-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathcalculate_rmse.py
202 lines (169 loc) · 9.35 KB
/
calculate_rmse.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
import numpy as np
import xarray as xr
import cmocean
from src.loaders import *
from src.colocate import *
from src.metrics import *
from src.filter import *
import os
import pandas as pd
import datetime
import pyinterp
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('--start', type = str, help = 'start date for stats calculation')
parser.add_argument('--end', type = str, help = 'end date for stats calculation')
parser.add_argument('--lon_bin_size', type = float, help = 'size of longitude bins')
parser.add_argument('--lat_bin_size', type = float, help = 'size of latitude bins')
parser.add_argument('--time_bin_size', type = int, help = 'size of time bins in days')
parser.add_argument('--lon_min', type = float, help = 'minimum longitude [0-360] for calculating stats on subdomain')
parser.add_argument('--lon_max', type = float, help = 'maximum longitude [0-360] for calculating stats on subdomain')
parser.add_argument('--lat_min', type = float, help = 'minimum latitude for calculating stats on subdomain')
parser.add_argument('--lat_max', type = float, help = 'maximum latitude for calculating stats on subdomain')
parser.add_argument('--swot_dir', type = str, help = 'path to directory containing SWOT L3 data')
parser.add_argument('--map_dir', type = str, help = 'path to directory containing mapped L4 data')
parser.add_argument('--output_dir', type = str, help = 'path to directory to save results')
parser.add_argument('--output_name', type = str, help = 'filename in which to save results')
parser.add_argument('--map_file_prefix', type = str, help = 'filename up to map date of map files')
args = parser.parse_args()
if args.start is None:
print('start not specified, defaulting to 2023-03-28...')
start_date = '2023-03-28'
else:
start_date = args.start
if np.datetime64(start_date) < np.datetime64('2023-03-28'):
raise ValueError("start must be no earlier than 2023-03-28")
if args.end is None:
print('end not specified, defaulting to 2024-09-16...')
end_date = '2024-09-16'
else:
end_date = args.end
if np.datetime64(end_date) > np.datetime64('2024-09-16'):
raise ValueError("end must be no later than 2024-09-16")
if (args.lon_min is None) or (args.lon_max is None) or (args.lat_min is None) or (args.lat_max is None):
print('At least one of [lon_min, lon_max, lat_min, lat_max] not specified, defaulting to global computation...')
subsetting = False
lon_bounds = (0, 360)
lat_bounds = (-90, 90)
else:
subsetting = True
lon_bounds = (args.lon_min, args.lon_max)
lat_bounds = (args.lat_min, args.lat_max)
if args.lon_bin_size is None:
print('lon_bin_size not specified, defaulting to 1 degree...')
lon_bin_size = 1
else:
lon_bin_size = args.lon_bin_size
if args.lat_bin_size is None:
print('lat_bin_size not specified, defaulting to 1 degree...')
lat_bin_size = 1
else:
lat_bin_size = args.lat_bin_size
if args.time_bin_size is None:
print('time_bin_size not specified, defaulting to 1 day...')
