def var_long_names(refdes):
# get science variable long names from the Data Review Database
stream_vars_dict = dict()
dr = cf.refdes_datareview_json(refdes)
for x in dr['instrument']['data_streams']:
dr_ms = '-'.join((x['method'], x['stream_name']))
sci_vars = dict()
for y in x['stream']['parameters']:
if (y['data_product_type'] == 'Science Data') or (y['data_product_type'] == 'Unprocessed Data'):
sci_vars.update({y['display_name']: dict(db_units=y['unit'], var_name=y['name'])})
if len(sci_vars) > 0:
stream_vars_dict.update({dr_ms: sci_vars})
return stream_vars_dict
def main(url_list, sDir, plot_type, start_time, end_time, deployment_num):
for i, u in enumerate(url_list):
elements = u.split('/')[-2].split('-')
r = '-'.join((elements[1], elements[2], elements[3], elements[4]))
ms = u.split(r + '-')[1].split('/')[0]
subsite = r.split('-')[0]
array = subsite[0:2]
main_sensor = r.split('-')[-1]
datasets = cf.get_nc_urls([u])
datasets_sel = cf.filter_collocated_instruments(main_sensor, datasets)
save_dir = os.path.join(sDir, array, subsite, r, plot_type)
cf.create_dir(save_dir)
sname = '-'.join((r, ms, 'track'))
print('Appending....')
sh = pd.DataFrame()
deployments = []
end_times = []
for ii, d in enumerate(datasets_sel):
print('\nDataset {} of {}: {}'.format(ii + 1, len(datasets_sel),
d.split('/')[-1]))
ds = xr.open_dataset(d, mask_and_scale=False)
ds = ds.swap_dims({'obs': 'time'})
if start_time is not None and end_time is not None:
ds = ds.sel(time=slice(start_time, end_time))
if len(ds['time'].values) == 0:
print(
'No data to plot for specified time range: ({} to {})'.
format(start_time, end_time))
continue
fname, subsite, refdes, method, stream, deployment = cf.nc_attributes(
d)
if deployment_num is not None:
if int(deployment.split('0')[-1]) is not deployment_num:
print(type(int(deployment.split('0')[-1])),
type(deployment_num))
continue
# get end times of deployments
ps_df, n_streams = cf.get_preferred_stream_info(r)
dr_data = cf.refdes_datareview_json(r)
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = cf.get_deployment_information(
dr_data, int(deploy[-4:]))
if int(deploy[-4:]) not in deployments:
deployments.append(int(deploy[-4:]))
if pd.to_datetime(deploy_info['stop_date']) not in end_times:
end_times.append(pd.to_datetime(deploy_info['stop_date']))
data = {'lat': ds['lat'].values, 'lon': ds['lon'].values}
new_r = pd.DataFrame(data,
columns=['lat', 'lon'],
index=ds['time'].values)
sh = sh.append(new_r)
xD = sh.lon.values
yD = sh.lat.values
tD = sh.index.values
clabel = 'Time'
ylabel = 'Latitude'
xlabel = 'Longitude'
fig, ax = pf.plot_profiles(xD,
yD,
tD,
ylabel,
xlabel,
clabel,
end_times,
deployments,
stdev=None)
ax.invert_yaxis()
ax.set_title('Glider Track - ' + r + '\n' + 'x: platform location',
fontsize=9)
ax.set_xlim(-71.75, -69.75)
ax.set_ylim(38.75, 40.75)
#cbar.ax.set_yticklabels(end_times)
# add Pioneer glider sampling area
ax.add_patch(
Rectangle((-71.5, 39.0),
1.58,
1.67,
linewidth=3,
edgecolor='b',
facecolor='none'))
ax.text(-71,
40.6,
'Pioneer Glider Sampling Area',
color='blue',
fontsize=8)
# add Pioneer AUV sampling area
# ax.add_patch(Rectangle((-71.17, 39.67), 0.92, 1.0, linewidth=3, edgecolor='m', facecolor='none'))
array_loc = cf.return_array_subsites_standard_loc(array)
ax.scatter(array_loc.lon,
array_loc.lat,
s=40,
marker='x',
color='k',
alpha=0.3)
#ax.legend(legn, array_loc.index, scatterpoints=1, loc='lower left', ncol=4, fontsize=8)
pf.save_fig(save_dir, sname)
def main(url_list, sDir, mDir, zcell_size, zdbar, start_time, end_time, inpercentile):
"""""
URL : path to instrument data by methods
sDir : path to the directory on your machine to save plots
mDir : path to the directory on your machine to save data ranges
zcell_size : depth cell size to group data
zdbar : define depth where suspect data are identified
start_time : select start date to slice timeseries
end_time : select end date to slice timeseries
"""""
rd_list = []
ms_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
ms = uu.split(rd + '-')[1].split('/')[0]
if rd not in rd_list:
rd_list.append(rd)
if ms not in ms_list:
ms_list.append(ms)
for r in rd_list:
print('\n{}'.format(r))
subsite = r.split('-')[0]
array = subsite[0:2]
main_sensor = r.split('-')[-1]
# read in the analysis file
dr_data = cf.refdes_datareview_json(r)
# get the preferred stream information
ps_df, n_streams = cf.get_preferred_stream_info(r)
# get science variable long names from the Data Review Database
stream_sci_vars = cd.sci_var_long_names(r)
# check if the science variable long names are the same for each stream and initialize empty arrays
sci_vars_dict0 = cd.sci_var_long_names_check(stream_sci_vars)
# get the list of data files and filter out collocated instruments and other streams
datasets = []
for u in url_list:
print(u)
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join((splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
fdatasets = cf.filter_collocated_instruments(main_sensor, datasets)
fdatasets = cf.filter_other_streams(r, ms_list, fdatasets)
# select the list of data files from the preferred dataset for each deployment
fdatasets_final = []
for ii in range(len(ps_df)):
for x in fdatasets:
if ps_df['deployment'][ii] in x and ps_df[0][ii] in x:
fdatasets_final.append(x)
# build dictionary of science data from the preferred dataset for each deployment
print('\nAppending data from files')
sci_vars_dict, y_unit, y_name, l0 = cd.append_evaluated_science_data(
sDir, ps_df, n_streams, r, fdatasets_final, sci_vars_dict0, zdbar, start_time, end_time)
# get end times of deployments
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = cf.get_deployment_information(dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
# create data range output folders
save_dir_stat = os.path.join(mDir, array, subsite)
cf.create_dir(save_dir_stat)
# create plots output folder
save_fdir = os.path.join(sDir, array, subsite, r, 'data_range')
cf.create_dir(save_fdir)
stat_df = pd.DataFrame()
"""
create data ranges csv file and figures
"""
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print('\n' + vinfo['var_name'])
if len(vinfo['t']) < 1:
print('no variable data to plot')
continue
else:
sv_units = vinfo['units'][0]
fv = vinfo['fv'][0]
t = vinfo['t']
z = vinfo['values']
y = vinfo['pressure']
# Check if the array is all NaNs
if sum(np.isnan(z)) == len(z):
print('Array of all NaNs - skipping plot.')
continue
# Check if the array is all fill values
elif len(z[z != fv]) == 0:
print('Array of all fill values - skipping plot.')
continue
else:
if len(y) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((vinfo['var_name'], r))
else:
sname = '-'.join((vinfo['var_name'], r, m))
"""
create data ranges for non - pressure data only
"""
if 'pressure' in vinfo['var_name']:
pass
else:
columns = ['tsec', 'dbar', str(vinfo['var_name'])]
# create depth ranges
min_r = int(round(min(y) - zcell_size))
max_r = int(round(max(y) + zcell_size))
ranges = list(range(min_r, max_r, zcell_size))
# group data by depth
groups, d_groups = gt.group_by_depth_range(t, y, z, columns, ranges)
print('writing data ranges for {}'.format(vinfo['var_name']))
stat_data = groups.describe()[vinfo['var_name']]
stat_data.insert(loc=0, column='parameter', value=sv, allow_duplicates=False)
t_deploy = deployments[0]
for i in range(len(deployments))[1:len(deployments)]:
t_deploy = '{}, {}'.format(t_deploy, deployments[i])
stat_data.insert(loc=1, column='deployments', value=t_deploy, allow_duplicates=False)
stat_df = stat_df.append(stat_data, ignore_index=False)
"""
plot full time range free from errors and suspect data
"""
clabel = sv + " (" + sv_units + ")"
ylabel = (y_name[0][0] + " (" + y_unit[0][0] + ")")
t_eng = None
m_water_depth = None
# plot non-erroneous -suspect data
fig, ax, bar = pf.plot_xsection(subsite, t, y, z, clabel, ylabel, t_eng, m_water_depth,
inpercentile, stdev=None)
title0 = 'Data colored using the upper and lower {} percentile.'.format(inpercentile)
ax.set_title(r+'\n'+title0, fontsize=9)
leg_text = ('{} % erroneous values removed after Human In the Loop review'.format(
(len(t)/l0) * 100),)
ax.legend(leg_text, loc='upper center', bbox_to_anchor=(0.5, -0.17), fontsize=6)
for ii in range(len(end_times)):
ax.axvline(x=end_times[ii], color='b', linestyle='--', linewidth=.8)
ax.text(end_times[ii], min(y)-5, 'End' + str(deployments[ii]),
fontsize=6, style='italic',
bbox=dict(boxstyle='round',
ec=(0., 0.5, 0.5),
fc=(1., 1., 1.),
))
# fig.tight_layout()
sfile = '_'.join(('data_range', sname))
pf.save_fig(save_fdir, sfile)
# write stat file
stat_df.to_csv('{}/{}_data_ranges.csv'.format(save_dir_stat, r), index=True, float_format='%11.6f')
def main(sDir, url_list):
reviewlist = pd.read_csv(
'https://raw.githubusercontent.com/ooi-data-lab/data-review-prep/master/review_list/data_review_list.csv'
)
rd_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
if rd not in rd_list:
rd_list.append(rd)
json_file_list = []
for r in rd_list:
dependencies = []
print('\n{}'.format(r))
data = OrderedDict(deployments=OrderedDict())
save_dir = os.path.join(sDir, r.split('-')[0], r)
cf.create_dir(save_dir)
# Deployment location test
deploy_loc_test = cf.deploy_location_check(r)
data['location_comparison'] = deploy_loc_test
for u in url_list:
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join(
(splitter[1], splitter[2], splitter[3], splitter[4]))
catalog_rms = '-'.join((r, splitter[-2], splitter[-1]))
# complete the analysis by reference designator
if rd_check == r:
udatasets = cf.get_nc_urls([u])
# check for the OOI 1.0 datasets for review
rl_filtered = reviewlist.loc[
(reviewlist['Reference Designator'] == r)
& (reviewlist['status'] == 'for review')]
review_deployments = rl_filtered['deploymentNumber'].tolist()
review_deployments_int = [
'deployment%04d' % int(x) for x in review_deployments
]
for rev_dep in review_deployments_int:
rdatasets = [s for s in udatasets if rev_dep in s]
if len(rdatasets) > 0:
datasets = []
for dss in rdatasets: # filter out collocated data files
if catalog_rms == dss.split('/')[-1].split(
'_20')[0][15:]:
datasets.append(dss)
else:
drd = dss.split('/')[-1].split('_20')[0][15:42]
if drd not in dependencies and drd != r:
dependencies.append(drd)
notes = []
time_ascending = ''
if len(datasets) == 1:
try:
ds = xr.open_dataset(datasets[0],
mask_and_scale=False)
ds = ds.swap_dims({'obs': 'time'})
fname, subsite, refdes, method, data_stream, deployment = cf.nc_attributes(
datasets[0])
except OSError:
print('OSError - skipping file {}'.format(
datasets[0]))
continue
elif len(datasets) > 1:
ds = xr.open_mfdataset(datasets,
mask_and_scale=False)
ds = ds.swap_dims({'obs': 'time'})
#ds = ds.chunk({'time': 100})
fname, subsite, refdes, method, data_stream, deployment = cf.nc_attributes(
datasets[0])
fname = fname.split('_20')[0]
notes.append('multiple deployment .nc files')
# when opening multiple datasets, don't check that the timestamps are in ascending order
time_ascending = 'not_tested'
else:
continue
print('\nAnalyzing file: {}'.format(fname))
# Get info from the data review database
dr_data = cf.refdes_datareview_json(refdes)
stream_vars = cf.return_stream_vars(data_stream)
sci_vars = cf.return_science_vars(data_stream)
node = refdes.split('-')[1]
if 'cspp' in data_stream or 'WFP' in node:
sci_vars.append('int_ctd_pressure')
# if 'FDCHP' in refdes:
# remove_vars = ['fdchp_wind_x', 'fdchp_wind_y', 'fdchp_wind_z', 'fdchp_speed_of_sound_sonic',
# 'fdchp_x_accel_g', 'fdchp_y_accel_g', 'fdchp_z_accel_g']
# rv_regex = re.compile('|'.join(remove_vars))
# rv_sci_vars = [nn for nn in sci_vars if not rv_regex.search(nn)]
# sci_vars = rv_sci_vars
deploy_info = get_deployment_information(
dr_data, int(deployment[-4:]))
# Grab deployment Variables
deploy_start = str(deploy_info['start_date'])
deploy_stop = str(deploy_info['stop_date'])
deploy_lon = deploy_info['longitude']
deploy_lat = deploy_info['latitude']
deploy_depth = deploy_info['deployment_depth']
# Calculate days deployed
if deploy_stop != 'None':
r_deploy_start = pd.to_datetime(
deploy_start).replace(hour=0,
minute=0,
second=0)
if deploy_stop.split('T')[1] == '00:00:00':
r_deploy_stop = pd.to_datetime(deploy_stop)
else:
r_deploy_stop = (pd.to_datetime(deploy_stop) +
timedelta(days=1)).replace(
hour=0,
minute=0,
second=0)
n_days_deployed = (r_deploy_stop -
r_deploy_start).days
else:
n_days_deployed = None
# Add reference designator to dictionary
try:
data['refdes']
except KeyError:
data['refdes'] = refdes
deployments = data['deployments'].keys()
data_start = pd.to_datetime(min(
ds['time'].values)).strftime('%Y-%m-%dT%H:%M:%S')
data_stop = pd.to_datetime(max(
ds['time'].values)).strftime('%Y-%m-%dT%H:%M:%S')
# Add deployment and info to dictionary and initialize delivery method sub-dictionary
if deployment not in deployments:
data['deployments'][deployment] = OrderedDict(
deploy_start=deploy_start,
deploy_stop=deploy_stop,
n_days_deployed=n_days_deployed,
lon=deploy_lon,
lat=deploy_lat,
deploy_depth=deploy_depth,
method=OrderedDict())
# Add delivery methods to dictionary and initialize stream sub-dictionary
methods = data['deployments'][deployment][
'method'].keys()
if method not in methods:
data['deployments'][deployment]['method'][
method] = OrderedDict(stream=OrderedDict())
# Add streams to dictionary and initialize file sub-dictionary
streams = data['deployments'][deployment]['method'][
method]['stream'].keys()
if data_stream not in streams:
data['deployments'][deployment]['method'][method][
'stream'][data_stream] = OrderedDict(
file=OrderedDict())
# Get a list of data gaps >1 day
time_df = pd.DataFrame(ds['time'].values,
columns=['time'])
gap_list = cf.timestamp_gap_test(time_df)
# Calculate the sampling rate to the nearest second
time_df['diff'] = time_df['time'].diff().astype(
'timedelta64[s]')
rates_df = time_df.groupby(['diff']).agg(['count'])
n_diff_calc = len(time_df) - 1
rates = dict(n_unique_rates=len(rates_df),
common_sampling_rates=dict())
for i, row in rates_df.iterrows():
percent = (float(row['time']['count']) /
float(n_diff_calc))
if percent > 0.1:
rates['common_sampling_rates'].update(
{int(i): '{:.2%}'.format(percent)})
sampling_rt_sec = None
for k, v in rates['common_sampling_rates'].items():
if float(v.strip('%')) > 50.00:
sampling_rt_sec = k
if not sampling_rt_sec:
sampling_rt_sec = 'no consistent sampling rate: {}'.format(
rates['common_sampling_rates'])
# Check that the timestamps in the file are unique
time = ds['time']
len_time = time.__len__()
len_time_unique = np.unique(time).__len__()
if len_time == len_time_unique:
time_test = 'pass'
else:
time_test = 'fail'
# Check that the timestamps in the file are in ascending order
if time_ascending != 'not_tested':
# convert time to number
time_in = [
dt.datetime.utcfromtimestamp(
np.datetime64(x).astype('O') / 1e9)
for x in ds['time'].values
]
time_data = nc.date2num(
time_in, 'seconds since 1900-01-01')
# Create a list of True or False by iterating through the array of time and checking
# if every time stamp is increasing
result = [(time_data[k + 1] - time_data[k]) > 0
for k in range(len(time_data) - 1)]
# Print outcome of the iteration with the list of indices when time is not increasing
if result.count(True) == len(time) - 1:
time_ascending = 'pass'
else:
ind_fail = {
k: time_in[k]
for k, v in enumerate(result) if v is False
}
time_ascending = 'fail: {}'.format(ind_fail)
# Count the number of days for which there is at least 1 timestamp
n_days = len(
np.unique(time.values.astype('datetime64[D]')))
# Compare variables in file to variables in Data Review Database
ds_variables = list(ds.data_vars.keys()) + list(
ds.coords.keys())
#ds_variables = [k for k in ds]
ds_variables = eliminate_common_variables(ds_variables)
ds_variables = [
x for x in ds_variables if 'qc' not in x
]
[_, unmatch1] = compare_lists(stream_vars,
ds_variables)
[_, unmatch2] = compare_lists(ds_variables,
stream_vars)
# Check deployment pressure from asset management against pressure variable in file
press = pf.pressure_var(ds, list(ds.coords.keys()))
if press is None:
press = pf.pressure_var(ds,
list(ds.data_vars.keys()))
# calculate mean pressure from data, excluding outliers +/- 3 SD
try:
pressure = ds[press]
num_dims = len(pressure.dims)
if len(pressure) > 1:
# if the pressure variable is an array of all zeros (as in the case of pressure_depth
# for OPTAAs on surface piercing profilers
if (len(np.unique(pressure)) == 1) & (
np.unique(pressure)[0] == 0.0):
try:
pressure = ds['int_ctd_pressure']
press = 'int_ctd_pressure'
except KeyError:
pressure = pressure
# reject NaNs
p_nonan = pressure.values[~np.isnan(pressure.
