This notebooks shows how to use the tmask from mesh_mask.nc file to mask variables.
It also compared the timing of using tmask vs searching for zeros.
In [1]:
import netCDF4 as nc
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
In [2]:
f = nc.Dataset('/results/SalishSea/nowcast/11jan15/SalishSea_1h_20150111_20150111_grid_T.nc')
sal = f.variables['vosaline'][:]
t = f.variables['votemper'][:]
Loading and manipulating mask¶
In [3]:
%%timeit -n1 -r1
mesh = nc.Dataset('/data/nsoontie/MEOPAR/NEMO-forcing/grid/mesh_mask_SalishSea2.nc')
tmask = mesh.variables['tmask'][:]
np_mask = np.abs(1-tmask) #tmask is zero on land, one on water. We need the opposite for numpy masking
np_mask=np_mask +np.zeros(sal.shape) #expanding time dimension of tmask so that it can be applied to full array.
1 loops, best of 1: 1.17 s per loop
In [4]:
mesh = nc.Dataset('/data/nsoontie/MEOPAR/NEMO-forcing/grid/mesh_mask_SalishSea2.nc')
tmask = mesh.variables['tmask'][:]
np_mask = np.abs(1-tmask)
np_mask=np_mask +np.zeros(sal.shape)
Masking one variable¶
a. Searching for zeros
In [5]:
%%timeit -n1 -r5
sal_m = np.ma.masked_values(sal,0)
1 loops, best of 5: 8.97 s per loop
b. Using tmask
In [6]:
%%timeit -n1 -r5
sal_t = np.ma.masked_array(sal,mask=np_mask)
1 loops, best of 5: 301 ms per loop
Masking two variables¶
a. Searching for zeros
In [7]:
%%timeit -n1 -r5
sal_m2 = np.ma.masked_values(sal,0)
t_m2 = np.ma.masked_values(t,0)
1 loops, best of 5: 13.6 s per loop
b. Using tmask
In [8]:
%%timeit -n1 -r5
sal_t2 = np.ma.masked_array(sal,mask=np_mask)
t_t2 = np.ma.masked_array(t,mask=np_mask)
1 loops, best of 5: 603 ms per loop
Check masking¶
In [9]:
sal_m = np.ma.masked_values(sal,0)
plt.pcolormesh(sal_m[0,0,:,:])
plt.colorbar()
Out[9]:
<matplotlib.colorbar.Colorbar at 0x7fa159872fd0>
In [10]:
sal_t = np.ma.masked_array(sal,mask=np_mask)
plt.pcolormesh(sal_t[0,0,:,:])
plt.colorbar()
Out[10]:
<matplotlib.colorbar.Colorbar at 0x7fa159723be0>
In [ ]: