-
Notifications
You must be signed in to change notification settings - Fork 0
/
cmip5.py
215 lines (193 loc) · 7.84 KB
/
cmip5.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
import os
from contextlib import closing
from datetime import datetime
import netCDF4
import numpy as np
from scipy.interpolate import griddata, interp1d
class CMIP5:
def __init__(self, filep, scale=1):
self._filep = filep
if isinstance(filep, list):
name = filep[0]
else:
name = filep
key = os.path.basename(name).split('_')[0]
if key.endswith('nc'):
key = os.path.basename(name).split('.')[0]
with closing(netCDF4.MFDataset(filep)) as ds:
lats = ds.variables['lat'][:]
lons = ds.variables['lon'][:]
lons[lons < 0] = lons[lons < 0] + 360
arctic_mask = np.array(lats >= 65)
if lats.ndim == 1:
lats = np.array(lats[arctic_mask])
self.lons, self.lats = np.meshgrid(lons, lats)
self.data = ds.variables[key][:][:, arctic_mask, ...]
else:
arctic_mask = np.any(arctic_mask, axis=1)
self.lats = np.array(lats[arctic_mask, ...])
self.lons = np.array(lons[arctic_mask, ...])
self.data = ds.variables[key][:][:, arctic_mask, ...]
self.data = self.data * scale
self.data = np.ma.array(
self.data,
mask=np.ma.getmaskarray(self.data)
)
ds_time = ds.variables['time']
self.dates = netCDF4.num2date(ds_time[:], ds_time.units)
self.start_date = self.dates[0]
self.end_date = self.dates[-1]
self.times = np.array(
[
(date - self.start_date).total_seconds()
for date in self.dates
]
)
self.units = ds.variables[key].units
self.long_name = ds.variables[key].long_name
self.key = key
self._interp_seconds = None
self._interp_mask = None
self._delta = None
def __repr__(self):
return f'{self.__class__.__name__}({self._filep})'
@property
def mask(self):
mask = np.ma.getmaskarray(self.data[0])
for data in self.data[1:]:
mask = mask & np.ma.getmaskarray(data)
return mask
def get_data(self, time):
time = time + self._delta
data = self._interp_seconds(time)
mask = self._interp_mask(time)
return np.ma.array(data, mask=mask)
def get_date(self, time):
delta = np.timedelta64(int(self._delta), 's')
time = time.astype('timedelta64[s]')
date = (np.datetime64(self.start_date, 's') + delta + time)
return date
def set_interpolation(self):
self._interp_seconds = interp1d(
self.times,
self.data,
axis=0,
fill_value='extrapolate',
)
self._interp_mask = interp1d(
self.times,
np.ma.getmaskarray(self.data),
axis=0,
fill_value='extrapolate',
)
def set_delta(self, ref_date):
self._delta = int((ref_date - self.start_date).total_seconds())
def set_grid_data(self, lats, lons):
same_lats = np.array_equal(lats, self.lats)
same_lons = np.array_equal(lons, self.lons)
if same_lats and same_lons:
return
lats = lats.copy()
lons = lons.copy()
datas = []
points = self.lats.copy().flatten(), self.lons.copy().flatten()
for datain in self.data:
data = griddata(
points=points,
values=datain.flatten(),
xi=(lats, lons),
method='nearest'
)
datas.append(data)
self.data = np.ma.vstack(
[np.ma.expand_dims(d, axis=0) for d in datas]
)
self.lons = lons
self.lats = lats
class CltCMIP5(CMIP5):
leap_years = np.arange(1972, 3000, 4).astype('str').astype('datetime64[Y]')
def __init__(self, filep, scale=1):
super().__init__(filep, scale)
dates = []
groups = []
assert self.dates.astype('datetime64[Y]')[0] in self.leap_years
for i in range(4):
group_years = self.leap_years + i
group_mask = np.isin(
self.dates.astype('datetime64[Y]'),
group_years
)
group_dates = self.dates[group_mask]
num_years = np.unique(group_dates.astype('datetime64[Y]')).size
days_in_year = 365 if i > 0 else 366
data_per_day = group_dates.size // (days_in_year * num_years)
group_shape = (num_years, days_in_year, data_per_day)
group_data = self.data[group_mask, ...]
data_shape = tuple(list(group_shape) + list(group_data.shape)[1:])
year_data = group_data.reshape(data_shape)
avg_data = np.mean(year_data, axis=0)
final_shape = tuple(
[data_per_day * days_in_year] + list(avg_data.shape)[2:]
)
data = avg_data.reshape(final_shape)
groups.append(data)
dates.append(group_dates[:data_per_day * days_in_year])
self.dates = np.concatenate(dates)
self.data = np.concatenate(groups)
self.start_date = self.dates[0]
self.times = np.array(
[
(date - self.start_date).total_seconds()
for date in self.dates
]
)
def set_delta(self, ref_date):
self._delta = np.datetime64(ref_date)
diff = (self.leap_years - np.datetime64(ref_date, 'Y')).astype(int)
ref_leap_year = ref_date.year - int(np.abs(diff[diff < 0]).min())
ref_leap_year = np.datetime64(str(ref_leap_year), 's')
self._ref_leap_year = ref_leap_year
def _fix_future_time(self, time):
if np.all(time <= self.times[-1]):
return time
four_years = np.timedelta64(365 * 3 + 366, 'D')
mask = time > self.times[-1]
time_to_change = time[mask]
dates = self._ref_leap_year + time_to_change.astype('timedelta64[s]')
new_dates = dates - four_years.astype('timedelta64[s]')
new_time = (new_dates - self._ref_leap_year).astype('timedelta64[s]')
new_time = new_time.astype(int)
time[mask] = self._fix_future_time(new_time)
return time
def _fix_past_time(self, time):
if np.all(time >= self.times[0]):
return time
four_years = np.timedelta64(365 * 3 + 366, 'D')
mask = time < self.times[0]
time_to_change = time[mask]
dates = self._ref_leap_year + time_to_change.astype('timedelta64[s]')
new_dates = dates + four_years.astype('timedelta64[s]')
new_time = (new_dates - self._ref_leap_year).astype('timedelta64[s]')
new_time = new_time.astype(int)
time[mask] = self._fix_past_time(new_time)
return time
def get_data(self, time):
time = time.astype('timedelta64[s]')
dates = self._delta + time
time = (dates - self._ref_leap_year).astype('timedelta64[s]')
time = time.astype(int)
time = self._fix_future_time(time)
time = self._fix_past_time(time)
data = self._interp_seconds(time)
mask = self._interp_mask(time)
return np.ma.array(data, mask=mask)
def get_date(self, time):
time = time.astype('timedelta64[s]')
dates = self._delta + time
time = (dates - self._ref_leap_year).astype('timedelta64[s]')
time = time.astype(int)
time = self._fix_future_time(time)
time = self._fix_past_time(time)
dates = np.datetime64(self.start_date) + time.astype('timedelta64[s]')
dates = dates.astype(datetime)
return dates