values)]
# reject fill values
p_nonan_nofv = p_nonan[
p_nonan != pressure._FillValue]
# reject data outside of global ranges
[pg_min,
pg_max] = cf.get_global_ranges(r, press)
if pg_min is not None and pg_max is not None:
pgr_ind = cf.reject_global_ranges(
p_nonan_nofv, pg_min, pg_max)
p_nonan_nofv_gr = p_nonan_nofv[pgr_ind]
else:
p_nonan_nofv_gr = p_nonan_nofv
if (len(p_nonan_nofv_gr) > 0) and (num_dims
== 1):
[
press_outliers, pressure_mean, _,
pressure_max, _, _
] = cf.variable_statistics(
p_nonan_nofv_gr, 3)
pressure_mean = round(pressure_mean, 2)
pressure_max = round(pressure_max, 2)
elif (len(p_nonan_nofv_gr) > 0) and (num_dims >
1):
print('variable has more than 1 dimension')
press_outliers = 'not calculated: variable has more than 1 dimension'
pressure_mean = round(
np.nanmean(p_nonan_nofv_gr), 2)
pressure_max = round(
np.nanmax(p_nonan_nofv_gr), 2)
else:
press_outliers = None
pressure_mean = None
pressure_max = None
if len(pressure) > 0 and len(p_nonan) == 0:
notes.append(
'Pressure variable all NaNs')
elif len(pressure) > 0 and len(
p_nonan) > 0 and len(
p_nonan_nofv) == 0:
notes.append(
'Pressure variable all fill values'
)
elif len(pressure) > 0 and len(
p_nonan) > 0 and len(
p_nonan_nofv) > 0 and len(
p_nonan_nofv_gr) == 0:
notes.append(
'Pressure variable outside of global ranges'
)
else: # if there is only 1 data point
press_outliers = 0
pressure_mean = round(
ds[press].values.tolist()[0], 2)
pressure_max = round(
ds[press].values.tolist()[0], 2)
try:
pressure_units = pressure.units
except AttributeError:
pressure_units = 'no units attribute for pressure'
if pressure_mean:
if ('WFP' in node) or ('MOAS' in subsite) or (
'SP' in node):
pressure_compare = int(round(pressure_max))
else:
pressure_compare = int(
round(pressure_mean))
if pressure_units == '0.001 dbar':
pressure_max = round((pressure_max / 1000),
2)
pressure_mean = round(
(pressure_mean / 1000), 2)
pressure_compare = round(
(pressure_compare / 1000), 2)
notes.append(
'Pressure converted from 0.001 dbar to dbar for pressure comparison'
)
elif pressure_units == 'daPa':
pressure_max = round((pressure_max / 1000),
2)
pressure_mean = round(
(pressure_mean / 1000), 2)
pressure_compare = round(
(pressure_compare / 1000), 2)
notes.append(
'Pressure converted from daPa to dbar for pressure comparison'
)
else:
pressure_compare = None
if (not deploy_depth) or (not pressure_mean):
pressure_diff = None
else:
pressure_diff = pressure_compare - deploy_depth
except KeyError:
press = 'no seawater pressure in file'
pressure_diff = None
pressure_mean = None
pressure_max = None
pressure_compare = None
press_outliers = None
pressure_units = None
# Add files and info to dictionary
filenames = data['deployments'][deployment]['method'][
method]['stream'][data_stream]['file'].keys()
if fname not in filenames:
data['deployments'][deployment]['method'][method][
'stream'][data_stream]['file'][
fname] = OrderedDict(
file_downloaded=pd.to_datetime(
splitter[0][0:15]).strftime(
'%Y-%m-%dT%H:%M:%S'),
file_coordinates=list(
ds.coords.keys()),
sampling_rate_seconds=sampling_rt_sec,
sampling_rate_details=rates,
data_start=data_start,
data_stop=data_stop,
time_gaps=gap_list,
unique_timestamps=time_test,
n_timestamps=len_time,
n_days=n_days,
notes=notes,
ascending_timestamps=time_ascending,
pressure_comparison=dict(
pressure_mean=pressure_mean,
units=pressure_units,
num_outliers=press_outliers,
diff=pressure_diff,
pressure_max=pressure_max,
variable=press,
pressure_compare=pressure_compare),
vars_in_file=ds_variables,
vars_not_in_file=[
x for x in unmatch1
if 'time' not in x
],
vars_not_in_db=unmatch2,
sci_var_stats=OrderedDict())
# calculate statistics for science variables, excluding outliers +/- 5 SD
for sv in sci_vars:
if sv != 't_max': # for ADCP
if sv != 'wavss_a_buoymotion_time':
print(sv)
try:
var = ds[sv]
# need to round SPKIR values to 1 decimal place to match the global ranges.
# otherwise, values that round to zero (e.g. 1.55294e-05) will be excluded by
# the global range test
# if 'spkir' in sv:
# vD = np.round(var.values, 1)
# else:
# vD = var.values
vD = var.values
if 'timedelta' not in str(
var.values.dtype):
# for OPTAA wavelengths: when multiple files are opened with xr.open_mfdataset
# xarray automatically forces all variables to have the same number of
# dimensions. So in this case wavelength_a and wavelength_c have 1 dimension
# in the individual files, so I'm forcing the analysis to treat them like
# they have 1 dimension (when there are multiple files for 1 deployment)
if sv == 'wavelength_a' or sv == 'wavelength_c':
[g_min,
g_max] = cf.get_global_ranges(
r, sv)
vnum_dims = len(var.dims)
if vnum_dims == 1:
n_all = len(var)
mean = list(vD)
else:
vnum_dims = 1
n_all = len(vD[0])
mean = list(vD[0])
num_outliers = None
vmin = None
vmax = None
sd = None
n_stats = 'not calculated'
var_units = var.units
n_nan = None
n_fv = None
n_grange = 'no global ranges'
fv = var._FillValue
else:
vnum_dims = len(var.dims)
if vnum_dims > 2:
print(
'variable has more than 2 dimensions'
)
num_outliers = None
mean = None
vmin = None
vmax = None
sd = None
n_stats = 'variable has more than 2 dimensions'
var_units = var.units
n_nan = None
n_fv = None
n_grange = None
fv = None
n_all = None
else:
if vnum_dims > 1:
n_all = [
len(vD),
len(vD.flatten())
]
else:
n_all = len(vD)
n_nan = int(
np.sum(np.isnan(vD)))
fv = var._FillValue
var_nofv = var.where(
var != fv)
n_fv = int(
np.sum(
np.isnan(
var_nofv.values
))) - n_nan
try:
var_units = var.units
except AttributeError:
var_units = 'no_units'
[g_min, g_max
] = cf.get_global_ranges(
r, sv)
if list(
np.unique(
np.isnan(
var_nofv))
) != [True]:
# reject data outside of global ranges
if g_min is not None and g_max is not None:
var_gr = var_nofv.where(
(var_nofv >=
g_min)
& (var_nofv <=
g_max))
n_grange = int(
np.sum(
np.isnan(
var_gr)
) - n_fv -
n_nan)
else:
n_grange = 'no global ranges'
var_gr = var_nofv
if list(
np.unique(
np.isnan(
var_gr)
)) != [True]:
if sv == 'spkir_abj_cspp_downwelling_vector':
# don't remove outliers from dataset
[
num_outliers,
mean, vmin,
vmax, sd,
n_stats
] = cf.variable_statistics_spkir(
var_gr)
else:
if vnum_dims > 1:
var_gr = var_gr.values.flatten(
)
# drop nans before calculating stats
var_gr = var_gr[
~np.isnan(
var_gr
)]
[
num_outliers,
mean, vmin,
vmax, sd,
n_stats
] = cf.variable_statistics(
var_gr, 5)
else:
num_outliers = None
mean = None
vmin = None
vmax = None
sd = None
n_stats = 0
n_grange = None
else:
num_outliers = None
mean = None
vmin = None
vmax = None
sd = None
n_stats = 0
n_grange = None
except KeyError:
if sv == 'int_ctd_pressure':
continue
else:
num_outliers = None
mean = None
vmin = None
vmax = None
sd = None
n_stats = 'variable not found in file'
var_units = None
n_nan = None
n_fv = None
fv = None
n_grange = None
n_all = None
if vnum_dims > 1:
sv = '{} (dims: {})'.format(
sv, list(var.dims))
else:
sv = sv
if 'timedelta' not in str(
var.values.dtype):
data['deployments'][deployment][
'method'][method]['stream'][
data_stream]['file'][fname][
'sci_var_stats'][sv] = dict(
n_outliers=num_outliers,
mean=mean,
min=vmin,
max=vmax,
stdev=sd,
n_stats=n_stats,
units=var_units,
n_nans=n_nan,
n_fillvalues=n_fv,
fill_value=str(fv),
global_ranges=[
g_min, g_max
],
n_grange=n_grange,
n_all=n_all)
sfile = os.path.join(save_dir, '{}-file_analysis.json'.format(r))
with open(sfile, 'w') as outfile:
json.dump(data, outfile)
depfile = os.path.join(save_dir, '{}-dependencies.txt'.format(r))
with open(depfile, 'w') as depf:
depf.write(str(dependencies))
json_file_list.append(str(sfile))
return json_file_list
def main(sDir, url_list, deployment_num):
reviewlist = pd.read_csv(
'https://raw.githubusercontent.com/ooi-data-lab/data-review-prep/master/review_list/data_review_list.csv')
rd_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
if rd not in rd_list:
rd_list.append(rd)
json_file_list = []
for r in rd_list:
dependencies = []
print('\n{}'.format(r))
data = OrderedDict(deployments=OrderedDict())
save_dir = os.path.join(sDir, r.split('-')[0], r)
cf.create_dir(save_dir)
# Deployment location test
deploy_loc_test = cf.deploy_location_check(r)
data['location_comparison'] = deploy_loc_test
for u in url_list:
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join((splitter[1], splitter[2], splitter[3], splitter[4]))
catalog_rms = '-'.join((r, splitter[-2], splitter[-1]))
# complete the analysis by reference designator
if rd_check == r:
udatasets = cf.get_nc_urls([u])
# check for the OOI 1.0 datasets for review
rl_filtered = reviewlist.loc[
(reviewlist['Reference Designator'] == r) & (reviewlist['status'] == 'for review')]
review_deployments = rl_filtered['deploymentNumber'].tolist()
review_deployments_int = ['deployment%04d' % int(x) for x in review_deployments]
for rev_dep in review_deployments_int:
if deployment_num is not None:
if int(rev_dep[-4:]) is not deployment_num:
print('\nskipping {}'.format(rev_dep))
continue
rdatasets = [s for s in udatasets if rev_dep in s]
rdatasets.sort()
if len(rdatasets) > 0:
datasets = []
for dss in rdatasets: # filter out collocated data files
if catalog_rms == dss.split('/')[-1].split('_20')[0][15:]:
datasets.append(dss)
else:
drd = dss.split('/')[-1].split('_20')[0][15:42]
if drd not in dependencies and drd != r:
dependencies.append(drd)
notes = []
time_ascending = ''
sci_vars_dict = {}
#datasets = datasets[0:2] #### for testing
for i in range(len(datasets)):
ds = xr.open_dataset(datasets[i], mask_and_scale=False)
ds = ds.swap_dims({'obs': 'time'})
print('\nAppending data from {}: file {} of {}'.format(rev_dep, i+1, len(datasets)))
# when opening multiple datasets, don't check that the timestamps are in ascending order
time_ascending = 'not_tested'
if i == 0:
fname, subsite, refdes, method, data_stream, deployment = cf.nc_attributes(datasets[0])
fname = fname.split('_20')[0]
# Get info from the data review database
dr_data = cf.refdes_datareview_json(refdes)
stream_vars = cf.return_stream_vars(data_stream)
sci_vars = cf.return_science_vars(data_stream)
node = refdes.split('-')[1]
if 'cspp' in data_stream or 'WFP' in node:
sci_vars.append('int_ctd_pressure')
# Add pressure to the list of science variables
press = pf.pressure_var(ds, list(ds.coords.keys()))
if press is None:
press = pf.pressure_var(ds, list(ds.data_vars.keys()))
if press is not None:
sci_vars.append(press)
sci_vars.append('time')
sci_vars = list(np.unique(sci_vars))
if 'ADCP' in r:
sci_vars = [x for x in sci_vars if 'beam' not in x]
for sci_var in sci_vars:
if sci_var == 'time':
sci_vars_dict.update(
{sci_var: dict(values=np.array([], dtype=np.datetime64), units=[], fv=[])})
else:
sci_vars_dict.update({sci_var: dict(values=np.array([]), units=[], fv=[])})
deploy_info = get_deployment_information(dr_data, int(deployment[-4:]))
# Grab deployment Variables
deploy_start = str(deploy_info['start_date'])
deploy_stop = str(deploy_info['stop_date'])
deploy_lon = deploy_info['longitude']
deploy_lat = deploy_info['latitude']
deploy_depth = deploy_info['deployment_depth']
# Calculate days deployed
if deploy_stop != 'None':
r_deploy_start = pd.to_datetime(deploy_start).replace(hour=0, minute=0, second=0)
if deploy_stop.split('T')[1] == '00:00:00':
r_deploy_stop = pd.to_datetime(deploy_stop)
else:
r_deploy_stop = (pd.to_datetime(deploy_stop) + timedelta(days=1)).replace(hour=0, minute=0, second=0)
n_days_deployed = (r_deploy_stop - r_deploy_start).days
else:
n_days_deployed = None
# Add reference designator to dictionary
try:
data['refdes']
except KeyError:
data['refdes'] = refdes
# append data for the deployment into a dictionary
for s_v in sci_vars_dict.keys():
vv = ds[s_v]
try:
if vv.units not in sci_vars_dict[s_v]['units']:
sci_vars_dict[s_v]['units'].append(vv.units)
except AttributeError:
print('')
try:
if vv._FillValue not in sci_vars_dict[s_v]['fv']:
sci_vars_dict[s_v]['fv'].append(vv._FillValue)
except AttributeError:
print('')
if len(vv.dims) == 1:
if s_v in ['wavelength_a', 'wavelength_c']:
# if the array is not same as the array that was already appended for these
# two OPTAA variables, append. if it's already there, don't append
if np.sum(vv.values == sci_vars_dict[s_v]['values']) != len(vv.values):
sci_vars_dict[s_v]['values'] = np.append(sci_vars_dict[s_v]['values'],
vv.values)
else:
sci_vars_dict[s_v]['values'] = np.append(sci_vars_dict[s_v]['values'], vv.values)
elif len(vv.dims) == 2: # appending 2D datasets
vD = vv.values.T
if len(sci_vars_dict[s_v]['values']) == 0:
sci_vars_dict[s_v]['values'] = vD
else:
sci_vars_dict[s_v]['values'] = np.concatenate((sci_vars_dict[s_v]['values'], vD), axis=1)
deployments = data['deployments'].keys()
data_start = pd.to_datetime(min(sci_vars_dict['time']['values'])).strftime('%Y-%m-%dT%H:%M:%S')
data_stop = pd.to_datetime(max(sci_vars_dict['time']['values'])).strftime('%Y-%m-%dT%H:%M:%S')
