def load_data(infile, basin_cube, basin_name):
"""Load, temporally aggregate and spatially slice input data"""
cube = iris.load_cube(infile, 'surface_downward_x_stress')
cube = timeseries.convert_to_annual(cube)
experiment = cube.attributes['experiment_id']
if experiment == 'historicalMisc':
experiment = 'historicalAA'
if not basin_name == 'globe':
if basin_cube:
ndim = cube.ndim
basin_array = uconv.broadcast_array(basin_cube.data,
[ndim - 2, ndim - 1],
cube.shape)
else:
basin_array = uconv.create_basin_array(cube)
cube.data.mask = numpy.where(
(cube.data.mask == False) & (basin_array == basins[basin_name]),
False, True)
sh_constraint = iris.Constraint(latitude=lambda cell: -30.0 <= cell < 0.0)
nh_constraint = iris.Constraint(latitude=lambda cell: 0.0 < cell <= 30.0)
scrit_constraint = iris.Constraint(
latitude=lambda cell: -17.0 <= cell < -13.0)
ncrit_constraint = iris.Constraint(
latitude=lambda cell: 13.0 < cell <= 17.0)
sh_cube = cube.extract(sh_constraint)
nh_cube = cube.extract(nh_constraint)
scrit_cube = cube.extract(scrit_constraint)
ncrit_cube = cube.extract(ncrit_constraint)
sh_mean = sh_cube.collapsed(['longitude', 'latitude'], iris.analysis.MEAN)
nh_mean = nh_cube.collapsed(['longitude', 'latitude'], iris.analysis.MEAN)
scrit_mean = scrit_cube.collapsed(['longitude', 'latitude'],
iris.analysis.MEAN)
ncrit_mean = ncrit_cube.collapsed(['longitude', 'latitude'],
iris.analysis.MEAN)
return sh_mean, nh_mean, scrit_mean, ncrit_mean, experiment
def main(inargs):
"""Run the program."""
# Basin data
hfbasin = True if inargs.var == 'northward_ocean_heat_transport' else False
if inargs.basin_file and not hfbasin:
basin_cube = iris.load_cube(inargs.basin_file)
else:
basin_cube = None
inargs.basin_file = None
# Heat transport data
data_cube = read_data(inargs.infiles, inargs.var, inargs.model, basin_cube)
orig_standard_name = data_cube.standard_name
orig_var_name = data_cube.var_name
history_attribute = get_history_attribute(inargs.infiles, data_cube,
inargs.basin_file, basin_cube)
data_cube.attributes['history'] = gio.write_metadata(
file_info=history_attribute)
# Regrid (if needed)
if inargs.regrid:
data_cube, coord_names, regrid_status = grids.curvilinear_to_rectilinear(
data_cube)
dim_coord_names = [coord.name() for coord in data_cube.dim_coords]
aux_coord_names = [coord.name() for coord in data_cube.aux_coords]
regular_grid = False if aux_coord_names else True
if hfbasin:
assert len(dim_coord_names) == 2
assert dim_coord_names[0] == 'time'
y_axis_name = dim_coord_names[1]
else:
assert len(dim_coord_names) == 3
assert dim_coord_names[0] == 'time'
y_axis_name, x_axis_name = dim_coord_names[1:]
for aux_coord in aux_coord_names:
data_cube.remove_coord(aux_coord)
# Basin array
if inargs.basin_file and not inargs.regrid:
ndim = cube.ndim
basin_array = uconv.broadcast_array(basin_cube.data,
[ndim - 2, ndim - 1], cube.shape)
basin_list = ['atlantic', 'pacific', 'indian', 'globe']
elif regular_grid and not hfbasin:
basin_array = uconv.create_basin_array(data_cube)
basin_list = ['atlantic', 'pacific', 'indian', 'globe']
else:
basin_list = ['globe']
# Calculate output for each basin
out_cubes = []
for basin_name in basin_list:
data_cube_copy = data_cube.copy()
if not basin_name == 'globe':
data_cube_copy.data.mask = numpy.where(
(data_cube_copy.data.mask == False) &
(basin_array == basins[basin_name]), False, True)
# Zonal mean
if hfbasin:
zonal_cube = data_cube_copy
else:
zonal_cube = data_cube_copy.collapsed(x_axis_name,
iris.analysis.SUM)
zonal_cube.remove_coord(x_axis_name)
# Convergence
zonal_cube = convergence(zonal_cube, y_axis_name)
# Attributes
standard_name = 'zonal_sum_%s_convergence_%s' % (orig_standard_name,
basin_name)
var_name = '%s_czs_%s' % (orig_var_name, basin_name)
iris.std_names.STD_NAMES[standard_name] = {
'canonical_units': zonal_cube.units
}
zonal_cube.standard_name = standard_name
zonal_cube.long_name = standard_name.replace('_', ' ')
zonal_cube.var_name = var_name
out_cubes.append(zonal_cube)
out_cubes = iris.cube.CubeList(out_cubes)
iris.save(out_cubes, inargs.outfile)
def main(inargs):
"""Run the program."""
