MemoryError: Unable to allocate 3.60 GiB for an array with shape (280000, 1728) and data type float64

[douyoujun]

Hi, i I would like to use kriging to interpolate station data to grid, following the code 00_ordinary.py in example directory, i made my code under windows, as follows:

import numpy as np
import pykrige.kriging_tools as kt
from pykrige.ok import OrdinaryKriging
import matplotlib.pyplot as plt

filename=r'D:\workspace\pyb\interp\data.xlsx'
df=pd.read_excel(filename,header=None, names = ['lat','lon','data'] )
lon=df['lon'] 
lat=df['lat']
PM=df['data'] 

olon=np.linspace(70,140,700)
olat=np.linspace(15,55,400)

OK = OrdinaryKriging(
    lon,
    lat,
    PM,
    variogram_model="linear",
    verbose=False,
    enable_plotting=False,
)

z, ss = OK.execute("grid", olon, olat)

but some error occurred when run the last line:
Traceback (most recent call last):

  File "<stdin>", line 1, in <module>
  File "C:\Users\Administrator\anaconda3\lib\site-packages\pykrige\ok.py", line 985, in execute
    zvalues, sigmasq = self._exec_vector(a, bd, mask)
  File "C:\Users\Administrator\anaconda3\lib\site-packages\pykrige\ok.py", line 656, in _exec_vector
    b[:, :n, 0] = -self.variogram_function(self.variogram_model_parameters, bd)
  File "C:\Users\Administrator\anaconda3\lib\site-packages\pykrige\variogram_models.py", line 29, in linear_variogram_model
    return slope * d + nugget
MemoryError: Unable to allocate 3.60 GiB for an array with shape (280000, 1728) and data type float64

it seems that the computer memory resulted in this error, so i try at another computer to run this code, the same error info showed, i don't know how to correct this problem.

any help will be appreciated.

[MuellerSeb]

You could try to use smaller chunks for olon for example:

olon = np.linspace(70,140,700)
olon1 = olon[0:100]
olon2 = olon[100:200]
olon3 = olon[200:300]
...
z1, ss1 = OK.execute("grid", olon1, olat)
z2, ss2 = OK.execute("grid", olon2, olat)
z3, ss3 = OK.execute("grid", olon3, olat)
...

And then you can merge the fields z1,z2,... back together.

You could also use the GSTools kriging routines, which provide a separate argument for chunking because of your exact problem:
https://geostat-framework.readthedocs.io/projects/gstools/en/stable/generated/gstools.krige.Ordinary.html#gstools.krige.Ordinary.__call__

Does this help?
Cheers, Sebastian