# Add deployment and info to dictionary and initialize delivery method sub-dictionary
if deployment not in deployments:
data['deployments'][deployment] = OrderedDict(deploy_start=deploy_start,
deploy_stop=deploy_stop,
n_days_deployed=n_days_deployed,
lon=deploy_lon,
lat=deploy_lat,
deploy_depth=deploy_depth,
method=OrderedDict())
# Add delivery methods to dictionary and initialize stream sub-dictionary
methods = data['deployments'][deployment]['method'].keys()
if method not in methods:
data['deployments'][deployment]['method'][method] = OrderedDict(
stream=OrderedDict())
# Add streams to dictionary and initialize file sub-dictionary
streams = data['deployments'][deployment]['method'][method]['stream'].keys()
if data_stream not in streams:
data['deployments'][deployment]['method'][method]['stream'][
data_stream] = OrderedDict(file=OrderedDict())
# Get a list of data gaps >1 day
time_df = pd.DataFrame(sci_vars_dict['time']['values'], columns=['time'])
time_df = time_df.sort_values(by=['time'])
gap_list = cf.timestamp_gap_test(time_df)
# Calculate the sampling rate to the nearest second
time_df['diff'] = time_df['time'].diff().astype('timedelta64[s]')
rates_df = time_df.groupby(['diff']).agg(['count'])
n_diff_calc = len(time_df) - 1
rates = dict(n_unique_rates=len(rates_df), common_sampling_rates=dict())
for i, row in rates_df.iterrows():
percent = (float(row['time']['count']) / float(n_diff_calc))
if percent > 0.1:
rates['common_sampling_rates'].update({int(i): '{:.2%}'.format(percent)})
sampling_rt_sec = None
for k, v in rates['common_sampling_rates'].items():
if float(v.strip('%')) > 50.00:
sampling_rt_sec = k
if not sampling_rt_sec:
sampling_rt_sec = 'no consistent sampling rate: {}'.format(rates['common_sampling_rates'])
# Don't do : Check that the timestamps in the file are unique
time_test = ''
# Count the number of days for which there is at least 1 timestamp
n_days = len(np.unique(sci_vars_dict['time']['values'].astype('datetime64[D]')))
# Compare variables in file to variables in Data Review Database
ds_variables = list(ds.data_vars.keys()) + list(ds.coords.keys())
ds_variables = eliminate_common_variables(ds_variables)
ds_variables = [x for x in ds_variables if 'qc' not in x]
[_, unmatch1] = compare_lists(stream_vars, ds_variables)
[_, unmatch2] = compare_lists(ds_variables, stream_vars)
# calculate mean pressure from data, excluding outliers +/- 3 SD
try:
pressure = sci_vars_dict[press]
if len(pressure) > 1:
# reject NaNs
p_nonan = pressure['values'][~np.isnan(pressure['values'])]
# reject fill values
p_nonan_nofv = p_nonan[p_nonan != pressure['fv'][0]]
# reject data outside of global ranges
[pg_min, pg_max] = cf.get_global_ranges(r, press)
if pg_min is not None and pg_max is not None:
pgr_ind = cf.reject_global_ranges(p_nonan_nofv, pg_min, pg_max)
p_nonan_nofv_gr = p_nonan_nofv[pgr_ind]
else:
p_nonan_nofv_gr = p_nonan_nofv
if (len(p_nonan_nofv_gr) > 0):
[press_outliers, pressure_mean, _, pressure_max, _, _] = cf.variable_statistics(p_nonan_nofv_gr, 3)
pressure_mean = round(pressure_mean, 2)
pressure_max = round(pressure_max, 2)
else:
press_outliers = None
pressure_mean = None
pressure_max = None
if len(pressure) > 0 and len(p_nonan) == 0:
notes.append('Pressure variable all NaNs')
elif len(pressure) > 0 and len(p_nonan) > 0 and len(p_nonan_nofv) == 0:
notes.append('Pressure variable all fill values')
elif len(pressure) > 0 and len(p_nonan) > 0 and len(p_nonan_nofv) > 0 and len(p_nonan_nofv_gr) == 0:
notes.append('Pressure variable outside of global ranges')
else: # if there is only 1 data point
press_outliers = 0
pressure_mean = round(ds[press].values.tolist()[0], 2)
pressure_max = round(ds[press].values.tolist()[0], 2)
try:
pressure_units = pressure['units'][0]
except AttributeError:
pressure_units = 'no units attribute for pressure'
if pressure_mean:
if 'SF' in node:
pressure_compare = int(round(pressure_max))
else:
pressure_compare = int(round(pressure_mean))
if pressure_units == '0.001 dbar':
pressure_max = round((pressure_max / 1000), 2)
pressure_mean = round((pressure_mean / 1000), 2)
pressure_compare = round((pressure_compare / 1000), 2)
notes.append('Pressure converted from 0.001 dbar to dbar for pressure comparison')
elif pressure_units == 'daPa':
pressure_max = round((pressure_max / 1000), 2)
pressure_mean = round((pressure_mean / 1000), 2)
pressure_compare = round((pressure_compare / 1000), 2)
notes.append('Pressure converted from daPa to dbar for pressure comparison')
else:
pressure_compare = None
if (not deploy_depth) or (not pressure_mean):
pressure_diff = None
else:
pressure_diff = pressure_compare - deploy_depth
except KeyError:
press = 'no seawater pressure in file'
pressure_diff = None
pressure_mean = None
pressure_max = None
pressure_compare = None
press_outliers = None
pressure_units = None
# Add files and info to dictionary
filenames = data['deployments'][deployment]['method'][method]['stream'][data_stream][
'file'].keys()
if fname not in filenames:
data['deployments'][deployment]['method'][method]['stream'][data_stream]['file'][
fname] = OrderedDict(
file_downloaded=pd.to_datetime(splitter[0][0:15]).strftime('%Y-%m-%dT%H:%M:%S'),
file_coordinates=list(ds.coords.keys()),
sampling_rate_seconds=sampling_rt_sec,
sampling_rate_details=rates,
data_start=data_start,
data_stop=data_stop,
time_gaps=gap_list,
unique_timestamps=time_test,
n_timestamps=len(sci_vars_dict['time']['values']),
n_days=n_days,
notes=notes,
ascending_timestamps=time_ascending,
pressure_comparison=dict(pressure_mean=pressure_mean, units=pressure_units,
num_outliers=press_outliers, diff=pressure_diff,
pressure_max=pressure_max, variable=press,
pressure_compare=pressure_compare),
vars_in_file=ds_variables,
vars_not_in_file=[x for x in unmatch1 if 'time' not in x],
vars_not_in_db=unmatch2,
sci_var_stats=OrderedDict())
# calculate statistics for science variables, excluding outliers +/- 5 SD
for sv in sci_vars_dict.keys():
if sv != 't_max': # for ADCP
if sv != 'time':
print(sv)
var = sci_vars_dict[sv]
vD = var['values']
var_units = var['units']
#if 'timedelta' not in str(vD.dtype):
vnum_dims = len(np.shape(vD))
# for OPTAA wavelengths, print the array
if sv == 'wavelength_a' or sv == 'wavelength_c':
[g_min, g_max] = cf.get_global_ranges(r, sv)
n_all = len(var)
mean = list(vD)
num_outliers = None
vmin = None
vmax = None
sd = None
n_stats = 'not calculated'
n_nan = None
n_fv = None
n_grange = 'no global ranges'
fv = var['fv'][0]
else:
if vnum_dims > 2:
print('variable has more than 2 dimensions')
num_outliers = None
mean = None
vmin = None
vmax = None
sd = None
n_stats = 'variable has more than 2 dimensions'
n_nan = None
n_fv = None
n_grange = None
fv = None
n_all = None
else:
if vnum_dims > 1:
n_all = [len(vD), len(vD.flatten())]
else:
n_all = len(vD)
n_nan = int(np.sum(np.isnan(vD)))
fv = var['fv'][0]
vD[vD == fv] = np.nan # turn fill values to nans
n_fv = int(np.sum(np.isnan(vD))) - n_nan
[g_min, g_max] = cf.get_global_ranges(r, sv)
if list(np.unique(np.isnan(vD))) != [True]:
# reject data outside of global ranges
if g_min is not None and g_max is not None:
# turn data outside of global ranges to nans
#var_gr = var_nofv.where((var_nofv >= g_min) & (var_nofv <= g_max))
vD[vD < g_min] = np.nan
vD[vD > g_max] = np.nan
n_grange = int(np.sum(np.isnan(vD)) - n_fv - n_nan)
else:
n_grange = 'no global ranges'
if list(np.unique(np.isnan(vD))) != [True]:
if sv == 'spkir_abj_cspp_downwelling_vector':
# don't remove outliers from dataset
[num_outliers, mean, vmin, vmax, sd, n_stats] = cf.variable_statistics_spkir(vD)
else:
if vnum_dims > 1:
var_gr = vD.flatten()
else:
var_gr = vD
# drop nans before calculating stats
var_gr = var_gr[~np.isnan(var_gr)]
[num_outliers, mean, vmin, vmax, sd, n_stats] = cf.variable_statistics(var_gr, 5)
else:
num_outliers = None
mean = None
vmin = None
vmax = None
sd = None
n_stats = 0
n_grange = None
else:
num_outliers = None
mean = None
vmin = None
vmax = None
sd = None
n_stats = 0
n_grange = None
if vnum_dims > 1:
sv = '{} (dims: {})'.format(sv, list(np.shape(var['values'])))
else:
sv = sv
#if 'timedelta' not in str(var.values.dtype):
data['deployments'][deployment]['method'][method]['stream'][data_stream]['file'][
fname]['sci_var_stats'][sv] = dict(n_outliers=num_outliers, mean=mean, min=vmin,
max=vmax, stdev=sd, n_stats=n_stats, units=var_units,
n_nans=n_nan, n_fillvalues=n_fv, fill_value=str(fv),
global_ranges=[g_min, g_max], n_grange=n_grange,
n_all=n_all)
sfile = os.path.join(save_dir, '{}-{}-file_analysis.json'.format(rev_dep, r))
with open(sfile, 'w') as outfile:
json.dump(data, outfile)
json_file_list.append(str(sfile))
depfile = os.path.join(save_dir, '{}-dependencies.txt'.format(r))
with open(depfile, 'w') as depf:
depf.write(str(dependencies))
return json_file_list
def main(url_list, sDir, plot_type, deployment_num, start_time, end_time, method_num, zdbar, n_std, inpercentile, zcell_size):
for i, u in enumerate(url_list):
print('\nUrl {} of {}: {}'.format(i + 1, len(url_list), u))
elements = u.split('/')[-2].split('-')
r = '-'.join((elements[1], elements[2], elements[3], elements[4]))
ms = u.split(r + '-')[1].split('/')[0]
subsite = r.split('-')[0]
array = subsite[0:2]
main_sensor = r.split('-')[-1]
# read URL to get data
datasets = cf.get_nc_urls([u])
datasets_sel = cf.filter_collocated_instruments(main_sensor, datasets)
# get sci data review list
dr_data = cf.refdes_datareview_json(r)
ps_df, n_streams = cf.get_preferred_stream_info(r)
# get end times of deployments
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = cf.get_deployment_information(dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
# create a dictionary for science variables from analysis file
stream_sci_vars_dict = dict()
for x in dr_data['instrument']['data_streams']:
dr_ms = '-'.join((x['method'], x['stream_name']))
if ms == dr_ms:
stream_sci_vars_dict[dr_ms] = dict(vars=dict())
sci_vars = dict()
for y in x['stream']['parameters']:
if y['data_product_type'] == 'Science Data':
sci_vars.update({y['name']: dict(db_units=y['unit'])})
if len(sci_vars) > 0:
stream_sci_vars_dict[dr_ms]['vars'] = sci_vars
for ii, d in enumerate(datasets_sel):
part_d = d.split('/')[-1]
print('\nDataset {} of {}: {}'.format(ii + 1, len(datasets_sel), part_d))
with xr.open_dataset(d, mask_and_scale=False) as ds:
ds = ds.swap_dims({'obs': 'time'})
fname, subsite, refdes, method, stream, deployment = cf.nc_attributes(d)
if method_num is not None:
if method != method_num:
print(method_num, method)
continue
if deployment_num is not None:
if int(deployment.split('0')[-1]) is not deployment_num:
print(type(int(deployment.split('0')[-1])), type(deployment_num))
continue
if start_time is not None and end_time is not None:
ds = ds.sel(time=slice(start_time, end_time))
if len(ds['time'].values) == 0:
print('No data to plot for specified time range: ({} to {})'.format(start_time, end_time))
continue
stime = start_time.strftime('%Y-%m-%d')
etime = end_time.strftime('%Y-%m-%d')
ext = stime + 'to' + etime # .join((ds0_method, ds1_method
save_dir = os.path.join(sDir, array, subsite, refdes, plot_type, ms.split('-')[0], deployment, ext)
else:
save_dir = os.path.join(sDir, array, subsite, refdes, plot_type, ms.split('-')[0], deployment)
cf.create_dir(save_dir)
texclude_dir = os.path.join(sDir, array, subsite, refdes, 'time_to_exclude')
cf.create_dir(texclude_dir)
# initialize an empty data array for science variables in dictionary
sci_vars_dict = cd.initialize_empty_arrays(stream_sci_vars_dict, ms)
for var in list(sci_vars_dict[ms]['vars'].keys()):
sh = sci_vars_dict[ms]['vars'][var]
if ds[var].units == sh['db_units']:
if ds[var]._FillValue not in sh['fv']:
sh['fv'].append(ds[var]._FillValue)
if ds[var].units not in sh['units']:
sh['units'].append(ds[var].units)
sh['t'] = np.append(sh['t'], ds['time'].values) # t = ds['time'].values
sh['values'] = np.append(sh['values'], ds[var].values) # z = ds[var].values
y, y_unit, y_name = cf.add_pressure_to_dictionary_of_sci_vars(ds)
sh['pressure'] = np.append(sh['pressure'], y)
stat_data = pd.DataFrame(columns=['deployments', 'time_to_exclude'])
file_exclude = '{}/{}_{}_{}_excluded_timestamps.csv'.format(texclude_dir,
deployment, refdes, method)
stat_data.to_csv(file_exclude, index=True)
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print(sv)
if len(vinfo['t']) < 1:
print('no variable data to plot')
else:
sv_units = vinfo['units'][0]
fv = vinfo['fv'][0]
t0 = pd.to_datetime(min(vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
colors = cm.rainbow(np.linspace(0, 1, len(vinfo['t'])))
t = vinfo['t']
z = vinfo['values']
y = vinfo['pressure']
# Check if the array is all NaNs
if sum(np.isnan(z)) == len(z):
print('Array of all NaNs - skipping plot.')
continue
# Check if the array is all fill values
elif len(z[z != fv]) == 0:
print('Array of all fill values - skipping plot.')
