some data using scipy.interpolate.griddata(x,y,z,xi,yi).

Since I want my unequal spaced original data on the X-Y grid map on an equal spaced XI-YI grid I have to use a meshgrid as:

X, Y = numpy.meshgrid([1,2,3], [2,5,6,8])
XI,YI = numpy.meshgrid([1,2,3]开发者_如何学C,[4,5,6,7])
print scipy.interpolate.griddata(X,Y,X**2+Y**2,XI,YI)

Unfortunately it seems as scipys' griddata does not accept matrices as input for x,y,z in contrast to matlab's griddata-function. Does anyone has a hint for me how to solve the problem?

The correct call sequence in your case is

print scipy.interpolate.griddata((X.ravel(),Y.ravel()), (X**2+Y**2).ravel(), (XI, YI))

I.e., you need to cast the input data points to 1-d. (This could be fixed to work without the .ravel()s in the next version of Scipy.)

I think you need to reshape your grids, griddata expects a list of points with coordinates in column form:

points = transpose(reshape((X,Y), (2,12)))
pointsI = transpose(reshape((XI,YI), (2,12)))
Z = reshape(X**2+Y**2, 12)

print scipy.interpolate.griddata(points, Z, pointsI)