Python Contour Plot irregular spaced data - Memory Error

Question

I am trying to plot a contour plot of non rectangular data without any interpolation (I do not want to use the meshgrid module here). Therefore I tried following code:

x_co=data_U[:,0] #-> 22450 rows
y_co=data_U[:,1] #-> 22450 rows
U_co=data_U[:,3] #-> 22450 rows
u_im=np.zeros((len(u_co),len(u_co))) #24450,22450 array
u_im[zip(y_co,x_co)]=U_co #22450,22450 array

fig1.plt.figure()
plate=pltcontourf(x_co,y_co,u_im,vmin=-10, vmax=10,extent=[x_co[0],x_co[-1],y_co[0],y_co[-1]],origin='upper',levels=level,cmap=plt.cm.jet,aspect='auto')   
plt.show

I do always run out of memory, so the idea is to decrease the resolution by np.meshgrid(xi,yi) with 1000,1000 points. Next step would be interpolating data U_co on the new meshgrid, but the triangulation here connects vertices where it should not. So my non regular grid is not realistic given by the interpolation.

Any advice?

Ed Smith Ed Smith · Accepted Answer · 2015-06-23T09:36:32

You can use scipy.interpolate.griddata to interpolate the data with the method=nearest (see here). Without the data it is difficult to help more but the code would look something like,

import numpy as np
from scipy.interpolate import griddata

x_co=data_U[:,0] #-> 22450 rows
y_co=data_U[:,1] #-> 22450 rows
U_co=data_U[:,3] #-> 22450 rows
interp_grid = np.meshgrid(1000,1000)
newgrid = griddata((x_co, y_co), U_co, interp_grid, method='linear')

plt.imshow(newgrid.T, origin='lower')
plt.show()

If you've got a uniform grid, I'd suggest using imshow instead of contourf as it should be faster.

Python Contour Plot irregular spaced data - Memory Error

1 Answers