continue
else:
# reject erroneous data
dtime, zpressure, ndata, lenfv, lennan, lenev, lengr, global_min, global_max = \
cf.reject_erroneous_data(r, sv, t, y, z, fv)
# create data groups
columns = ['tsec', 'dbar', str(sv)]
min_r = int(round(min(zpressure) - zcell_size))
max_r = int(round(max(zpressure) + zcell_size))
ranges = list(range(min_r, max_r, zcell_size))
groups, d_groups = gt.group_by_depth_range(dtime, zpressure, ndata, columns, ranges)
# ... excluding timestamps
if 'scatter' in sv:
n_std = None #to use percentile
else:
n_std = n_std
# rejecting timestamps from percentile analysis
y_avg, n_avg, n_min, n_max, n0_std, n1_std, l_arr, time_ex, \
t_nospct, z_nospct, y_nospct = cf.reject_timestamps_in_groups(groups, d_groups, n_std,
dtime, zpressure, ndata,
inpercentile)
print('{} using {} percentile of data grouped in {} dbar segments'.format(
len(zpressure) - len(z_nospct), inpercentile, zcell_size))
"""
writing timestamps to .csv file to use with data_range.py script
"""
if len(time_ex) != 0:
t_exclude = time_ex[0]
for i in range(len(time_ex))[1:len(time_ex)]:
t_exclude = '{}, {}'.format(t_exclude, time_ex[i])
stat_data = pd.DataFrame({'deployments': deployment,
'time_to_exclude': t_exclude}, index=[sv])
stat_data.to_csv(file_exclude, index=True, mode='a', header=False)
# reject time range from data portal file export
t_portal, z_portal, y_portal = cf.reject_timestamps_dataportal(subsite, r,
t_nospct, z_nospct, y_nospct)
print('{} using visual inspection of data'.format(len(z_nospct) - len(z_portal),
inpercentile, zcell_size))
# reject data in a depth range
if zdbar is not None:
y_ind = y_portal < zdbar
t_array = t_portal[y_ind]
y_array = y_portal[y_ind]
z_array = z_portal[y_ind]
else:
y_ind = []
t_array = t_portal
y_array = y_portal
z_array = z_portal
print('{} in water depth > {} dbar'.format(len(y_ind), zdbar))
"""
Plot data
"""
if len(t_array) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((sv, r))
else:
sname = '-'.join((sv, r, m))
xlabel = sv + " (" + sv_units + ")"
ylabel = y_name[0] + " (" + y_unit[0] + ")"
clabel = 'Time'
title = ' '.join((deployment, r, m))
# plot non-erroneous data
fig, ax = pf.plot_profiles(ndata, zpressure, dtime,
ylabel, xlabel, clabel, end_times, deployments, stdev=None)
ax.set_title(title, fontsize=9)
ax.plot(n_avg, y_avg, '-k')
ax.fill_betweenx(y_avg, n0_std, n1_std, color='m', alpha=0.2)
leg_text = (
'removed {} fill values, {} NaNs, {} Extreme Values (1e7), {} Global ranges [{} - {}]'.format(
len(z) - lenfv, len(z) - lennan, len(z) - lenev, lengr, global_min, global_max) + '\n' +
('(black) data average in {} dbar segments'.format(zcell_size)) + '\n' +
('(magenta) upper and lower {} percentile envelope in {} dbar segments'.format(
inpercentile, zcell_size)),)
ax.legend(leg_text, loc='upper center', bbox_to_anchor=(0.5, -0.17), fontsize=6)
fig.tight_layout()
sfile = '_'.join(('rm_erroneous_data', sname))
pf.save_fig(save_dir, sfile)
# plot excluding time ranges for suspect data
if len(z_nospct) != len(zpressure):
fig, ax = pf.plot_profiles(z_nospct, y_nospct, t_nospct,
ylabel, xlabel, clabel, end_times, deployments, stdev=None)
ax.set_title(title, fontsize=9)
leg_text = (
'removed {} in the upper and lower {} percentile of data grouped in {} dbar segments'.format(
len(zpressure) - len(z_nospct), inpercentile, zcell_size),)
ax.legend(leg_text, loc='upper center', bbox_to_anchor=(0.5, -0.17), fontsize=6)
fig.tight_layout()
sfile = '_'.join(('rm_suspect_data', sname))
pf.save_fig(save_dir, sfile)
# plot excluding time ranges from data portal export
if len(z_nospct) - len(z_portal):
fig, ax = pf.plot_profiles(z_portal, y_portal, t_portal,
ylabel, xlabel, clabel, end_times, deployments, stdev=None)
ax.set_title(title, fontsize=9)
leg_text = ('excluded {} suspect data when inspected visually'.format(
len(z_nospct) - len(z_portal)),)
ax.legend(leg_text, loc='upper center', bbox_to_anchor=(0.5, -0.17), fontsize=6)
fig.tight_layout()
sfile = '_'.join(('rm_v_suspect_data', sname))
pf.save_fig(save_dir, sfile)
# Plot excluding a selected depth value
if len(z_array) != len(z_array):
fig, ax = pf.plot_profiles(z_array, y_array, t_array,
ylabel, xlabel, clabel, end_times, deployments, stdev=None)
ax.set_title(title, fontsize=9)
leg_text = ('excluded {} suspect data in water depth greater than {} dbar'.format(len(y_ind), zdbar),)
ax.legend(leg_text, loc='upper center', bbox_to_anchor=(0.5, -0.17), fontsize=6)
fig.tight_layout()
sfile = '_'.join(('rm_depth_range', sname))
pf.save_fig(save_dir, sfile)
def main(url_list, sDir, mDir, zcell_size, zdbar, start_time, end_time):
"""""
URL : path to instrument data by methods
sDir : path to the directory on your machine to save files
plot_type: folder name for a plot type
""" ""
rd_list = []
ms_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
ms = uu.split(rd + '-')[1].split('/')[0]
if rd not in rd_list:
rd_list.append(rd)
if ms not in ms_list:
ms_list.append(ms)
'''
separate different instruments
'''
for r in rd_list:
print('\n{}'.format(r))
subsite = r.split('-')[0]
array = subsite[0:2]
main_sensor = r.split('-')[-1]
# read in the analysis file
dr_data = cf.refdes_datareview_json(r)
# get the preferred stream information
ps_df, n_streams = cf.get_preferred_stream_info(r)
# get science variable long names from the Data Review Database
stream_sci_vars = cd.sci_var_long_names(r)
#stream_vars = cd.var_long_names(r)
# check if the science variable long names are the same for each stream and initialize empty arrays
sci_vars_dict0 = cd.sci_var_long_names_check(stream_sci_vars)
# get the list of data files and filter out collocated instruments and other streams
datasets = []
for u in url_list:
print(u)
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join(
(splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
fdatasets = cf.filter_collocated_instruments(main_sensor, datasets)
fdatasets = cf.filter_other_streams(r, ms_list, fdatasets)
# select the list of data files from the preferred dataset for each deployment
fdatasets_final = []
for ii in range(len(ps_df)):
for x in fdatasets:
if ps_df['deployment'][ii] in x and ps_df[0][ii] in x:
fdatasets_final.append(x)
# build dictionary of science data from the preferred dataset for each deployment
print('\nAppending data from files')
et = []
sci_vars_dict, y_unit, y_name = cd.append_evaluated_science_data(
sDir, ps_df, n_streams, r, fdatasets_final, sci_vars_dict0, et,
start_time, end_time)
# get end times of deployments
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = cf.get_deployment_information(
dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
"""
create a data-ranges table and figure for full data time range
"""
# create a folder to save data ranges
save_dir_stat = os.path.join(mDir, array, subsite)
cf.create_dir(save_dir_stat)
save_fdir = os.path.join(sDir, array, subsite, r, 'data_range')
cf.create_dir(save_fdir)
stat_df = pd.DataFrame()
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print(vinfo['var_name'])
if len(vinfo['t']) < 1:
print('no variable data to plot')
continue
else:
sv_units = vinfo['units'][0]
fv = vinfo['fv'][0]
t = vinfo['t']
z = vinfo['values']
y = vinfo['pressure']
# Check if the array is all NaNs
if sum(np.isnan(z)) == len(z):
print('Array of all NaNs - skipping plot.')
continue
# Check if the array is all fill values
elif len(z[z != fv]) == 0:
print('Array of all fill values - skipping plot.')
continue
else:
"""
clean up data
"""
# reject erroneous data
dtime, zpressure, ndata, lenfv, lennan, lenev, lengr, global_min, global_max = \
cf.reject_erroneous_data(r, sv, t, y, z, fv)
# reject timestamps from stat analysis
Dpath = '{}/{}/{}/{}/{}'.format(sDir, array, subsite, r,
'time_to_exclude')
onlyfiles = []
for item in os.listdir(Dpath):
if not item.startswith('.') and os.path.isfile(
os.path.join(Dpath, item)):
onlyfiles.append(join(Dpath, item))
dre = pd.DataFrame()
for nn in onlyfiles:
dr = pd.read_csv(nn)
dre = dre.append(dr, ignore_index=True)
drn = dre.loc[dre['Unnamed: 0'] == vinfo['var_name']]
list_time = []
for itime in drn.time_to_exclude:
ntime = itime.split(', ')
list_time.extend(ntime)
u_time_list = np.unique(list_time)
if len(u_time_list) != 0:
t_nospct, z_nospct, y_nospct = cf.reject_suspect_data(
dtime, zpressure, ndata, u_time_list)
print(
'{} using {} percentile of data grouped in {} dbar segments'
.format(
len(zpressure) - len(z_nospct), inpercentile,
zcell_size))
# reject time range from data portal file export
t_portal, z_portal, y_portal = cf.reject_timestamps_dataportal(
subsite, r, t_nospct, y_nospct, z_nospct)
print('{} using visual inspection of data'.format(
len(z_nospct) - len(z_portal), inpercentile,
zcell_size))
# reject data in a depth range
if zdbar is not None:
y_ind = y_portal < zdbar
t_array = t_portal[y_ind]
y_array = y_portal[y_ind]
z_array = z_portal[y_ind]
else:
y_ind = []
t_array = t_portal
y_array = y_portal
z_array = z_portal
print('{} in water depth > {} dbar'.format(
len(y_ind), zdbar))
if len(y_array) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((vinfo['var_name'], r))
else:
sname = '-'.join((vinfo['var_name'], r, m))
"""
create data ranges for non - pressure data only
"""
if 'pressure' in vinfo['var_name']:
pass
else:
columns = ['tsec', 'dbar', str(vinfo['var_name'])]
# create depth ranges
min_r = int(round(min(y_array) - zcell_size))
max_r = int(round(max(y_array) + zcell_size))
ranges = list(range(min_r, max_r, zcell_size))
# group data by depth
groups, d_groups = gt.group_by_depth_range(
t_array, y_array, z_array, columns, ranges)
print('writing data ranges for {}'.format(
vinfo['var_name']))
stat_data = groups.describe()[vinfo['var_name']]
stat_data.insert(loc=0,
column='parameter',
value=sv,
allow_duplicates=False)
t_deploy = deployments[0]
for i in range(
len(deployments))[1:len(deployments)]:
t_deploy = '{}, {}'.format(
t_deploy, deployments[i])
stat_data.insert(loc=1,
column='deployments',
value=t_deploy,
allow_duplicates=False)
stat_df = stat_df.append(stat_data, ignore_index=True)
"""
plot full time range free from errors and suspect data
"""
clabel = sv + " (" + sv_units + ")"
ylabel = (y_name[0][0] + " (" + y_unit[0][0] + ")")
title = ' '.join((r, m))
# plot non-erroneous -suspect data
fig, ax, bar = pf.plot_xsection(subsite,
t_array,
y_array,
z_array,
clabel,
ylabel,
inpercentile=None,
stdev=None)
ax.set_title(title, fontsize=9)
leg_text = (
'removed {} fill values, {} NaNs, {} Extreme Values (1e7), {} Global ranges [{} - {}]'
.format(
len(z) - lenfv,
len(z) - lennan,
len(z) - lenev, lengr, global_min,
global_max) + '\n' +
('removed {} in the upper and lower {} percentile of data grouped in {} dbar segments'
.format(
len(zpressure) - len(z_nospct), inpercentile,
zcell_size)), )
ax.legend(leg_text,
loc='upper center',
bbox_to_anchor=(0.5, -0.17),
fontsize=6)
for ii in range(len(end_times)):
ax.axvline(x=end_times[ii],
color='b',
linestyle='--',
linewidth=.8)
ax.text(end_times[ii],
min(y_array) - 5,
'End' + str(deployments[ii]),
fontsize=6,
style='italic',
bbox=dict(
boxstyle='round',
ec=(0., 0.5, 0.5),
fc=(1., 1., 1.),
))
fig.tight_layout()
sfile = '_'.join(('data_range', sname))
pf.save_fig(save_fdir, sfile)
# write stat file
stat_df.to_csv('{}/{}_data_ranges.csv'.format(save_dir_stat, r),
index=True,
float_format='%11.6f')
def main(sDir, url_list, start_time, end_time, deployment_num, interval):
rd_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
if rd not in rd_list:
rd_list.append(rd)
for r in rd_list:
print('\n{}'.format(r))
datasets = []
deployments = []
for u in url_list:
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join((splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
for ud in udatasets:
if ud.split('/')[-1].split('_')[0] not in deployments:
deployments.append(ud.split('/')[-1].split('_')[0])
datasets = list(itertools.chain(*datasets))
datasets = cf.filter_collocated_instruments(r, datasets)
deployments.sort()
fdatasets = np.unique(datasets).tolist()
for deploy in deployments:
if deployment_num is not None:
if int(deploy[-4:]) is not deployment_num:
print('\nskipping {}'.format(deploy))
continue
rdatasets = [s for s in fdatasets if deploy in s]
# break deployment into 4 segments or make a list of the time range specified
if start_time is not None and end_time is not None:
dt_range = [dt.datetime.strftime(start_time, '%Y-%m-%d'), dt.datetime.strftime(end_time, '%Y-%m-%d')]
else:
# Get deployment info from the data review database
dr_data = cf.refdes_datareview_json(r)
d_info = [x for x in dr_data['instrument']['deployments'] if x['deployment_number'] == int(deploy[-4:])]
d_info = d_info[0]
deploy_start = dt.datetime.strptime(str(d_info['start_date']).split('T')[0], '%Y-%m-%d')
deploy_stop = dt.datetime.strptime(str(d_info['stop_date']).split('T')[0], '%Y-%m-%d') + dt.timedelta(
days=1)
dt_range = list(date_range(deploy_start, deploy_stop, 4))
sci_vars_dict = {'time': dict(values=np.array([], dtype=np.datetime64), fv=[], ln=[]),
'bin_depths': dict(values=np.array([]), units=[], fv=[], ln=[])}
percentgood = {'percent_good_beam1': dict(values=np.array([])),
'percent_good_beam2': dict(values=np.array([])),
'percent_good_beam3': dict(values=np.array([])),
'percent_good_beam4': dict(values=np.array([]))}
if interval is None:
toplot = range(len(dt_range) - 1)
else:
toplot = [interval - 1]
for dtri in toplot:
stime = dt.datetime.strptime(dt_range[dtri], '%Y-%m-%d')
etime = dt.datetime.strptime(dt_range[dtri + 1], '%Y-%m-%d')
if len(rdatasets) > 0:
for i in range(len(rdatasets)):
#for i in range(0, 2): ##### for testing
ds = xr.open_dataset(rdatasets[i], mask_and_scale=False)
ds = ds.swap_dims({'obs': 'time'})
print('\nAppending data from {}: file {} of {}'.format(deploy, i + 1, len(rdatasets)))
ds = ds.sel(time=slice(stime, etime))
if len(ds['time'].values) == 0:
print('No data to plot for specified time range: ({} to {})'.format(start_time, end_time))
continue
try:
print(fname)
except NameError:
fname, subsite, refdes, method, stream, deployment = cf.nc_attributes(rdatasets[0])
array = subsite[0:2]
sci_vars = cf.return_science_vars(stream)
# drop the following list of key words from science variables list
sci_vars = notin_list(sci_vars, ['salinity', 'temperature', 'bin_depths', 'beam'])
sci_vars = [name for name in sci_vars if ds[name].units != 'mm s-1']
for sci_var in sci_vars:
sci_vars_dict.update({sci_var: dict(values=np.array([]), units=[], fv=[], ln=[])})
# append data for the deployment into a dictionary
for s_v, info in sci_vars_dict.items():
print(s_v)
vv = ds[s_v]
try:
if vv.units not in info['units']:
info['units'].append(vv.units)
except AttributeError:
print('no units')
try:
if vv._FillValue not in info['fv']:
info['fv'].append(vv._FillValue)
except AttributeError:
print('no fill value')
try:
if vv.long_name not in info['ln']:
info['ln'].append(vv.long_name)
except AttributeError:
print('no long name')
if len(vv.dims) == 1:
info['values'] = np.append(info['values'], vv.values)
else:
if len(info['values']) == 0:
info['values'] = vv.values.T
else:
info['values'] = np.concatenate((info['values'], vv.values.T), axis=1)
# append percent good beams
for j, k in percentgood.items():
pgvv = ds[j]
fv_pgvv = pgvv._FillValue
pgvv = pgvv.values.T.astype(float)
pgvv[pgvv == fv_pgvv] = np.nan
if len(k['values']) == 0:
k['values'] = pgvv
else:
k['values'] = np.concatenate((k['values'], pgvv), axis=1)
if len(sci_vars_dict['time']['values']) > 0:
filename = '_'.join(fname.split('_')[:-1])
save_dir = os.path.join(sDir, array, subsite, refdes, 'plots', deployment)
cf.create_dir(save_dir)
tm = sci_vars_dict['time']['values']
t0 = pd.to_datetime(tm.min()).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(tm.max()).strftime('%Y-%m-%dT%H:%M:%S')
title_text = ' '.join((deployment, refdes, method))
bd = sci_vars_dict['bin_depths']
ylabel = 'bin_depths ({})'.format(bd['units'][0])
print('\nPlotting interval {}'.format(int(dtri) + 1))
for var in sci_vars:
print('----{}'.format(var))
v = sci_vars_dict[var]
fv = v['fv'][0]
v_name = v['ln'][0]
units = v['units'][0]
if len(np.shape(v['values'])) == 1:
v, n_nan, n_fv, n_ev, n_grange, g_min, g_max, n_std = reject_err_data_1_dims(v['values'], fv, r, var, n=5)