try:
time_constraint = gio.get_time_constraint(inargs.time)
except AttributeError:
time_constraint = iris.Constraint()
with iris.FUTURE.context(cell_datetime_objects=True):
data_cubes = iris.load(inargs.infiles, inargs.var & time_constraint, callback=save_history)
equalise_attributes(data_cubes)
climatology_cube = read_climatology(inargs.climatology_file, inargs.var)
basin_cube = read_optional(inargs.basin_file)
depth_cube = read_optional(inargs.depth_file)
atts = set_attributes(inargs, data_cubes[0], climatology_cube, basin_cube, depth_cube)
out_cubes = []
for data_cube in data_cubes:
standard_name = data_cube.standard_name
var_name = data_cube.var_name
if climatology_cube:
data_cube = data_cube - climatology_cube
if basin_cube:
data_cube = uconv.mask_marginal_seas(data_cube, basin_cube)
data_cube, coord_names, regrid_status = regrid_cube(data_cube)
if regrid_status:
depth_cube = None
# FIXME: Could delete depth file from atts
assert coord_names[-3:] == ['depth', 'latitude', 'longitude']
depth_axis = data_cube.coord('depth')
assert depth_axis.units in ['m', 'dbar'], "Unrecognised depth axis units"
out_list = iris.cube.CubeList([])
start_indexes, step = uconv.get_chunks(data_cube.shape, coord_names, chunk=inargs.chunk)
for index in start_indexes:
cube_slice = data_cube[index:index+step, 0:1000, ...]
# Vertical
for layer in vertical_layers.keys():
vertical_mean = calc_vertical_mean(cube_slice, layer, coord_names, atts, standard_name, var_name)
out_list.append(vertical_mean)
if layer in ['surface', 'argo']:
for basin in basins.keys():
if basin_cube and not regrid_status:
basin_array = basin_cube.data
else:
basin_array = uconv.create_basin_array(vertical_mean)
out_list.append(calc_zonal_vertical_mean(vertical_mean.copy(), depth_cube, basin_array, basin, layer, atts, standard_name, var_name))
# Zonal
if basin_cube and not regrid_status:
ndim = cube_slice.ndim
basin_array = uconv.broadcast_array(basin_cube.data, [ndim - 2, ndim - 1], cube_slice.shape)
else:
basin_array = uconv.create_basin_array(cube_slice)
for basin in basins.keys():
out_list.append(calc_zonal_mean(cube_slice.copy(), basin_array, basin, atts, standard_name, var_name))
out_cubes.append(out_list.concatenate())
del out_list
del cube_slice
del basin_array
cube_list = []
nvars = len(vertical_layers.keys()) + len(basins.keys()) + 2*len(basins.keys())
for var_index in range(0, nvars):
temp_list = []
for infile_index in range(0, len(inargs.infiles)):
temp_list.append(out_cubes[infile_index][var_index])
temp_list = iris.cube.CubeList(temp_list)
cube_list.append(temp_list.concatenate_cube())
cube_list = iris.cube.CubeList(cube_list)
assert cube_list[0].data.dtype == numpy.float32
if not 'time' in coord_names:
iris.FUTURE.netcdf_no_unlimited = True
iris.save(cube_list, inargs.outfile)
def main(inargs):
"""Run the program."""
