def interpolate_u_geo_field(self):
from utils import interpolate_scalar_field
from constants import data_dir_path, mdt_interp_method
from constants import lat_min, lat_max, n_lat, lon_min, lon_max, n_lon
interp_filename_suffix = 'lat' + str(lat_min) + '-' + str(lat_max) + '_n' + str(n_lat) + '_' \
+ 'lon' + str(lon_min) + '-' + str(lon_max) + '_n' + str(n_lon) + '.pickle'
ugeo_interp_filename = 'mdt_cnes_cls2013_global' + '_interp_ugeo_' + interp_filename_suffix
vgeo_interp_filename = 'mdt_cnes_cls2013_global' + '_interp_vgeo_' + interp_filename_suffix
ugeo_interp_filepath = path.join(data_dir_path,
'mdt_cnes_cls2013_global',
ugeo_interp_filename)
vgeo_interp_filepath = path.join(data_dir_path,
'mdt_cnes_cls2013_global',
vgeo_interp_filename)
# TODO: Properly check for masked/filled values.
mask_value_cond = lambda x: x < -100
repeat0tile1 = True
convert_lon_range = True
ugeo_interp, latgrid_interp, longrid_interp = interpolate_scalar_field(
self.u_geo, self.lats, self.lons, ugeo_interp_filepath,
mask_value_cond, 'latlon', mdt_interp_method, repeat0tile1,
convert_lon_range)
vgeo_interp, latgrid_interp, longrid_interp = interpolate_scalar_field(
self.v_geo, self.lats, self.lons, vgeo_interp_filepath,
mask_value_cond, 'latlon', mdt_interp_method, repeat0tile1,
convert_lon_range)
self.ugeo_interp = ugeo_interp
self.vgeo_interp = vgeo_interp
self.latgrid_interp = latgrid_interp
self.longrid_interp = longrid_interp
def interpolate_wind_field(self):
from utils import interpolate_scalar_field
from constants import u_wind_interp_method
from constants import lat_min, lat_max, n_lat, lon_min, lon_max, n_lon
interp_filename_suffix = 'lat' + str(lat_min) + '-' + str(lat_max) + '_n' + str(n_lat) + '_' \
+ 'lon' + str(lon_min) + '-' + str(lon_max) + '_n' + str(n_lon) + '.pickle'
day_of_year = str(self.date.timetuple().tm_yday)
u_wind_interp_filename = 'uwnd.10m.gauss.' + str(self.date.year) + 'day' + day_of_year + \
'_interp_uwind_' + interp_filename_suffix
v_wind_interp_filename = 'vwnd.10m.gauss.' + str(self.date.year) + 'day' + day_of_year + \
'_interp_vwind_' + interp_filename_suffix
u_wind_interp_filepath = path.join(self.interp_dir,
str(self.date.year),
u_wind_interp_filename)
v_wind_interp_filepath = path.join(self.interp_dir,
str(self.date.year),
v_wind_interp_filename)
# TODO: Properly check for masked/filled values.
mask_value_cond = lambda x: np.full(x.shape, False, dtype=bool)
logger.info('lats.shape={}, lons.shape={}, u_wind.shape={}'.format(
self.lats.shape, self.lons.shape, self.u_wind.shape))
u_wind_interp, latgrid_interp, longrid_interp = interpolate_scalar_field(
data=self.u_wind,
x=self.lats,
y=self.lons,
pickle_filepath=u_wind_interp_filepath,
mask_value_cond=mask_value_cond,
grid_type='latlon',
interp_method=u_wind_interp_method,
repeat0tile1=True,
convert_lon_range=True)
v_wind_interp, latgrid_interp, longrid_interp = interpolate_scalar_field(
data=self.v_wind,
x=self.lats,
y=self.lons,
pickle_filepath=v_wind_interp_filepath,
mask_value_cond=mask_value_cond,
grid_type='latlon',
interp_method=u_wind_interp_method,
repeat0tile1=True,
convert_lon_range=True)