time_bin_size = 1
else:
time_bin_size = args.time_bin_size
if args.swot_dir is None:
print('swot_dir not specified, defaulting to /dat1/smart1n/SWOT/data/SWOT_L3_LR_SSH_EXPERT_1.0.2/')
swot_dir = '/dat1/smart1n/SWOT/data/SWOT_L3_LR_SSH_EXPERT_1.0.2/'
else:
swot_dir = args.swot_dir
if args.map_dir is None:
print('map_dir not specified, defaulting to /dat1/smart1n/NeurOST_SSH-SST/')
map_dir = '/dat1/smart1n/NeurOST_SSH-SST/'
else:
map_dir = args.map_dir
if args.output_dir is None:
print('output_dir not specified, defaulting to ./results/')
output_dir = './results/'
else:
output_dir = args.output_dir
if output_dir[-1] != '/':
output_dir = output_dir + '/'
if args.map_file_prefix is None:
print('map_file_prefix not specified, defaulting to NeurOST_SSH-SST_')
map_file_prefix = 'NeurOST_SSH-SST_'
else:
map_file_prefix = args.map_file_prefix
name_convention = {'prefix': map_file_prefix,
'date_hyphenated': False,
'suffix_format': '_YYYYMMDD.nc',
}
if args.output_name is None:
n_outputs = len(os.listdir(output_dir))
print('output_name not specified, defaulting to ' + f'rmse_ssh{n_outputs}.nc')
output_name = f'rmse_ssh{n_outputs}.nc'
else:
output_name = args.output_name
if '.nc' not in output_name:
output_name = output_name + '.nc'
output_path = output_dir + output_name
start_dt = np.datetime64(start_date)
end_dt = np.datetime64(end_date)
num_days = (((end_dt - start_dt)//time_bin_size).astype('timedelta64[D]') + 1).astype('int')
date_array = np.array([(start_dt + np.timedelta64(n * time_bin_size, 'D')).astype('datetime64[s]') for n in range(num_days)]) #np.arange(start_dt, end_dt + np.timedelta64(1, 'D')).astype('datetime64[s]')
N_lon = int((lon_bounds[1] - lon_bounds[0]) / lon_bin_size) + 1
N_lat = int((lat_bounds[1] - lat_bounds[0]) / lat_bin_size) + 1
lon_bins = np.linspace(lon_bounds[0], lon_bounds[1], N_lon)
lat_bins = np.linspace(lat_bounds[0], lat_bounds[1], N_lat)
ds_results = xr.Dataset(
{
"ssha_sum": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_sum_squares": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_count": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_filtered_sum": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_filtered_sum_squares": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_filtered_count": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"sla_map_sum": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"sla_map_sum_squares": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"sla_map_count": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"sla_map_filtered_sum": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"sla_map_filtered_sum_squares": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"sla_map_filtered_count": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_diff_sum": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_diff_sum_squares": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_diff_count": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_filtered_diff_sum": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_filtered_diff_sum_squares": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_filtered_diff_count": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"sla_map_variance": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_variance": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_mse": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_R2": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"sla_map_filtered_variance": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_filtered_variance": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_filtered_mse": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
"ssha_filtered_R2": (["time", "lat", "lon"], np.full((len(date_array), N_lat, N_lon), np.nan)),
},
coords={
"time": date_array,
"lat": lat_bins,
"lon": lon_bins
}
)
for i, t in enumerate(date_array):
start = t
end = t + np.timedelta64(time_bin_size, 'D')
print(f'Processing: {start} to {end}')
swot_data = SWOT_L3_Dataset(swot_dir, start, end, file_prefix = 'SWOT_L3_LR_SSH_Expert_XXX_YYY_')
map_data = Map_L4_Dataset(map_dir, start, end, name_convention = name_convention)
interp = interp_L4_to_L3(map_data, swot_data)
del swot_data, map_data
if subsetting:
interp = interp.subset(lon_min = lon_min, lon_max = lon_max, lat_min = lat_min, lat_max = lat_max)
if interp is not None:
interp = along_track_filter(interp, scale = 100e3, filt_type = 'high_pass', filt_vars = ['ssha', 'sla_map'])
agg, stats = calc_aggregate_stats(data = interp.ds,
single_vars = ['sla_map', 'ssha', 'sla_map_filtered', 'ssha_filtered'],
pairwise_vars = [['ssha', 'sla_map'], ['ssha_filtered', 'sla_map_filtered']],
lon_bin_size = 1,
lat_bin_size = 1,
lon_bounds = lon_bounds,
lat_bounds = lat_bounds,
)
for var in agg.data_vars:
ds_results[var][i,:,:] = agg[var].values
for var in stats.data_vars:
ds_results[var][i,:,:] = stats[var].values
ds_results.astype('float32').to_netcdf(output_path)