region = inargs.region.replace('-', '_')
# Basin data
hfbasin = True if inargs.var == 'northward_ocean_heat_transport' else False
if not hfbasin:
assert inargs.basin_file, "Must provide a basin file for hfy data"
basin_cube = iris.load_cube(inargs.basin_file)
else:
basin_cube = None
inargs.basin_file = None
# Heat transport data
data_cube = read_data(inargs.infiles, inargs.var, basin_cube, region)
orig_standard_name = data_cube.standard_name
orig_var_name = data_cube.var_name
history_attribute = get_history_attribute(inargs.infiles, data_cube,
inargs.basin_file, basin_cube)
data_cube.attributes['history'] = gio.write_metadata(
file_info=history_attribute)
# Regrid (if needed)
if inargs.regrid:
data_cube, coord_names, regrid_status = grids.curvilinear_to_rectilinear(
data_cube)
dim_coord_names = [coord.name() for coord in data_cube.dim_coords]
aux_coord_names = [coord.name() for coord in data_cube.aux_coords]
regular_grid = False if aux_coord_names else True
if hfbasin:
assert len(dim_coord_names) == 2
assert dim_coord_names[0] == 'time'
y_axis_name = dim_coord_names[1]
else:
assert len(dim_coord_names) == 3
assert dim_coord_names[0] == 'time'
y_axis_name, x_axis_name = dim_coord_names[1:]
for aux_coord in aux_coord_names:
data_cube.remove_coord(aux_coord)
# Basin array
if inargs.basin_file and not inargs.regrid:
ndim = data_cube.ndim
basin_array = uconv.broadcast_array(basin_cube.data,
[ndim - 2, ndim - 1],
data_cube.shape)
elif regular_grid and not hfbasin:
basin_array = uconv.create_basin_array(data_cube)
# Calculate the zonal sum (if required)
data_cube_copy = data_cube.copy()
if hfbasin:
zonal_cube = data_cube_copy
else:
zonal_cube = data_cube_copy.collapsed(x_axis_name, iris.analysis.SUM)
zonal_cube.remove_coord(x_axis_name)
# Attributes
try:
zonal_cube.remove_coord('region')
except iris.exceptions.CoordinateNotFoundError:
pass
standard_name = 'northward_ocean_heat_transport'
var_name = 'hfbasin'
zonal_cube.standard_name = standard_name
zonal_cube.long_name = standard_name.replace('_', ' ')
zonal_cube.var_name = var_name
iris.save(zonal_cube, inargs.outfile)
def main(inargs):
"""Run the program."""