# Convert everything to a numpy array otherwise the argmin functions below have to create a new numpy array
# every time, slowing down lookup considerably.
self.u_wind_interp = np.array(u_wind_interp)
self.v_wind_interp = np.array(v_wind_interp)
self.latgrid_interp = np.array(latgrid_interp)
self.longrid_interp = np.array(longrid_interp)
def interpolate_u_geo_field(self):
from utils import interpolate_scalar_field
from constants import lat_min, lat_max, n_lat, lon_min, lon_max, n_lon, u_geo_interp_method
interp_filename_prefix = 'dt_global_allsat_msla_h_20150701_' + str(self.date.year) \
+ str(self.date.month).zfill(2) + str(self.date.day).zfill(2) + '_'
interp_filename_suffix = 'lat' + str(lat_min) + '-' + str(lat_max) + '_n' + str(n_lat) + '_' \
+ 'lon' + str(lon_min) + '-' + str(lon_max) + '_n' + str(n_lon) + '.pickle'
u_geo_interp_filename = interp_filename_prefix + '_interp_u_geo_' + interp_filename_suffix
v_geo_interp_filename = interp_filename_prefix + '_interp_v_geo_' + interp_filename_suffix
u_geo_interp_filepath = path.join(self.u_geo_interp_dir,
str(self.date.year),
u_geo_interp_filename)
v_geo_interp_filepath = path.join(self.u_geo_interp_dir,
str(self.date.year),
v_geo_interp_filename)
# TODO: Properly check for masked/filled values.
mask_value_cond = lambda x: x < -100
u_geo_interp, lats_interp, lons_interp = interpolate_scalar_field(
data=self.u_geo,
x=self.lats,
y=self.lons,
pickle_filepath=u_geo_interp_filepath,
mask_value_cond=mask_value_cond,
grid_type='latlon',
interp_method=u_geo_interp_method,
repeat0tile1=True,
convert_lon_range=True)
v_geo_interp, lats_interp, lons_interp = interpolate_scalar_field(
data=self.v_geo,
x=self.lats,
y=self.lons,
pickle_filepath=v_geo_interp_filepath,
mask_value_cond=mask_value_cond,
grid_type='latlon',
interp_method=u_geo_interp_method,
repeat0tile1=True,
convert_lon_range=True)
# Convert everything to a numpy array otherwise the argmin functions below have to create a new numpy array
# every time, slowing down lookup considerably.
self.u_geo_interp = np.array(u_geo_interp)
self.v_geo_interp = np.array(v_geo_interp)
self.lats_interp = np.array(lats_interp)
self.lons_interp = np.array(lons_interp)
def interpolate_mdt_field(self):
from utils import interpolate_scalar_field
from constants import data_dir_path
from constants import lat_min, lat_max, n_lat, lon_min, lon_max, n_lon
mdt_interp_filename = 'mdt_cnes_cls2013_global' + '_interp_mdt_' \
+ 'lat' + str(lat_min) + '-' + str(lat_max) + '_n' + str(n_lat) + '_' \
+ 'lon' + str(lon_min) + '-' + str(lon_max) + '_n' + str(n_lon) + '.pickle'
mdt_interp_filepath = path.join(data_dir_path,
'mdt_cnes_cls2013_global',
mdt_interp_filename)
# TODO: Properly check for masked/filled values.
mask_value_cond = lambda x: x < -100
repeat0tile1 = True
convert_lon_range = False
mdt_interp, latgrid_interp, longrid_interp = interpolate_scalar_field(
self.mdt, self.lats, self.lons, mdt_interp_filepath,
mask_value_cond, 'latlon', 'cubic', repeat0tile1,
convert_lon_range)
self.mdt_interp = mdt_interp
self.latgrid_interp = latgrid_interp
self.longrid_interp = longrid_interp
def interpolate_geostrophic_current_field(self):
from utils import interpolate_scalar_field
from constants import n_row, n_col, dot_interp_method
interp_filename_prefix = 'CS2_combined_Southern_Ocean' + '_DOT_' + str(
self.month_idx)
interp_filename_suffix = str(n_row) + 'rows_' + str(
n_col) + 'cols.pickle'
interp_filename = interp_filename_prefix + interp_filename_suffix
interp_filepath = path.join(self.interp_dir, interp_filename)
