def buildT3():
lon = pyg.regularlon(60)
lat = pyg.gausslat(42)
tm = pyg.modeltime365n('1 Jan 2000', 200)
def eps_like(v):
return pyg.Var(v.axes, values=np.random.randn(*v.shape))
X1 = (5. * tm / 2000.) * pyg.cosd(lat)
X2 = pyg.cosd(2 * np.pi * tm / 500.) * pyg.sind(0.5 * lon)
X3 = pyg.sind(2 * np.pi * tm / 120.) * pyg.cosd(3 * lon)**2
X1e = X1 + 0.1 * eps_like(X1)
X2e = X2 + 0.2 * eps_like(X2)
X3e = X3 + 0.2 * eps_like(X3)
Y1 = -1. * pyg.sind(lon) * X1e + eps_like(X1).smooth('lat', 4)
Y2 = -0.2 * X1e + 0.8 * X2e + -0.6 * X3e + 0.1 * eps_like(X1)
X1 = X1.rename('X1')
X2 = X2.rename('X2')
X3 = X3.rename('X3')
X1e = X1e.rename('X1e')
X2e = X2e.rename('X2e')
X3e = X3e.rename('X3e')
Y1 = Y1.rename('Y1')
Y2 = Y2.rename('Y2')
Y3 = Y2.rename('Y3')
return pyg.Dataset(
[X1e, X2e, X3e, Y1, Y2],
atts={'history': 'Synthetic dataset generated by pygeode'})
def buildT2():
nyrs = 10
lat = pyg.regularlat(31)
lon = pyg.regularlon(60)
time = pyg.ModelTime365(values=np.arange(nyrs*365), \
units='days', startdate={'year':2011, 'month':1, 'day':1})
pres = pyg.Pres(np.arange(1000, 0, -50.))
z = 6.6 * pyg.log(1000. / pres)
ts1 = 2 * pyg.sin(2 * np.pi * time / 365.) + 4 * pyg.Var(
(time, ), values=np.random.randn(nyrs * 365))
ts1 = ts1.smooth('time', 20)
ts2 = -5 + 0.6 * time / 365. + 5 * pyg.Var(
(time, ), values=np.random.randn(nyrs * 365))
ts2 = ts2.smooth('time', 20)
T_c = 260. + 40. * pyg.exp(-(
(lat - 10 * np.sin(2 * np.pi * time / 365)) / 45.)**2)
T_wave = 0.05 * lat * pyg.sind(6 * lon - time) # * ts1
T_lapse = -5 * z
Tf = (T_lapse + T_c + T_wave).transpose('time', 'pres', 'lat', 'lon')
Tf.name = 'Temp'
U_c = 40 * pyg.sind(2 * lat)**2 * pyg.sin(2 * np.pi * z / 12)**2
U_wave = 0.08 * lat * pyg.sind(6 * lon - time)
U = (U_c + ts2 * U_wave).transpose('time', 'pres', 'lat', 'lon')
U.name = 'U'
return pyg.Dataset(
[Tf, U],
atts={
'history':
'Synthetic Temperature and Wind data generated by pygeode'
})
def buildT1():
lat = pyg.regularlat(31)
lon = pyg.regularlon(60)
T_c = 260. + 40. * pyg.exp(-(lat/45.)**2)
T_wave = 0.05 * lat * pyg.sind(6*lon)
T = T_c + T_wave
T.name = 'Temp'
T.units = 'K'
return pyg.Dataset([T], atts={'history':'Synthetic Temperature data generated by pygeode'})
import pygeode as pyg
import numpy as np
import pylab as pyl
from pygeode.formats import netcdf as nc
lat = pyg.gausslat(32)
lon = pyg.Lon(np.arange(0, 360, 360 / 64.))
ln_grid, lt_grid = np.meshgrid(lon.values, lat.values)
T_values = 260. + 40 * np.exp(-(lt_grid / 45.)**2) + 0.05 * lt_grid * np.sin(
3 * ln_grid * np.pi / 180.)
T_c = 260. + 40 * np.exp(-(lt_grid / 45.)**2)
T_wave = 0.05 * lt_grid * np.sin(3 * ln_grid * np.pi / 180.)
T = pyg.Var((lat, lon), name='Temp', values=T_c + T_wave, atts={'units': 'K'})
d = pyg.Dataset(
[T], atts={'history': 'Synthetic Temperature data generated by pygeode'})
nc.save('t_sample.nc', d)