if len(tm) > np.sum(np.isnan(v)): # only plot if the array contains values
# Plot all data
fig, ax = pf.plot_timeseries(tm, v, v_name, stdev=None)
ax.set_title((title_text + '\n' + t0 + ' - ' + t1), fontsize=9)
sfile = '-'.join((filename, v_name, t0[:10]))
pf.save_fig(save_dir, sfile)
# Plot data with outliers removed
fig, ax = pf.plot_timeseries(tm, v, v_name, stdev=5)
title_i = 'removed: {} nans, {} fill values, {} extreme values, {} GR [{}, {}],' \
' {} outliers +/- 5 SD'.format(n_nan, n_fv , n_ev, n_grange, g_min, g_max, n_std)
ax.set_title((title_text + '\n' + t0 + ' - ' + t1 + '\n' + title_i), fontsize=8)
sfile = '-'.join((filename, v_name, t0[:10])) + '_rmoutliers'
pf.save_fig(save_dir, sfile)
else:
print('Array of all nans - skipping plot')
else:
v, n_nan, n_fv, n_ev, n_bb, n_grange, g_min, g_max = reject_err_data_2_dims(v['values'], percentgood, fv, r, var)
clabel = '{} ({})'.format(var, units)
# check bin depths for extreme values
y = bd['values']
# if all the values are negative, take the absolute value (cabled data bin depths are negative)
if int(np.nanmin(y)) < 0 and int(np.nanmax(y)) < 0:
y = abs(y)
y_nan = np.sum(np.isnan(y))
y = np.where(y < 6000, y, np.nan) # replace extreme bin_depths by nans
bin_nan = np.sum(np.isnan(y)) - y_nan
bin_title = 'removed: {} bin depths > 6000'.format(bin_nan)
if 'echo' in var:
color = 'BuGn'
else:
color = 'RdBu'
new_y = dropna(y, axis=1) # convert to DataFrame to drop nan
y_mask = new_y.loc[list(new_y.index), list(new_y.columns)]
v_new = pd.DataFrame(v)
v_mask = v_new.loc[list(new_y.index), list(new_y.columns)]
tm_mask = tm[new_y.columns]
fig, ax, __ = pf.plot_adcp(tm_mask, np.array(y_mask), np.array(v_mask), ylabel, clabel, color,
n_stdev=None)
if bin_nan > 0:
ax.set_title((title_text + '\n' + t0 + ' - ' + t1 + '\n' + bin_title), fontsize=8)
else:
ax.set_title((title_text + '\n' + t0 + ' - ' + t1), fontsize=8)
sfile = '-'.join((filename, var, t0[:10]))
pf.save_fig(save_dir, sfile)
fig, ax, n_nans_all = pf.plot_adcp(tm_mask, np.array(y_mask), np.array(v_mask), ylabel, clabel, color, n_stdev=5)
title_i = 'removed: {} nans, {} fill values, {} extreme values, {} bad beams, {} GR [{}, {}]'.format(
n_nan, n_fv, n_ev, n_bb, n_grange, g_min, g_max)
if bin_nan > 0:
ax.set_title((title_text + '\n' + t0 + ' - ' + t1 + '\n' + title_i + '\n' + bin_title), fontsize=8)
else:
ax.set_title((title_text + '\n' + t0 + ' - ' + t1 + '\n' + title_i), fontsize=8)
sfile = '-'.join((filename, var, t0[:10])) + '_rmoutliers'
pf.save_fig(save_dir, sfile)
def main(sDir, url_list, start_time, end_time):
rd_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
if rd not in rd_list:
rd_list.append(rd)
for r in rd_list:
print('\n{}'.format(r))
datasets = []
for u in url_list:
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join((splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
fdatasets = []
# get the preferred stream information
ps_df, n_streams = cf.get_preferred_stream_info(r)
for index, row in ps_df.iterrows():
for ii in range(n_streams):
try:
rms = '-'.join((r, row[ii]))
except TypeError:
continue
for dd in datasets:
spl = dd.split('/')[-2].split('-')
catalog_rms = '-'.join((spl[1], spl[2], spl[3], spl[4], spl[5], spl[6]))
fdeploy = dd.split('/')[-1].split('_')[0]
if rms == catalog_rms and fdeploy == row['deployment']:
fdatasets.append(dd)
main_sensor = r.split('-')[-1]
fdatasets_sel = cf.filter_collocated_instruments(main_sensor, fdatasets)
# get science variable long names from the Data Review Database
#stream_sci_vars = cd.sci_var_long_names(r)
if 'SPKIR' in r or 'PRESF' in r: # only get the main science variable for SPKIR
stream_vars = cd.sci_var_long_names(r)
else:
stream_vars = var_long_names(r)
# check if the science variable long names are the same for each stream and initialize empty arrays
sci_vars_dict = cd.sci_var_long_names_check(stream_vars)
# get the preferred stream information
ps_df, n_streams = cf.get_preferred_stream_info(r)
# build dictionary of science data from the preferred dataset for each deployment
print('\nAppending data from files')
et = []
sci_vars_dict, __, __ = cd.append_science_data(ps_df, n_streams, r, fdatasets_sel, sci_vars_dict, et, start_time, end_time)
# get end times of deployments
dr_data = cf.refdes_datareview_json(r)
deployments = []
dend_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = get_deployment_information(dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
dend_times.append(pd.to_datetime(deploy_info['stop_date']))
subsite = r.split('-')[0]
array = subsite[0:2]
save_dir = os.path.join(sDir, array, subsite, r, 'timeseries_plots_preferred_all')
cf.create_dir(save_dir)
print('\nPlotting data')
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print(sv)
if 'SPKIR' in r:
fv_lst = np.unique(vinfo['fv']).tolist()
if len(fv_lst) == 1:
fill_value = fv_lst[0]
else:
print(fv_lst)
print('No unique fill value for {}'.format(sv))
sv_units = np.unique(vinfo['units']).tolist()
t = vinfo['t']
if len(t) > 1:
data = vinfo['values']
[dd_data, g_min, g_max] = index_dataset_2d(r, 'spkir_abj_cspp_downwelling_vector', data, fill_value)
t0 = pd.to_datetime(min(t)).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(t)).strftime('%Y-%m-%dT%H:%M:%S')
deploy_final = vinfo['deployments']
deploy = list(np.unique(deploy_final))
deployments = [int(dd) for dd in deploy]
sname = '-'.join((r, sv))
fig, ax = pf.plot_spkir(t, dd_data, sv, sv_units[0])
ax.set_title((r + '\nDeployments: ' + str(sorted(deployments)) + '\n' + t0 + ' - ' + t1 + '\n'
+ 'removed global ranges +/- [{} - {}]'.format(g_min, g_max)), fontsize=8)
for etimes in dend_times:
ax.axvline(x=etimes, color='k', linestyle='--', linewidth=.6)
pf.save_fig(save_dir, sname)
# plot each wavelength
wavelengths = ['412nm', '443nm', '490nm', '510nm', '555nm', '620nm', '683nm']
for wvi in range(len(dd_data)):
fig, ax = pf.plot_spkir_wv(t, dd_data[wvi], sv, sv_units[0], wvi)
ax.set_title(
(r + '\nDeployments: ' + str(sorted(deployments)) + '\n' + t0 + ' - ' + t1 + '\n'
+ 'removed global ranges +/- [{} - {}]'.format(g_min, g_max)), fontsize=8)
for etimes in dend_times:
ax.axvline(x=etimes, color='k', linestyle='--', linewidth=.6)
snamewvi = '-'.join((sname, wavelengths[wvi]))
pf.save_fig(save_dir, snamewvi)
elif 'presf_abc_wave_burst' in m:
fv_lst = np.unique(vinfo['fv']).tolist()
if len(fv_lst) == 1:
fill_value = fv_lst[0]
else:
print(fv_lst)
print('No unique fill value for {}'.format(sv))
sv_units = np.unique(vinfo['units']).tolist()
t = vinfo['t']
if len(t) > 1:
data = vinfo['values']
[dd_data, g_min, g_max] = index_dataset_2d(r, 'presf_wave_burst_pressure', data, fill_value)
t0 = pd.to_datetime(min(t)).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(t)).strftime('%Y-%m-%dT%H:%M:%S')
deploy_final = vinfo['deployments']
deploy = list(np.unique(deploy_final))
deployments = [int(dd) for dd in deploy]
sname = '-'.join((r, sv))
fig, ax = pf.plot_presf_2d(t, dd_data, sv, sv_units[0])
ax.set_title((r + '\nDeployments: ' + str(sorted(deployments)) + '\n' + t0 + ' - ' + t1 + '\n'
+ 'removed global ranges +/- [{} - {}]'.format(g_min, g_max)), fontsize=8)
for etimes in dend_times:
ax.axvline(x=etimes, color='k', linestyle='--', linewidth=.6)
pf.save_fig(save_dir, sname)
else:
if type(vinfo['values']) != dict: # if the variable is not a 2D array
if 'Spectra' not in sv:
if len(vinfo['t']) < 1:
print('no variable data to plot')
else:
sv_units = vinfo['units'][0]
sv_name = vinfo['var_name']
t0 = pd.to_datetime(min(vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
x = vinfo['t']
y = vinfo['values']
# reject NaNs and values of 0.0
nan_ind = (~np.isnan(y)) & (y != 0.0)
x_nonan = x[nan_ind]
y_nonan = y[nan_ind]
# reject fill values
fv_ind = y_nonan != vinfo['fv'][0]
x_nonan_nofv = x_nonan[fv_ind]
y_nonan_nofv = y_nonan[fv_ind]
# reject extreme values
Ev_ind = cf.reject_extreme_values(y_nonan_nofv)
y_nonan_nofv_nE = y_nonan_nofv[Ev_ind]
x_nonan_nofv_nE = x_nonan_nofv[Ev_ind]
# reject values outside global ranges:
global_min, global_max = cf.get_global_ranges(r, sv_name)
if any(e is None for e in [global_min, global_max]):
y_nonan_nofv_nE_nogr = y_nonan_nofv_nE
x_nonan_nofv_nE_nogr = x_nonan_nofv_nE
else:
gr_ind = cf.reject_global_ranges(y_nonan_nofv_nE, global_min, global_max)
y_nonan_nofv_nE_nogr = y_nonan_nofv_nE[gr_ind]
x_nonan_nofv_nE_nogr = x_nonan_nofv_nE[gr_ind]
if len(y_nonan_nofv) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((r, sv))
else:
sname = '-'.join((r, m, sv))
plt_deploy = [int(x) for x in list(np.unique(vinfo['deployments']))]
# plot hourly averages for cabled and FDCHP data
if 'streamed' in sci_vars_dict[list(sci_vars_dict.keys())[0]]['ms'][0] or 'FDCHP' in r:
sname = '-'.join((sname, 'hourlyavg'))
df = pd.DataFrame({'dfx': x_nonan_nofv_nE_nogr, 'dfy': y_nonan_nofv_nE_nogr})
dfr = df.resample('H', on='dfx').mean()
# Plot all data
fig, ax = pf.plot_timeseries_all(dfr.index, dfr['dfy'], sv, sv_units, stdev=None)
ax.set_title((r + '\nDeployments: ' + str(plt_deploy) + '\n' + t0 + ' - ' + t1),
fontsize=8)
# if plotting a specific time range, plot deployment lines only for those deployments
if type(start_time) == dt.datetime:
for e in list(np.unique(vinfo['deployments'])):
etime = dend_times[int(e) - 1]
ax.axvline(x=etime, color='b', linestyle='--', linewidth=.6)
else:
for etime in dend_times:
ax.axvline(x=etime, color='b', linestyle='--', linewidth=.6)
pf.save_fig(save_dir, sname)
else:
# Plot all data
fig, ax = pf.plot_timeseries_all(x_nonan_nofv, y_nonan_nofv, sv, sv_units, stdev=None)
ax.set_title((r + '\nDeployments: ' + str(plt_deploy) + '\n' + t0 + ' - ' + t1),
fontsize=8)
# if plotting a specific time range, plot deployment lines only for those deployments
if type(start_time) == dt.datetime:
# for e in list(np.unique(vinfo['deployments'])):
# etime = dend_times[int(e) - 1]
# ax.axvline(x=etime, color='b', linestyle='--', linewidth=.6)
etime = dend_times[int(list(np.unique(vinfo['deployments']))[0]) - 1]
ax.axvline(x=etime, color='b', linestyle='--', linewidth=.6)
else:
for etime in dend_times:
ax.axvline(x=etime, color='b', linestyle='--', linewidth=.6)
# if not any(e is None for e in [global_min, global_max]):
# ax.axhline(y=global_min, color='r', linestyle='--', linewidth=.6)
# ax.axhline(y=global_max, color='r', linestyle='--', linewidth=.6)
# else:
# maxpoint = x[np.argmax(y_nonan_nofv)], max(y_nonan_nofv)
# ax.annotate('No Global Ranges', size=8,
# xy=maxpoint, xytext=(5, 5), textcoords='offset points')
pf.save_fig(save_dir, sname)
# Plot data with outliers removed
fig, ax = pf.plot_timeseries_all(x_nonan_nofv_nE_nogr, y_nonan_nofv_nE_nogr, sv, sv_units,
stdev=5)
ax.set_title((r + '\nDeployments: ' + str(plt_deploy) + '\n' + t0 + ' - ' + t1),
fontsize=8)
# if plotting a specific time range, plot deployment lines only for those deployments
if type(start_time) == dt.datetime:
# for e in list(np.unique(vinfo['deployments'])):
# etime = dend_times[int(e) - 1]
# ax.axvline(x=etime, color='b', linestyle='--', linewidth=.6)
etime = dend_times[int(list(np.unique(vinfo['deployments']))[0]) - 1]
ax.axvline(x=etime, color='b', linestyle='--', linewidth=.6)
else:
for etime in dend_times:
ax.axvline(x=etime, color='b', linestyle='--', linewidth=.6)
# if not any(e is None for e in [global_min, global_max]):
# ax.axhline(y=global_min, color='r', linestyle='--', linewidth=.6)
# ax.axhline(y=global_max, color='r', linestyle='--', linewidth=.6)
# else:
# maxpoint = x[np.argmax(y_nonan_nofv_nE_nogr)], max(y_nonan_nofv_nE_nogr)
# ax.annotate('No Global Ranges', size=8,
# xy=maxpoint, xytext=(5, 5), textcoords='offset points')
sfile = '_'.join((sname, 'rmoutliers'))
pf.save_fig(save_dir, sfile)
def main(url_list, sDir, plot_type):
"""""
URL : path to instrument data by methods
sDir : path to the directory on your machine to save files
plot_type: folder name for a plot type
""" ""
rd_list = []
ms_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
ms = uu.split(rd + '-')[1].split('/')[0]
if rd not in rd_list:
rd_list.append(rd)
if ms not in ms_list:
ms_list.append(ms)
'''
separate different instruments
'''
for r in rd_list:
print('\n{}'.format(r))
subsite = r.split('-')[0]
array = subsite[0:2]
main_sensor = r.split('-')[-1]
ps_df, n_streams = cf.get_preferred_stream_info(r)
# read in the analysis file
dr_data = cf.refdes_datareview_json(r)
# get end times of deployments
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = get_deployment_information(dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
# get the list of data files and filter out collocated instruments and other streams chat
datasets = []
for u in url_list:
print(u)
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join(
(splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
fdatasets = cf.filter_collocated_instruments(main_sensor, datasets)
fdatasets = cf.filter_other_streams(r, ms_list, fdatasets)
'''
separate the data files by methods
'''
for ms in ms_list: # np.unique(methodstream)
fdatasets_sel = [x for x in fdatasets if ms in x]
# create a folder to save figures
save_dir = os.path.join(sDir, array, subsite, r, plot_type,
ms.split('-')[0])
cf.create_dir(save_dir)
# create a dictionary for science variables from analysis file
stream_sci_vars_dict = dict()
for x in dr_data['instrument']['data_streams']:
dr_ms = '-'.join((x['method'], x['stream_name']))
if ms == dr_ms:
stream_sci_vars_dict[dr_ms] = dict(vars=dict())
sci_vars = dict()
for y in x['stream']['parameters']:
if y['data_product_type'] == 'Science Data':
sci_vars.update(
{y['name']: dict(db_units=y['unit'])})
if len(sci_vars) > 0:
stream_sci_vars_dict[dr_ms]['vars'] = sci_vars
# initialize an empty data array for science variables in dictionary
sci_vars_dict = cd.initialize_empty_arrays(stream_sci_vars_dict,
ms)
y_unit = []
y_name = []
for fd in fdatasets_sel:
ds = xr.open_dataset(fd, mask_and_scale=False)
print('\nAppending data file: {}'.format(fd.split('/')[-1]))
for var in list(sci_vars_dict[ms]['vars'].keys()):
sh = sci_vars_dict[ms]['vars'][var]
if ds[var].units == sh['db_units']:
if ds[var]._FillValue not in sh['fv']:
sh['fv'].append(ds[var]._FillValue)
if ds[var].units not in sh['units']:
sh['units'].append(ds[var].units)
# time
t = ds['time'].values
t0 = pd.to_datetime(
t.min()).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(
t.max()).strftime('%Y-%m-%dT%H:%M:%S')
# sci variable
z = ds[var].values
sh['t'] = np.append(sh['t'], t)
sh['values'] = np.append(sh['values'], z)
# add pressure to dictionary of sci vars
if 'MOAS' in subsite:
if 'CTD' in main_sensor: # for glider CTDs, pressure is a coordinate
pressure = 'sci_water_pressure_dbar'
y = ds[pressure].values
if ds[pressure].units not in y_unit:
y_unit.append(ds[pressure].units)
if ds[pressure].long_name not in y_name:
y_name.append(ds[pressure].long_name)
else:
pressure = 'int_ctd_pressure'
y = ds[pressure].values
if ds[pressure].units not in y_unit:
y_unit.append(ds[pressure].units)
if ds[pressure].long_name not in y_name:
y_name.append(ds[pressure].long_name)
else:
pressure = pf.pressure_var(ds, ds.data_vars.keys())
y = ds[pressure].values
if ds[pressure].units not in y_unit:
y_unit.append(ds[pressure].units)
if ds[pressure].long_name not in y_name:
y_name.append(ds[pressure].long_name)
sh['pressure'] = np.append(sh['pressure'], y)
if len(y_unit) != 1:
print('pressure unit varies!')
else:
y_unit = y_unit[0]
if len(y_name) != 1:
print('pressure long name varies!')
else:
y_name = y_name[0]
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print('\nWorking on variable: {}'.format(sv))
if len(vinfo['t']) < 1:
print('no variable data to plot')
else:
sv_units = vinfo['units'][0]
fv = vinfo['fv'][0]
t0 = pd.to_datetime(min(
vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(
vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t = vinfo['t']
x = vinfo['values']
y = vinfo['pressure']
# Check if the array is all NaNs
if sum(np.isnan(x)) == len(x):
print('Array of all NaNs - skipping plot.')
continue
# Check if the array is all fill values
elif len(x[x != fv]) == 0:
print('Array of all fill values - skipping plot.')