# Mask NaN values.
mask_value_cond = lambda x: np.isnan(x)
dot_interp, row_interp, col_interp = interpolate_scalar_field(
data=np.rot90(self.dot, k=1, axes=(1, 0)),
x=np.arange(len(self.y)),
y=np.arange(len(self.x)),
pickle_filepath=interp_filepath,
mask_value_cond=mask_value_cond,
grid_type='ease_rowcol',
interp_method=dot_interp_method,
repeat0tile1=False,
convert_lon_range=False,
debug_plots=False)
# Convert everything to a numpy array otherwise the argmin functions below have to create a new numpy array
# every time, slowing down lookup considerably.
self.dot_interp = np.array(dot_interp)
self.row_interp = np.array(row_interp)
self.col_interp = np.array(col_interp)
def interpolate_alpha_field(self):
from utils import interpolate_scalar_field
from constants import n_x, n_y, alpha_interp_method
interp_filename_prefix = 'seaice_conc_daily_sh_f17_' + str(self.date.year) \
+ str(self.date.month).zfill(2) + str(self.date.day).zfill(2) \
+ '_v03r00'
interp_filename_suffix = str(n_x) + 'x_' + str(n_y) + 'y.pickle'
alpha_interp_filename = interp_filename_prefix + '_interp_alpha_' + interp_filename_suffix
alpha_interp_filepath = path.join(self.sic_interp_dir,
str(self.date.year),
alpha_interp_filename)
# TODO: Properly check for masked/filled values.
mask_value_cond = lambda x: x > 1
alpha_interp, xgrid_interp, ygrid_interp = interpolate_scalar_field(
data=self.alpha,
x=self.xgrid,
y=self.ygrid,
pickle_filepath=alpha_interp_filepath,
mask_value_cond=mask_value_cond,
grid_type='polar_stereographic_xy',
interp_method=alpha_interp_method,
repeat0tile1=False,
convert_lon_range=False)
# Convert everything to a numpy array otherwise the argmin functions below have to create a new numpy array
# every time, slowing down lookup considerably.
self.alpha_interp = np.array(alpha_interp)
self.xgrid_interp = np.array(xgrid_interp)
self.ygrid_interp = np.array(ygrid_interp)
def interpolate_seaice_motion_field(self):
from utils import interpolate_scalar_field
from constants import n_row, n_col, u_ice_interp_method
if self.monthly:
interp_filename_prefix = 'icemotion.grid.month.' + str(self.date.year) + '.' \
+ str(self.date.month).zfill(2) + '.s.v3'
else:
interp_filename_prefix = 'icemotion.grid.daily.' + str(self.date.year) \
+ str(self.date.timetuple().tm_yday).zfill(3) + '.s.v3'
interp_filename_suffix = str(n_row) + 'rows_' + str(n_col) + 'cols.pickle'
u_ice_interp_filename = interp_filename_prefix + '_interp_u_ice_' + interp_filename_suffix
v_ice_interp_filename = interp_filename_prefix + '_interp_v_ice_' + interp_filename_suffix
u_ice_interp_filepath = path.join(self.seaice_motion_interp_dir, str(self.date.year), u_ice_interp_filename)
v_ice_interp_filepath = path.join(self.seaice_motion_interp_dir, str(self.date.year), v_ice_interp_filename)
# Throw out NaN values and extreme anomalies (u_ice or v_ice > 0.5 m/s).
mask_value_cond = lambda x: np.isnan(x) | (np.abs(x) > 0.5)
u_ice_interp, row_interp, col_interp = interpolate_scalar_field(data=self.u_ice, x=self.x[0], y=self.y[:, 0],
pickle_filepath=u_ice_interp_filepath,
mask_value_cond=mask_value_cond,
grid_type='ease_rowcol',
interp_method=u_ice_interp_method,
repeat0tile1=True, convert_lon_range=False)
v_ice_interp, row_interp, col_interp = interpolate_scalar_field(data=self.v_ice, x=self.x[0], y=self.y[:, 0],
pickle_filepath=v_ice_interp_filepath,
mask_value_cond=mask_value_cond,
grid_type='ease_rowcol',
interp_method=u_ice_interp_method,
repeat0tile1=True, convert_lon_range=False)
# Convert everything to a numpy array otherwise the argmin functions below have to create a new numpy array
# every time, slowing down lookup considerably.
self.u_ice_interp = np.array(u_ice_interp)
self.v_ice_interp = np.array(v_ice_interp)
self.row_interp = np.array(row_interp)
self.col_interp = np.array(col_interp)