cube = iris.load(inargs.infiles,
gio.check_iris_var(inargs.var),
callback=save_history)
atts = cube[0].attributes
equalise_attributes(cube)
iris.util.unify_time_units(cube)
cube = cube.concatenate_cube()
cube = gio.check_time_units(cube)
cube.attributes = atts
orig_long_name = cube.long_name
if cube.standard_name == None:
orig_standard_name = orig_long_name.replace(' ', '_')
else:
orig_standard_name = cube.standard_name
orig_var_name = cube.var_name
# Temporal smoothing
cube = timeseries.convert_to_annual(cube, full_months=True)
# Mask marginal seas
if inargs.basin:
if '.nc' in inargs.basin:
basin_cube = iris.load_cube(inargs.basin_file)
cube = uconv.mask_marginal_seas(cube, basin_cube)
else:
basin_cube = 'create'
else:
basin_cube = None
# Regrid (if needed)
if inargs.regrid:
cube, coord_names, regrid_status = grids.curvilinear_to_rectilinear(
cube)
# Change units (remove m-2)
if inargs.area:
cube = multiply_by_area(cube, inargs.area)
cube.attributes = atts
cube.long_name = orig_long_name
cube.standard_name = orig_standard_name
cube.var_name = orig_var_name
# History
history_attribute = get_history_attribute(inargs.infiles[0], history[0])
cube.attributes['history'] = gio.write_metadata(
file_info=history_attribute)
# Calculate output for each basin
if type(basin_cube) == iris.cube.Cube:
ndim = cube.ndim
basin_array = uconv.broadcast_array(basin_cube.data,
[ndim - 2, ndim - 1], cube.shape)
basin_list = ['atlantic', 'pacific', 'indian', 'globe']
elif type(basin_cube) == str:
basin_array = uconv.create_basin_array(cube)
basin_list = ['atlantic', 'pacific', 'indian', 'globe']
else:
basin_array = None
basin_list = ['globe']
dim_coord_names = [coord.name() for coord in cube.dim_coords]
aux_coord_names = [coord.name() for coord in cube.aux_coords]
assert len(dim_coord_names) == 3
assert dim_coord_names[0] == 'time'
x_axis_name = dim_coord_names[2]
for aux_coord in aux_coord_names:
cube.remove_coord(aux_coord)
out_cubes = []
for basin_name in basin_list:
data_cube = cube.copy()
if not basin_name == 'globe':
data_cube.data.mask = numpy.where(
(data_cube.data.mask == False) &
(basin_array == basins[basin_name]), False, True)
# Zonal statistic
zonal_cube = data_cube.collapsed(
x_axis_name, aggregation_functions[inargs.zonal_stat])
zonal_cube.remove_coord(x_axis_name)
# Attributes
standard_name = 'zonal_%s_%s_%s' % (inargs.zonal_stat,
orig_standard_name, basin_name)
var_name = '%s_%s_%s' % (orig_var_name,
aggregation_abbreviations[inargs.zonal_stat],
basin_name)
iris.std_names.STD_NAMES[standard_name] = {
'canonical_units': zonal_cube.units
}
zonal_cube.standard_name = standard_name
zonal_cube.long_name = standard_name.replace('_', ' ')
zonal_cube.var_name = var_name
out_cubes.append(zonal_cube)
out_cubes = iris.cube.CubeList(out_cubes)
iris.save(out_cubes, inargs.outfile)
def main(inargs):
"""Run the program."""
file_dict, tas_dict, area_dict, basin_dict = read_data(inargs)
metadata_dict = {}
climatology_dict = {}
time_trend_dict = {}
tas_scaled_trend_dict = {}
branch_dict = {}
for experiment in [
'historical', 'historicalGHG', 'historicalAA', 'historicalnoAA',
'piControl'
]:
filenames = file_dict[experiment]
if not filenames:
climatology_dict[experiment] = None
time_trend_dict[experiment] = None
tas_scaled_trend_dict[experiment] = None
else:
print(experiment)
try:
time_constraint = gio.get_time_constraint(inargs.total_time)
except (AttributeError, TypeError):
time_constraint = iris.Constraint()
with iris.FUTURE.context(cell_datetime_objects=True):
cube = iris.load(filenames, gio.check_iris_var(inargs.var))
# Merge cubes
metadata_dict[filenames[0]] = cube[0].attributes['history']