continue
else:
# reject fill values
fv_ind = x != fv
y_nofv = y[fv_ind]
t_nofv = t[fv_ind]
c_nofv = cm.rainbow(np.linspace(0, 1, len(t[fv_ind])))
x_nofv = x[fv_ind]
print(len(x) - len(fv_ind), ' fill values')
# reject NaNs
nan_ind = ~np.isnan(x)
t_nofv_nonan = t_nofv[nan_ind]
c_nofv_nonan = c_nofv[nan_ind]
y_nofv_nonan = y_nofv[nan_ind]
x_nofv_nonan = x_nofv[nan_ind]
print(len(x) - len(nan_ind), ' NaNs')
# reject extreme values
ev_ind = cf.reject_extreme_values(x_nofv_nonan)
t_nofv_nonan_noev = t_nofv_nonan[ev_ind]
c_nofv_nonan_noev = c_nofv_nonan[ev_ind]
y_nofv_nonan_noev = y_nofv_nonan[ev_ind]
x_nofv_nonan_noev = x_nofv_nonan[ev_ind]
print(len(z) - len(ev_ind), ' Extreme Values', '|1e7|')
# reject values outside global ranges:
global_min, global_max = cf.get_global_ranges(r, sv)
# platform not in qc-table (parad_k_par)
# global_min = 0
# global_max = 2500
print('global ranges for : {}-{} {} - {}'.format(
r, sv, global_min, global_max))
if isinstance(global_min, (int, float)) and isinstance(
global_max, (int, float)):
gr_ind = cf.reject_global_ranges(
x_nofv_nonan_noev, global_min, global_max)
t_nofv_nonan_noev_nogr = t_nofv_nonan_noev[gr_ind]
y_nofv_nonan_noev_nogr = y_nofv_nonan_noev[gr_ind]
x_nofv_nonan_noev_nogr = x_nofv_nonan_noev[gr_ind]
else:
t_nofv_nonan_noev_nogr = t_nofv_nonan_noev
y_nofv_nonan_noev_nogr = y_nofv_nonan_noev
x_nofv_nonan_noev_nogr = x_nofv_nonan_noev
if len(x_nofv_nonan_noev) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((r, sv))
else:
sname = '-'.join((r, m, sv))
if sv != 'pressure':
columns = ['tsec', 'dbar', str(sv)]
bin_size = 10
min_r = int(round(min(y_nofv_nonan_noev) - bin_size))
max_r = int(round(max(y_nofv_nonan_noev) + bin_size))
ranges = list(range(min_r, max_r, bin_size))
groups, d_groups = gt.group_by_depth_range(
t_nofv_nonan_noev_nogr, y_nofv_nonan_noev_nogr,
x_nofv_nonan_noev_nogr, columns, ranges)
y_avg, n_avg, n_min, n_max, n0_std, n1_std, l_arr = [], [], [], [], [], [], []
tm = 1
for ii in range(len(groups)):
nan_ind = d_groups[ii + tm].notnull()
xtime = d_groups[ii + tm][nan_ind]
colors = cm.rainbow(np.linspace(0, 1, len(xtime)))
ypres = d_groups[ii + tm + 1][nan_ind]
nval = d_groups[ii + tm + 2][nan_ind]
tm += 2
l_arr.append(len(
nval)) # count of data to filter out small groups
y_avg.append(ypres.mean())
n_avg.append(nval.mean())
n_min.append(nval.min())
n_max.append(nval.max())
n_std = 3
n0_std.append(nval.mean() + n_std * nval.std())
n1_std.append(nval.mean() - n_std * nval.std())
# Plot all data
ylabel = y_name + " (" + y_unit + ")"
xlabel = sv + " (" + sv_units + ")"
clabel = 'Time'
fig, ax = pf.plot_profiles(x_nofv_nonan_noev_nogr,
y_nofv_nonan_noev_nogr,
t_nofv_nonan_noev_nogr,
ylabel,
xlabel,
clabel,
end_times,
deployments,
stdev=None)
title_text = ' '.join((r, ms.split('-')[-1])) + '\n' \
+ t0 + ' - ' + t1 + '\n' + str(bin_size) +\
' m average and ' + str(n_std) + ' std shown'
ax.set_title(title_text, fontsize=9)
ax.plot(n_avg, y_avg, '-k')
ax.fill_betweenx(y_avg,
n0_std,
n1_std,
color='m',
alpha=0.2)
pf.save_fig(save_dir, sname)
# Plot data with outliers removed
fig, ax = pf.plot_profiles(x_nofv_nonan_noev_nogr,
y_nofv_nonan_noev_nogr,
t_nofv_nonan_noev_nogr,
ylabel,
xlabel,
clabel,
end_times,
deployments,
stdev=5)
ax.set_title(' '.join((r, ms.split('-')[-1])) + '\n' \
+ t0 + ' - ' + t1, fontsize=9)
sfile = '_'.join((sname, 'rmoutliers'))
pf.save_fig(save_dir, sfile)
def main(sDir, url_list):
rd_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
if rd not in rd_list:
rd_list.append(rd)
for r in rd_list:
print('\n{}'.format(r))
subsite = r.split('-')[0]
array = subsite[0:2]
ps_df, n_streams = cf.get_preferred_stream_info(r)
# get end times of deployments
dr_data = cf.refdes_datareview_json(r)
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = get_deployment_information(dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
# filter datasets
datasets = []
for u in url_list:
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join((splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
main_sensor = r.split('-')[-1]
fdatasets = cf.filter_collocated_instruments(main_sensor, datasets)
methodstream = []
for f in fdatasets:
methodstream.append('-'.join((f.split('/')[-2].split('-')[-2], f.split('/')[-2].split('-')[-1])))
for ms in np.unique(methodstream):
fdatasets_sel = [x for x in fdatasets if ms in x]
check_ms = ms.split('-')[1]
if 'recovered' in check_ms:
check_ms = check_ms.split('_recovered')
save_dir = os.path.join(sDir, array, subsite, r, 'timeseries_monthly_plot',
check_ms[0], ms.split('-')[0])
cf.create_dir(save_dir)
stream_sci_vars_dict = dict()
for x in dr_data['instrument']['data_streams']:
dr_ms = '-'.join((x['method'], x['stream_name']))
if ms == dr_ms:
stream_sci_vars_dict[dr_ms] = dict(vars=dict())
sci_vars = dict()
for y in x['stream']['parameters']:
if y['data_product_type'] == 'Science Data':
sci_vars.update({y['name']: dict(db_units=y['unit'])})
if len(sci_vars) > 0:
stream_sci_vars_dict[dr_ms]['vars'] = sci_vars
sci_vars_dict = cd.initialize_empty_arrays(stream_sci_vars_dict, ms)
print('\nAppending data from files: {}'.format(ms))
for fd in fdatasets_sel:
ds = xr.open_dataset(fd, mask_and_scale=False)
for var in list(sci_vars_dict[ms]['vars'].keys()):
sh = sci_vars_dict[ms]['vars'][var]
if ds[var].units == sh['db_units']:
if ds[var]._FillValue not in sh['fv']:
sh['fv'].append(ds[var]._FillValue)
if ds[var].units not in sh['units']:
sh['units'].append(ds[var].units)
tD = ds['time'].values
varD = ds[var].values
sh['t'] = np.append(sh['t'], tD)
sh['values'] = np.append(sh['values'], varD)
print('\nPlotting data')
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print(sv)
if len(vinfo['t']) < 1:
print('no variable data to plot')
else:
sv_units = vinfo['units'][0]
t0 = pd.to_datetime(min(vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
x = vinfo['t']
y = vinfo['values']
# reject NaNs
nan_ind = ~np.isnan(y)
x_nonan = x[nan_ind]
y_nonan = y[nan_ind]
# reject fill values
fv_ind = y_nonan != vinfo['fv'][0]
x_nonan_nofv = x_nonan[fv_ind]
y_nonan_nofv = y_nonan[fv_ind]
# reject extreme values
Ev_ind = cf.reject_extreme_values(y_nonan_nofv)
y_nonan_nofv_nE = y_nonan_nofv[Ev_ind]
x_nonan_nofv_nE = x_nonan_nofv[Ev_ind]
# reject values outside global ranges:
global_min, global_max = cf.get_global_ranges(r, sv)
gr_ind = cf.reject_global_ranges(y_nonan_nofv_nE, global_min, global_max)
y_nonan_nofv_nE_nogr = y_nonan_nofv_nE[gr_ind]
x_nonan_nofv_nE_nogr = x_nonan_nofv_nE[gr_ind]
title = ' '.join((r, ms.split('-')[0]))
if len(y_nonan_nofv) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((r, sv))
else:
sname = '-'.join((r, m, sv))
# 1st group by year
ygroups, gy_data = gt.group_by_timerange(x_nonan_nofv_nE_nogr, y_nonan_nofv_nE_nogr, 'A')
tn = 1
for n in range(len(ygroups)):
x_time = gy_data[n+tn].dropna(axis=0)
y_data = gy_data[n+(tn+1)].dropna(axis=0)
y_data = y_data.astype(float)
# 2nd group by month
mgroups, gm_data = gt.group_by_timerange(x_time.values, y_data.values, 'M')
x_year = x_time[0].year
print(x_year)
#
# create bins for histogram
mgroups_min = min(mgroups.describe()['DO']['min'])
mgroups_max = max(mgroups.describe()['DO']['max'])
lower_bound = int(round(mgroups_min))
upper_bound = int(round(mgroups_max + (mgroups_max / 50)))
step_bound = int(round((mgroups_max - mgroups_min) / 10))
lower_bound = int(round(global_min))
upper_bound = int(round(global_max + (global_max / 50)))
step_bound = int(round((global_max - global_min) / 10))
if step_bound == 0:
step_bound += 1
if (upper_bound - lower_bound) == step_bound:
lower_bound -= 1
upper_bound += 1
if (upper_bound - lower_bound) < step_bound:
step_bound = int(round(step_bound / 10))
bin_range = list(range(lower_bound, upper_bound, step_bound))
print(bin_range)
# create color palette
colors = color_names[:len(mgroups)]
print('1--- ', len(colors))
print(colors)
fig0, ax0 = pyplot.subplots(nrows=2, ncols=1)
# # subplot for histogram and basic statistics table
ax0[0].axis('off')
ax0[0].axis('tight')
the_table = ax0[0].table(cellText=mgroups.describe().round(2).values,
rowLabels=mgroups.describe().index.month,
rowColours=colors,
colLabels=mgroups.describe().columns.levels[1], loc='center')
the_table.set_fontsize(5)
fig, ax = pyplot.subplots(nrows=12, ncols=1, sharey=True)
for kk in list(range(0, 12)):
ax[kk].tick_params(axis='both', which='both', color='r', labelsize=7,
labelcolor='m', rotation=0, pad=0.1, length=1)
month_name = calendar.month_abbr[kk + 1]
ax[kk].set_ylabel(month_name, rotation=0, fontsize=8, color='b', labelpad=20)
if kk == 0:
ax[kk].set_title(str(x_year) + '\n ' + sv + " (" + sv_units + ")" +
' Global Range: [' + str(int(global_min)) + ',' + str(int(global_max)) + ']' +
'\n End of deployments are marked with a vertical line \n ' +
'Plotted: Data, Mean and STD (Method: 1 day' +
' rolling window calculations)',
fontsize=8)
if kk < 11:
ax[kk].tick_params(labelbottom=False)
if kk == 11:
ax[kk].set_xlabel('Days', rotation=0, fontsize=8, color='b')
tm = 1
for mt in range(len(mgroups)):
x_time = gm_data[mt+tm].dropna(axis=0)
y_data = gm_data[mt+(tm+1)].dropna(axis=0)
if len(x_time) == 0:
# ax[plt_index].tick_params(which='both', labelbottom=False, labelleft=False,
# pad=0.1, length=1)
continue
x_month = x_time[0].month
col_name = str(x_month)
series_m = pd.DataFrame(columns=[col_name], index=x_time)
series_m[col_name] = list(y_data[:])
# serie_n.plot.hist(ax=ax0[0], bins=bin_range,
# histtype='bar', color=colors[ny], stacked=True)
series_m.plot.kde(ax=ax0[0], color=colors[mt])
ax0[0].legend(fontsize=8, bbox_to_anchor=(0., 1.12, 1., .102), loc=3,
ncol=len(mgroups), mode="expand", borderaxespad=0.)
# ax0[0].set_xticks(bin_range)
ax0[0].set_xlabel('Observation Ranges' + ' (' + sv + ', ' + sv_units + ')', fontsize=8)
ax0[0].set_ylabel('Density', fontsize=8) # 'Number of Observations'
ax0[0].set_title('Kernel Density Estimates', fontsize=8)
ax0[0].tick_params(which='both', labelsize=7, pad=0.1, length=1, rotation=0)
plt_index = x_month - 1
# Plot data
series_m.plot(ax=ax[plt_index], linestyle='None', marker='.', markersize=1)
ax[plt_index].legend().set_visible(False)
ma = series_m.rolling('86400s').mean()
mstd = series_m.rolling('86400s').std()
ax[plt_index].plot(ma.index, ma[col_name].values, 'b')
ax[plt_index].fill_between(mstd.index, ma[col_name].values-3*mstd[col_name].values,
ma[col_name].values+3*mstd[col_name].values,
color='b', alpha=0.2)
# prepare the time axis parameters
mm, nod = monthrange(x_year, x_month)
datemin = datetime.date(x_year, x_month, 1)
datemax = datetime.date(x_year, x_month, nod)
ax[plt_index].set_xlim(datemin, datemax)
xlocator = mdates.DayLocator() # every day
myFmt = mdates.DateFormatter('%d')
ax[plt_index].xaxis.set_major_locator(xlocator)
ax[plt_index].xaxis.set_major_formatter(myFmt)
ax[plt_index].xaxis.set_minor_locator(pyplot.NullLocator())
ax[plt_index].xaxis.set_minor_formatter(pyplot.NullFormatter())
# data_min = min(ma.DO_n.dropna(axis=0) - 5 * mstd.DO_n.dropna(axis=0))
# 0data_max = max(ma.DO_n.dropna(axis=0) + 5 * mstd.DO_n.dropna(axis=0))
# ax[plt_index].set_ylim([data_min, data_max])
ylocator = MaxNLocator(prune='both', nbins=3)
ax[plt_index].yaxis.set_major_locator(ylocator)
if x_month != 12:
ax[plt_index].tick_params(which='both', labelbottom=False, pad=0.1, length=1)
ax[plt_index].set_xlabel(' ')
else:
ax[plt_index].tick_params(which='both', color='r', labelsize=7, labelcolor='m',
pad=0.1, length=1, rotation=0)
ax[plt_index].set_xlabel('Days', rotation=0, fontsize=8, color='b')
dep = 1
for etimes in end_times:
ax[plt_index].axvline(x=etimes, color='b', linestyle='--', linewidth=.8)
if ma[col_name].values.any():
ax[plt_index].text(etimes, max(ma[col_name].dropna(axis=0)), 'End' + str(dep),
fontsize=6, style='italic',
bbox=dict(boxstyle='round',
ec=(0., 0.5, 0.5),
fc=(1., 1., 1.),
))
else:
ax[plt_index].text(etimes, min(series_m['DO_n']), 'End' + str(dep),
fontsize=6, style='italic',
bbox=dict(boxstyle='round',
ec=(0., 0.5, 0.5),
fc=(1., 1., 1.),
))
dep += 1
tm += 1
tn += 1
# pyplot.show()
sfile = '_'.join((str(x_year), sname))
save_file = os.path.join(save_dir, sfile)
fig.savefig(str(save_file), dpi=150)
sfile = '_'.join(('Statistics', str(x_year), sname))
save_file = os.path.join(save_dir, sfile)
fig0.savefig(str(save_file), dpi=150)
def main(sDir, url_list, preferred_only):
rd_list = []
ms_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
ms = uu.split(rd + '-')[1].split('/')[0]
if rd not in rd_list:
rd_list.append(rd)
if ms not in ms_list:
ms_list.append(ms)
for r in rd_list:
print('\n{}'.format(r))
subsite = r.split('-')[0]
array = subsite[0:2]
# filter datasets
datasets = []
for u in url_list:
print(u)
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join((splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
fdatasets = []
if preferred_only == 'yes':
# get the preferred stream information
ps_df, n_streams = cf.get_preferred_stream_info(r)
for index, row in ps_df.iterrows():
for ii in range(n_streams):
try:
rms = '-'.join((r, row[ii]))
except TypeError:
continue
for dd in datasets:
spl = dd.split('/')[-2].split('-')
catalog_rms = '-'.join((spl[1], spl[2], spl[3], spl[4], spl[5], spl[6]))
fdeploy = dd.split('/')[-1].split('_')[0]
if rms == catalog_rms and fdeploy == row['deployment']:
fdatasets.append(dd)
else:
fdatasets = datasets
main_sensor = r.split('-')[-1]
fdatasets = cf.filter_collocated_instruments(main_sensor, fdatasets)
# ps_df, n_streams = cf.get_preferred_stream_info(r)
# get end times of deployments
dr_data = cf.refdes_datareview_json(r)
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = get_deployment_information(dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
# filter datasets
# datasets = []
# for u in url_list:
# splitter = u.split('/')[-2].split('-')
# rd_check = '-'.join((splitter[1], splitter[2], splitter[3], splitter[4]))
# if rd_check == r:
# udatasets = cf.get_nc_urls([u])
# datasets.append(udatasets)
# datasets = list(itertools.chain(*datasets))
# main_sensor = r.split('-')[-1]
# fdatasets = cf.filter_collocated_instruments(main_sensor, datasets)
# fdatasets = cf.filter_other_streams(r, ms_list, fdatasets)
methodstream = []
for f in fdatasets:
methodstream.append('-'.join((f.split('/')[-2].split('-')[-2], f.split('/')[-2].split('-')[-1])))
for ms in np.unique(methodstream):
fdatasets_sel = [x for x in fdatasets if ms in x]
save_dir = os.path.join(sDir, array, subsite, r, 'timeseries_yearly_plot', ms.split('-')[0])
cf.create_dir(save_dir)
stream_sci_vars_dict = dict()
for x in dr_data['instrument']['data_streams']:
dr_ms = '-'.join((x['method'], x['stream_name']))
if ms == dr_ms:
stream_sci_vars_dict[dr_ms] = dict(vars=dict())
sci_vars = dict()
for y in x['stream']['parameters']:
if 'light' in y['name']:
sci_vars.update({y['name']: dict(db_units=y['unit'])})
# if y['data_product_type'] == 'Science Data':
# sci_vars.update({y['name']: dict(db_units=y['unit'])})
if len(sci_vars) > 0:
stream_sci_vars_dict[dr_ms]['vars'] = sci_vars
sci_vars_dict = cd.initialize_empty_arrays(stream_sci_vars_dict, ms)
print('\nAppending data from files: {}'.format(ms))
fcount = 0
for fd in fdatasets_sel:
ds = xr.open_dataset(fd, mask_and_scale=False)
for var in list(sci_vars_dict[ms]['vars'].keys()):
sh = sci_vars_dict[ms]['vars'][var]
if ds[var].units == sh['db_units']:
if ds[var]._FillValue not in sh['fv']:
sh['fv'].append(ds[var]._FillValue)
if ds[var].units not in sh['units']:
sh['units'].append(ds[var].units)
tD = ds['time'].values
varD = ds[var].values
sh['t'] = np.append(sh['t'], tD) # put deployments time series together
if fcount == 0:
if varD.ndim > 1:
sh['values'] = np.zeros(shape=[varD.shape[0], varD.shape[1]])
sh['values'][:] = varD
else:
sh['values'] = np.append(sh['values'], varD)
else:
if varD.ndim > 1:
sh['values'] = np.vstack([sh['values'], varD])
else:
sh['values'] = np.append(sh['values'], varD)
print(fcount, var, sh['values'].shape, sh['t'].shape)
fcount += 1
print('\nPlotting data')
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print(sv)
if len(vinfo['t']) < 1:
print('no variable data to plot')
else:
sv_units = vinfo['units'][0]
t0 = pd.to_datetime(min(vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
x = vinfo['t']
y = vinfo['values']
if len(y) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((r, sv))
else:
sname = '-'.join((r, m, sv))
if y.ndim == 1:
num_col = 1
else:
num_col = y.shape[1]
col = [str(x) for x in list(range(1, num_col + 1))]
col.insert(0, 'time')
groups, d_groups = gt.group_by_time_frequency(x, y, col, 'A')
if len(d_groups.columns) == len(col):
time_ind = [1]
data_ind0 = [2]
data_ind1 = [len(col)]
else:
time_ind = list(range(1, len(d_groups.columns), len(col)))
data_ind0 = list(range(2, len(d_groups.columns), len(col)))
if data_ind0[-1] == d_groups.columns.values[-(len(col)+1)]:
data_ind0.insert(len(groups) - 1, len(d_groups.columns))
data_ind1 = list(range(len(col), len(d_groups.columns), len(col)))
data_ind1.insert(len(groups)-1, len(d_groups.columns))
save_file = os.path.join(save_dir, sname)
print('\ncreating images')
fig, ax = pyplot.subplots(nrows=len(groups), ncols=1)
if len(groups) == 1:
ax = [ax]
colors = color_names[:len(groups)]
images = []
for group in range(len(groups)):
nan_ind = d_groups[time_ind[group]].notnull()
xtime = d_groups[time_ind[group]][nan_ind]
ycol = list(range(data_ind0[group], data_ind1[group] + 1))
ydata = d_groups[ycol][nan_ind]
if len(xtime) != 0 and len(ydata) != 0:
n_year = xtime[0].year
print(n_year)
ydata = ydata.set_index(xtime)
if len(ydata.columns) > 1:
print('more than one col: ', len(ydata.columns))
b = fsplt.split_by_timegap(ydata, 86400)
if b:
print('gaps exist, splitting data')
for ib in (range(len(b))):
iydata = b[ib][b[ib].columns.values[0:-2]]
Y = iydata.columns.values
X = iydata.index.values
Z = iydata.values
if Z.shape[0] == 1:
X = np.repeat(X[0], len(Y), axis=0)
df = pd.DataFrame(dict(a=list(X), b=list(Y), c=list(Z[0])))
images.append(ax[group].scatter(df['a'].values, df['b'].values, c=df['c'].values, cmap='Blues', s=1))
else:
Z = Z.T
x, y = np.meshgrid(X, Y)
images.append(ax[group].contourf(x, y, Z, alpha=0.7, cmap='Blues'))
else:
print('no gaps exist, not splitting data')
iydata = ydata[ydata.columns.values[0:-2]]
Y = iydata.columns.values
X = iydata.index.values
Z = iydata.values
if Z.shape[0] == 1:
X = np.repeat(X[0], len(Y), axis=0)
df = pd.DataFrame(dict(a=list(X), b=list(Y), c=list(Z[0])))
images.append(
ax[group].scatter(df['a'].values, df['b'].values, c=df['c'].values,
cmap='Blues', s=1))
elif Z.shape[1] < 2:
Y = np.repeat(Y[0], len(X), axis=1)
df = pd.DataFrame(dict(a=list(X), b=list(Y), c=list(Z[0])))
images.append(
ax[group].scatter(df['a'].values, df['b'].values, c=df['c'].values,
cmap='Blues', s=1))
else:
Z = Z.T
x, y = np.meshgrid(X, Y)
im = ax[group].contourf(x, y, Z, alpha=0.7, cmap='Blues') #pyplot.cm.jet
images.append(ax[group].contourf(x, y, Z, alpha=0.7, cmap='Blues'))
else:
print('with one column:', len(ydata.columns))
ax[group].contourf(x, y, Z, alpha=0.7, cmap='Blues')(ydata.plot(ax=ax[group],
linestyle='None',
marker='.',
markersize=0.5,
color=colors[group]))
ax[group].legend().set_visible(False)
# plot Mean and Standard deviation
ma = ydata.rolling('86400s').mean()
mstd = ydata.rolling('86400s').std()
m_mstd_min = ma[ycol].values - 2 * mstd[ycol].values
m_mstd_max = ma[ycol].values + 2 * mstd[ycol].values
ax[group].plot(ma.index.values, ma[ycol].values, 'k', linewidth=0.15)
ax[group].fill_between(mstd.index.values, m_mstd_min, m_mstd_max, color='b', alpha=0.2)
# flag deployments end-time for reference
ymin, ymax = ax[group].get_ylim()
dep = 1
for etimes in end_times:
if etimes.year == n_year:
ax[group].axvline(x=etimes, color='b', linestyle='--', linewidth=.6)
ax[group].text(etimes, ymin, 'End' + str(dep), fontsize=6, style='italic',
bbox=dict(boxstyle='round', ec=(0., 0.5, 0.5), fc=(1., 1., 1.)))
dep += 1
# prepare the time axis parameters
datemin = datetime.date(n_year, 1, 1)
datemax = datetime.date(n_year, 12, 31)
ax[group].set_xlim(datemin, datemax)
xlocator = mdates.MonthLocator() # every month
myFmt = mdates.DateFormatter('%m')
ax[group].xaxis.set_minor_locator(xlocator)
ax[group].xaxis.set_major_formatter(myFmt)
# prepare the y axis parameters
ax[group].set_ylabel(n_year, rotation=0, fontsize=8, color='b', labelpad=20)
ax[group].yaxis.set_label_position("right")
ylocator = MaxNLocator(prune='both', nbins=3)
ax[group].yaxis.set_major_locator(ylocator)
ax[group].yaxis.set_ticklabels([]) #range(1, len(col), 1)
# format figure
ax[group].tick_params(axis='both', color='r', labelsize=7, labelcolor='m')
if group == 0:
ax[group].set_title(sv + '( ' + sv_units + ')', fontsize=8)
if group < len(groups) - 1:
ax[group].tick_params(which='both', pad=0.1, length=1, labelbottom=False)
ax[group].set_xlabel(' ')
else:
ax[group].tick_params(which='both', color='r', labelsize=7, labelcolor='m',
pad=0.1, length=1, rotation=0)
ax[group].set_xlabel('Months', rotation=0, fontsize=8, color='b')
vmin = min(image.get_array().min() for image in images)
vmax = max(image.get_array().max() for image in images)
norm = mcolors.Normalize(vmin=vmin, vmax=vmax)
for im in images:
im.set_norm(norm)
fig.colorbar(images[0], ax=ax, orientation='horizontal', fraction=.1, spacing='proportional')
fig.savefig(str(save_file), dpi=150)
pyplot.close()
def main(sDir, url_list):
rd_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
if rd not in rd_list:
rd_list.append(rd)
for r in rd_list:
print('\n{}'.format(r))
subsite = r.split('-')[0]
array = subsite[0:2]
ps_df, n_streams = cf.get_preferred_stream_info(r)
# get end times of deployments
dr_data = cf.refdes_datareview_json(r)
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = get_deployment_information(dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
# filter datasets
datasets = []
for u in url_list:
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join(
(splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
main_sensor = r.split('-')[-1]
fdatasets = cf.filter_collocated_instruments(main_sensor, datasets)
methodstream = []
for f in fdatasets:
methodstream.append('-'.join((f.split('/')[-2].split('-')[-2],
f.split('/')[-2].split('-')[-1])))
for ms in np.unique(methodstream):
fdatasets_sel = [x for x in fdatasets if ms in x]
save_dir = os.path.join(sDir, array, subsite, r,
'timeseries_daily_plots',
ms.split('-')[0])
cf.create_dir(save_dir)
stream_sci_vars_dict = dict()
for x in dr_data['instrument']['data_streams']:
dr_ms = '-'.join((x['method'], x['stream_name']))
if ms == dr_ms:
stream_sci_vars_dict[dr_ms] = dict(vars=dict())
sci_vars = dict()
for y in x['stream']['parameters']:
if y['data_product_type'] == 'Science Data':
sci_vars.update(
{y['name']: dict(db_units=y['unit'])})
if len(sci_vars) > 0:
stream_sci_vars_dict[dr_ms]['vars'] = sci_vars
sci_vars_dict = cd.initialize_empty_arrays(stream_sci_vars_dict,
ms)
print('\nAppending data from files: {}'.format(ms))
for fd in fdatasets_sel:
ds = xr.open_dataset(fd, mask_and_scale=False)
for var in list(sci_vars_dict[ms]['vars'].keys()):
sh = sci_vars_dict[ms]['vars'][var]
if ds[var].units == sh['db_units']:
if ds[var]._FillValue not in sh['fv']:
sh['fv'].append(ds[var]._FillValue)
if ds[var].units not in sh['units']:
sh['units'].append(ds[var].units)
tD = ds['time'].values
varD = ds[var].values
sh['t'] = np.append(sh['t'], tD)
sh['values'] = np.append(sh['values'], varD)
print('\nPlotting data')
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print(sv)
if len(vinfo['t']) < 1:
print('no variable data to plot')
else:
sv_units = vinfo['units'][0]
t0 = pd.to_datetime(min(
vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(
vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
x = vinfo['t']
y = vinfo['values']
# reject NaNs
nan_ind = ~np.isnan(y)
x_nonan = x[nan_ind]
y_nonan = y[nan_ind]
# reject fill values
fv_ind = y_nonan != vinfo['fv'][0]
x_nonan_nofv = x_nonan[fv_ind]
y_nonan_nofv = y_nonan[fv_ind]
# reject extreme values
Ev_ind = cf.reject_extreme_values(y_nonan_nofv)
y_nonan_nofv_nE = y_nonan_nofv[Ev_ind]
x_nonan_nofv_nE = x_nonan_nofv[Ev_ind]
# reject values outside global ranges:
global_min, global_max = cf.get_global_ranges(r, sv)
gr_ind = cf.reject_global_ranges(
y_nonan_nofv_nE, global_min, global_max)
y_nonan_nofv_nE_nogr = y_nonan_nofv_nE[gr_ind]
x_nonan_nofv_nE_nogr = x_nonan_nofv_nE[gr_ind]
if len(y_nonan_nofv) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((r, sv))
else:
sname = '-'.join((r, m, sv))
# 1st group by year
ygroups, gy_data = gt.group_by_timerange(
x_nonan_nofv_nE_nogr, y_nonan_nofv_nE_nogr,
'A')
tn = 1
for n in range(len(ygroups)):
x_time = gy_data[n + tn].dropna(axis=0)
y_data = gy_data[n + (tn + 1)].dropna(axis=0)
# 2nd group by month
mgroups, gm_data = gt.group_by_timerange(
x_time.values, y_data.values, 'M')
if len(x_time) == 0:
continue
td = 1
for jj in range(len(mgroups)):
x_time = gm_data[jj + td].dropna(axis=0)
y_data = gm_data[jj +
(td + 1)].dropna(axis=0)
if len(x_time) == 0:
continue
# 3rd group by day
dgroups, gd_data = gt.group_by_timerange(
x_time.values, y_data.values, 'D')
x_year = x_time[0].year
x_month = x_time[0].month
month_name = calendar.month_abbr[x_month]
print(x_year, x_month)
sfile = '_'.join(
(str(x_year), str(x_month), sname))
# prepare plot layout
fig, ax = pyplot.subplots(nrows=7,
ncols=5,
sharey=True)
title_in = month_name + '-' + str(x_year) + \
' calendar days \n Parameter: ' + \
sv + " (" + sv_units + ")"
ax[0][2].text(0.5,
1.5,
title_in,
horizontalalignment='center',
fontsize=8,
transform=ax[0][2].transAxes)
num_i = 0
day_i = {}
for kk in list(range(0, 7)):
for ff in list(range(0, 5)):
num_i += 1
day_i[num_i] = [kk, ff]
ax[kk][ff].tick_params(
axis='both',
which='both',
color='r',
labelsize=7,
labelcolor='m',
rotation=0)
ax[kk][ff].text(
0.1,
0.75,