equalise_attributes(cube)
iris.util.unify_time_units(cube)
cube = cube.concatenate_cube()
cube = gio.check_time_units(cube)
# Time extraction and branch time info
coord_names = [coord.name() for coord in cube.dim_coords]
assert coord_names[0] == 'time'
if 'historical' in experiment:
original_time_length = cube.shape[0]
cube = cube.extract(time_constraint)
new_time_length = cube.shape[0]
branch_time_index_offset = original_time_length - new_time_length
branch_time = cube.attributes['branch_time']
time_length = cube.shape[0]
branch_dict[experiment] = (branch_time, time_length,
branch_time_index_offset)
elif experiment == 'piControl':
branch_time, time_length, branch_time_index_offset = branch_dict[
'historical']
start_index, error = uconv.find_nearest(
cube.coord('time').points,
float(branch_time) + 15.5,
index=True)
if abs(error) > 15:
print(
"WARNING: Large error of %f in locating branch time"
% (error))
start_index = 0
start_index = start_index + branch_time_index_offset
cube = cube[start_index:start_index + time_length, ::]
# Temporal smoothing
cube = timeseries.convert_to_annual(cube, full_months=True)
# Mask marginal seas
if basin_dict[experiment]:
basin_cube = iris.load_cube(basin_dict[experiment])
cube = uconv.mask_marginal_seas(cube, basin_cube)
# Regrid and select basin
cube, coord_names, regrid_status = grids.curvilinear_to_rectilinear(
cube)
if not inargs.basin == 'globe':
if basin_dict[experiment] and not regrid_status:
ndim = cube.ndim
basin_array = uconv.broadcast_array(
basin_cube.data, [ndim - 2, ndim - 1], cube.shape)
else:
basin_array = uconv.create_basin_array(cube)
cube.data.mask = numpy.where(
(cube.data.mask == False) &
(basin_array == basins[inargs.basin]), False, True)
# Scale
cube, units = scale_data(cube,
inargs.var,
reverse_sign=inargs.reverse_sign)
# Zonal statistic
if inargs.area_adjust:
if regrid_status:
area_dict[experiment] = None
cube, units, metadata_dict = area_ajustment(
cube, area_dict[experiment], metadata_dict)
zonal_cube = cube.collapsed('longitude', iris.analysis.SUM)
aggregation = 'Zonally integrated'
else:
zonal_cube = cube.collapsed('longitude',
iris.analysis.MEAN)
aggregation = 'Zonal mean'
zonal_cube.remove_coord('longitude')
# Climatology and trends
climatology_dict[experiment] = calculate_climatology(
zonal_cube, inargs.climatology_time, experiment)
time_trend_dict[experiment] = get_trend_cube(zonal_cube)
if tas_dict[experiment]:
tas_cube = iris.load_cube(
tas_dict[experiment],
'air_temperature' & time_constraint)
scale_factor = get_scale_factor(tas_cube)
print(experiment, 'warming:', scale_factor)
tas_scaled_trend_dict[experiment] = time_trend_dict[
experiment] * (1. / abs(scale_factor))
metadata_dict[tas_dict[experiment]
[0]] = tas_cube.attributes['history']
else:
tas_scaled_trend_dict[experiment] = None
# Create the plots
tas_scaled_trend_flag = tas_scaled_trend_dict[
'historicalGHG'] and tas_scaled_trend_dict['historicalAA']
fig = plt.figure(figsize=[15, 20])
gs = set_plot_grid(tas_trend=tas_scaled_trend_flag)
ax_main = plt.subplot(gs[0])
plt.sca(ax_main)
plot_climatology(climatology_dict, inargs.var, units, inargs.legloc,
aggregation)
plt.title('%s (%s), %s' % (inargs.model, inargs.run, inargs.basin))
ax_diff = plt.subplot(gs[1])
plt.sca(ax_diff)
plot_difference(climatology_dict)
ax_time_trend = plt.subplot(gs[2])
plt.sca(ax_time_trend)
plot_trend(time_trend_dict, units)
if tas_scaled_trend_flag:
ax_tas_trend = plt.subplot(gs[3])
plt.sca(ax_tas_trend)
plot_trend(tas_scaled_trend_dict, units, scaled=True)
plt.xlabel('latitude')
plt.savefig(inargs.outfile, bbox_inches='tight')
gio.write_metadata(inargs.outfile, file_info=metadata_dict)