str(num_i),
horizontalalignment='center',
fontsize=7,
transform=ax[kk][ff].transAxes,
bbox=dict(
boxstyle="round",
ec=(0., 0.5, 0.5),
fc=(1., 1., 1.),
))
if kk is not 6:
ax[kk][ff].tick_params(
labelbottom=False)
if ff is not 0:
ax[kk][ff].tick_params(
labelright=False)
if kk is 6 and ff is 0:
ax[kk][ff].set_xlabel(
'Hours',
rotation=0,
fontsize=8,
color='b')
if kk is 6 and ff in list(
range(1, 5)):
fig.delaxes(ax[kk][ff])
tm = 1
for mt in range(len(dgroups)):
x_time = gd_data[mt +
tm].dropna(axis=0)
y_DO = gd_data[mt +
(tm + 1)].dropna(axis=0)
series_m = pd.DataFrame(
columns=['DO_n'], index=x_time)
series_m['DO_n'] = list(y_DO[:])
if len(x_time) == 0:
continue
x_day = x_time[0].day
print(x_time[0].year, x_time[0].month,
x_day)
i0 = day_i[x_day][0]
i1 = day_i[x_day][1]
# Plot data
series_m.plot(ax=ax[i0][i1],
linestyle='None',
marker='.',
markersize=1)
ax[i0][i1].legend().set_visible(False)
ma = series_m.rolling('3600s').mean()
mstd = series_m.rolling('3600s').std()
ax[i0][i1].plot(ma.index,
ma.DO_n,
'b',
linewidth=0.25)
ax[i0][i1].fill_between(
mstd.index,
ma.DO_n - 3 * mstd.DO_n,
ma.DO_n + 3 * mstd.DO_n,
color='b',
alpha=0.2)
# prepare the time axis parameters
datemin = datetime.datetime(
x_year, x_month, x_day, 0)
datemax = datetime.datetime(
x_year, x_month, x_day, 23)
ax[i0][i1].set_xlim(datemin, datemax)
xLocator = mdates.HourLocator(
interval=4) # every hour
myFmt = mdates.DateFormatter('%H')
ax[i0][i1].xaxis.set_minor_locator(
xLocator)
ax[i0][i1].xaxis.set_minor_formatter(
myFmt)
ax[i0][i1].xaxis.set_major_locator(
pyplot.NullLocator())
ax[i0][i1].xaxis.set_major_formatter(
pyplot.NullFormatter())
yLocator = MaxNLocator(prune='both',
nbins=3)
ax[i0][i1].yaxis.set_major_locator(
yLocator)
if x_day is not 31:
ax[i0][i1].tick_params(
labelbottom=False)
ax[i0][i1].set_xlabel(' ')
else:
ax[i0][i1].tick_params(
which='both',
color='r',
labelsize=7,
labelcolor='m',
length=0.1,
pad=0.1)
ax[i0][i1].set_xlabel('Hours',
rotation=0,
fontsize=8,
color='b')
ymin, ymax = ax[i0][i1].get_ylim()
dep = 1
for etimes in end_times:
ax[i0][i1].axvline(x=etimes,
color='b',
linestyle='--',
linewidth=.6)
ax[i0][i1].text(
etimes,
ymin + 50,
str(dep),
fontsize=6,
style='italic',
bbox=dict(
boxstyle="round",
ec=(0., 0.5, 0.5),
fc=(1., 1., 1.),
))
dep += 1
tm += 1
td += 1
pf.save_fig(save_dir, sfile)
tn += 1
def main(url_list, sDir, plot_type):
"""""
URL : path to instrument data by methods
sDir : path to the directory on your machine to save files
plot_type: folder name for a plot type
""" ""
rd_list = []
ms_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
ms = uu.split(rd + '-')[1].split('/')[0]
if rd not in rd_list:
rd_list.append(rd)
if ms not in ms_list:
ms_list.append(ms)
'''
separate different instruments
'''
for r in rd_list:
print('\n{}'.format(r))
subsite = r.split('-')[0]
array = subsite[0:2]
main_sensor = r.split('-')[-1]
ps_df, n_streams = cf.get_preferred_stream_info(r)
# read in the analysis file
dr_data = cf.refdes_datareview_json(r)
# get preferred stream
ps_df, n_streams = cf.get_preferred_stream_info(r)
# get end times of deployments
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = cf.get_deployment_information(
dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
# get the list of data files and filter out collocated instruments and other streams
datasets = []
for u in url_list:
print(u)
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join(
(splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
fdatasets = cf.filter_collocated_instruments(main_sensor, datasets)
fdatasets = cf.filter_other_streams(r, ms_list, fdatasets)
'''
separate data files by methods
'''
for ms in ms_list:
fdatasets_sel = [x for x in fdatasets if ms in x]
# create a folder to save figures
save_dir = os.path.join(sDir, array, subsite, r, plot_type,
ms.split('-')[0])
cf.create_dir(save_dir)
# create a dictionary for science variables from analysis file
stream_sci_vars_dict = dict()
for x in dr_data['instrument']['data_streams']:
dr_ms = '-'.join((x['method'], x['stream_name']))
if ms == dr_ms:
stream_sci_vars_dict[dr_ms] = dict(vars=dict())
sci_vars = dict()
for y in x['stream']['parameters']:
if y['data_product_type'] == 'Science Data':
sci_vars.update(
{y['name']: dict(db_units=y['unit'])})
if len(sci_vars) > 0:
stream_sci_vars_dict[dr_ms]['vars'] = sci_vars
# initialize an empty data array for science variables in dictionary
sci_vars_dict = cd.initialize_empty_arrays(stream_sci_vars_dict,
ms)
print('\nAppending data from files: {}'.format(ms))
y_unit = []
y_name = []
for fd in fdatasets_sel:
ds = xr.open_dataset(fd, mask_and_scale=False)
print('\nAppending data file: {}'.format(fd.split('/')[-1]))
for var in list(sci_vars_dict[ms]['vars'].keys()):
sh = sci_vars_dict[ms]['vars'][var]
if ds[var].units == sh['db_units']:
if ds[var]._FillValue not in sh['fv']:
sh['fv'].append(ds[var]._FillValue)
if ds[var].units not in sh['units']:
sh['units'].append(ds[var].units)
# time
t = ds['time'].values
t0 = pd.to_datetime(
t.min()).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(
t.max()).strftime('%Y-%m-%dT%H:%M:%S')
# sci variable
z = ds[var].values
sh['t'] = np.append(sh['t'], t)
sh['values'] = np.append(sh['values'], z)
# add pressure to dictionary of sci vars
if 'MOAS' in subsite:
if 'CTD' in main_sensor: # for glider CTDs, pressure is a coordinate
pressure = 'sci_water_pressure_dbar'
y = ds[pressure].values
if ds[pressure].units not in y_unit:
y_unit.append(ds[pressure].units)
if ds[pressure].long_name not in y_name:
y_name.append(ds[pressure].long_name)
else:
pressure = 'int_ctd_pressure'
y = ds[pressure].values
if ds[pressure].units not in y_unit:
y_unit.append(ds[pressure].units)
if ds[pressure].long_name not in y_name:
y_name.append(ds[pressure].long_name)
else:
pressure = pf.pressure_var(ds, ds.data_vars.keys())
y = ds[pressure].values
sh['pressure'] = np.append(sh['pressure'], y)
try:
ds[pressure].units
if ds[pressure].units not in y_unit:
y_unit.append(ds[pressure].units)
except AttributeError:
print('pressure attributes missing units')
if 'pressure unit missing' not in y_unit:
y_unit.append('pressure unit missing')
try:
ds[pressure].long_name
if ds[pressure].long_name not in y_name:
y_name.append(ds[pressure].long_name)
except AttributeError:
print('pressure attributes missing long_name')
if 'pressure long name missing' not in y_name:
y_name.append('pressure long name missing')
# create a csv file with diagnostic results:
if len(y_unit) != 1:
print('pressure unit varies')
if 'dbar' in y_unit:
y_unit = 'dbar'
print(y_unit)
else:
y_unit = y_unit[0]
if len(y_name) != 1:
print('pressure long name varies')
if 'Seawater Pressure' in y_name:
y_name = 'Seawater Pressure'
print(y_name)
else:
y_name = y_name[0]
# create a folder to save variables statistics
mDir = '/Users/leila/Documents/NSFEduSupport/github/data-review-tools/data_review/final_stats'
save_dir_stat = os.path.join(mDir, array, subsite)
cf.create_dir(save_dir_stat)
stat_df = pd.DataFrame()
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print(sv)
if len(vinfo['t']) < 1:
print('no variable data to plot')
else:
sv_units = vinfo['units'][0]
fv = vinfo['fv'][0]
t0 = pd.to_datetime(min(
vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(
vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t = vinfo['t']
z = vinfo['values']
y = vinfo['pressure']
title = ' '.join((r, ms))
# Check if the array is all NaNs
if sum(np.isnan(z)) == len(z):
print('Array of all NaNs - skipping plot.')
continue
# Check if the array is all fill values
elif len(z[z != fv]) == 0:
print('Array of all fill values - skipping plot.')
continue
else:
# reject fill values
fv_ind = z != fv
y_nofv = y[fv_ind]
t_nofv = t[fv_ind]
z_nofv = z[fv_ind]
print(len(z) - len(fv_ind), ' fill values')
# reject NaNs
nan_ind = ~np.isnan(z_nofv)
t_nofv_nonan = t_nofv[nan_ind]
y_nofv_nonan = y_nofv[nan_ind]
z_nofv_nonan = z_nofv[nan_ind]
print(len(z) - len(nan_ind), ' NaNs')
# reject extreme values
ev_ind = cf.reject_extreme_values(z_nofv_nonan)
t_nofv_nonan_noev = t_nofv_nonan[ev_ind]
y_nofv_nonan_noev = y_nofv_nonan[ev_ind]
z_nofv_nonan_noev = z_nofv_nonan[ev_ind]
print(
len(z) - len(ev_ind), ' Extreme Values',
'|1e7|')
# reject values outside global ranges:
global_min, global_max = cf.get_global_ranges(
r, sv)
# platform not in qc-table (parad_k_par)
# global_min = 0
# global_max = 2500
print('global ranges for : {}-{} {} - {}'.format(
r, sv, global_min, global_max))
if isinstance(global_min,
(int, float)) and isinstance(
global_max, (int, float)):
gr_ind = cf.reject_global_ranges(
z_nofv_nonan_noev, global_min, global_max)
t_nofv_nonan_noev_nogr = t_nofv_nonan_noev[
gr_ind]
y_nofv_nonan_noev_nogr = y_nofv_nonan_noev[
gr_ind]
z_nofv_nonan_noev_nogr = z_nofv_nonan_noev[
gr_ind]
else:
t_nofv_nonan_noev_nogr = t_nofv_nonan_noev
y_nofv_nonan_noev_nogr = y_nofv_nonan_noev
z_nofv_nonan_noev_nogr = z_nofv_nonan_noev
if len(z_nofv_nonan_noev) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((r, sv))
else:
sname = '-'.join((r, m, sv))
# group by depth range
sname = '_'.join((sname, sv_units))
# if sv != 'pressure':
# columns = ['tsec', 'dbar', str(sv)]
#
# # select depth bin size for the data group function
# bin_size = 10
# min_r = int(round(min(y_nofv_nonan_noev) - bin_size))
# max_r = int(round(max(y_nofv_nonan_noev) + bin_size))
# ranges = list(range(min_r, max_r, bin_size))
# groups, d_groups = gt.group_by_depth_range(t_nofv_nonan_noev_nogr, y_nofv_nonan_noev_nogr,
# z_nofv_nonan_noev_nogr, columns, ranges)
#
# if (ms.split('-')[0]) == (ps_df[0].values[0].split('-')[0]):
# if 'pressure' not in sv:
# print('final_stats_{}-{}-{}-{}'.format(r,
# ms.split('-')[0],
# ps_df[0].values[0].split('-')[0],
# sv))
# stat_data = groups.describe()[sv]
# stat_data.insert(loc=0, column='parameter', value=sv, allow_duplicates=False)
# stat_df = stat_df.append(stat_data)
# if sv == 'optical_backscatter':
# less_ind = z_nofv_nonan_noev < 0.0004
# print(sv, ' < 0.0004', len(less_ind))
# more_ind = z_nofv_nonan_noev > 0.01
# print(sv, ' > 0.01', len(more_ind))
# Plot all data
clabel = sv + " (" + sv_units + ")"
ylabel = y_name + " (" + y_unit + ")"
fig, ax = pf.plot_xsection(subsite,
t_nofv_nonan_noev,
y_nofv_nonan_noev,
z_nofv_nonan_noev,
clabel,
ylabel,
stdev=None)
ax.set_title((title + '\n' + t0 + ' - ' + t1),
fontsize=9)
pf.save_fig(save_dir, sname)
# Plot data with outliers removed
fig, ax = pf.plot_xsection(subsite,
t_nofv_nonan_noev_nogr,
y_nofv_nonan_noev_nogr,
z_nofv_nonan_noev_nogr,
clabel,
ylabel,
stdev=5)
ax.set_title((title + '\n' + t0 + ' - ' + t1),
fontsize=9)
sfile = '_'.join((sname, 'rmoutliers'))
pf.save_fig(save_dir, sfile)
# plot data with excluded time range removed
dr = pd.read_csv(
'https://datareview.marine.rutgers.edu/notes/export'
)
drn = dr.loc[dr.type == 'exclusion']
if len(drn) != 0:
subsite_node = '-'.join((subsite, r.split('-')[1]))
drne = drn.loc[drn.reference_designator.isin(
[subsite, subsite_node, r])]
t_ex = t_nofv_nonan_noev_nogr
y_ex = y_nofv_nonan_noev_nogr
z_ex = z_nofv_nonan_noev_nogr
for i, row in drne.iterrows():
sdate = cf.format_dates(row.start_date)
edate = cf.format_dates(row.end_date)
ts = np.datetime64(sdate)
te = np.datetime64(edate)
ind = np.where((t_ex < ts) | (t_ex > te), True,
False)
if len(ind) != 0:
t_ex = t_ex[ind]
z_ex = z_ex[ind]
y_ex = y_ex[ind]
fig, ax = pf.plot_xsection(subsite,
t_ex,
y_ex,
z_ex,
clabel,
ylabel,
stdev=None)
ax.set_title((title + '\n' + t0 + ' - ' + t1),
fontsize=9)
sfile = '_'.join((sname, 'rmsuspectdata'))
pf.save_fig(save_dir, sfile)
def main(sDir, plotting_sDir, url_list, sd_calc):
dr = pd.read_csv('https://datareview.marine.rutgers.edu/notes/export')
drn = dr.loc[dr.type == 'exclusion']
rd_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
if rd not in rd_list:
rd_list.append(rd)
for r in rd_list:
print('\n{}'.format(r))
datasets = []
for u in url_list:
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join(
(splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
fdatasets = []
# get the preferred stream information
ps_df, n_streams = cf.get_preferred_stream_info(r)
pms = []
for index, row in ps_df.iterrows():
for ii in range(n_streams):
try:
rms = '-'.join((r, row[ii]))
pms.append(row[ii])
except TypeError:
continue
for dd in datasets:
spl = dd.split('/')[-2].split('-')
catalog_rms = '-'.join(
(spl[1], spl[2], spl[3], spl[4], spl[5], spl[6]))
fdeploy = dd.split('/')[-1].split('_')[0]
if rms == catalog_rms and fdeploy == row['deployment']:
fdatasets.append(dd)
main_sensor = r.split('-')[-1]
fdatasets_sel = cf.filter_collocated_instruments(
main_sensor, fdatasets)
# find time ranges to exclude from analysis for data review database
subsite = r.split('-')[0]
subsite_node = '-'.join((subsite, r.split('-')[1]))
drne = drn.loc[drn.reference_designator.isin(
[subsite, subsite_node, r])]
et = []
for i, row in drne.iterrows():
sdate = cf.format_dates(row.start_date)
edate = cf.format_dates(row.end_date)
et.append([sdate, edate])
# get science variable long names from the Data Review Database
stream_sci_vars = cd.sci_var_long_names(r)
# check if the science variable long names are the same for each stream
sci_vars_dict = cd.sci_var_long_names_check(stream_sci_vars)
# get the preferred stream information
ps_df, n_streams = cf.get_preferred_stream_info(r)
# build dictionary of science data from the preferred dataset for each deployment
print('\nAppending data from files')
sci_vars_dict, pressure_unit, pressure_name = cd.append_science_data(
ps_df, n_streams, r, fdatasets_sel, sci_vars_dict, et)
# analyze combined dataset
print('\nAnalyzing combined dataset and writing summary file')
array = subsite[0:2]
save_dir = os.path.join(sDir, array, subsite)
cf.create_dir(save_dir)
rows = []
if ('FLM' in r) and (
'CTDMO' in r
): # calculate Flanking Mooring CTDMO stats based on pressure
headers = [
'common_stream_name', 'preferred_methods_streams',
'deployments', 'long_name', 'units', 't0', 't1', 'fill_value',
'global_ranges', 'n_all', 'press_min_max',
'n_excluded_forpress', 'n_nans', 'n_fillvalues', 'n_grange',
'define_stdev', 'n_outliers', 'n_stats', 'mean', 'min', 'max',
'stdev', 'note'
]
else:
headers = [
'common_stream_name', 'preferred_methods_streams',
'deployments', 'long_name', 'units', 't0', 't1', 'fill_value',
'global_ranges', 'n_all', 'n_nans', 'n_fillvalues', 'n_grange',
'define_stdev', 'n_outliers', 'n_stats', 'mean', 'min', 'max',
'stdev'
]
for m, n in sci_vars_dict.items():
print('\nSTREAM: ', m)
if m == 'common_stream_placeholder':
m = 'science_data_stream'
if m == 'metbk_hourly': # don't calculate ranges for metbk_hourly
continue
if ('FLM' in r) and (
'CTDMO' in r
): # calculate Flanking Mooring CTDMO stats based on pressure
# index the pressure variable to filter and calculate stats on the rest of the variables
sv_press = 'Seawater Pressure'
vinfo_press = n['vars'][sv_press]
# first, index where data are nans, fill values, and outside of global ranges
fv_press = list(np.unique(vinfo_press['fv']))[0]
pdata = vinfo_press['values']
[pind, __, __, __, __,
__] = index_dataset(r, vinfo_press['var_name'], pdata,
fv_press)
pdata_filtered = pdata[pind]
[__, pmean, __, __, psd,
__] = cf.variable_statistics(pdata_filtered, None)
# index of pressure = average of all 'valid' pressure data +/- 1 SD
ipress_min = pmean - psd
ipress_max = pmean + psd
ind_press = (pdata >= ipress_min) & (pdata <= ipress_max)
# calculate stats for all variables
print('\nPARAMETERS:')
for sv, vinfo in n['vars'].items():
print(sv)
fv_lst = np.unique(vinfo['fv']).tolist()
if len(fv_lst) == 1:
fill_value = fv_lst[0]
else:
print('No unique fill value for {}'.format(sv))
lunits = np.unique(vinfo['units']).tolist()
n_all = len(vinfo['t'])
# filter data based on pressure index
t_filtered = vinfo['t'][ind_press]
data_filtered = vinfo['values'][ind_press]
deploy_filtered = vinfo['deployments'][ind_press]
n_excluded = n_all - len(t_filtered)
[dataind, g_min, g_max, n_nan, n_fv,
n_grange] = index_dataset(r, vinfo['var_name'],
data_filtered, fill_value)
t_final = t_filtered[dataind]
data_final = data_filtered[dataind]
deploy_final = deploy_filtered[dataind]
t0 = pd.to_datetime(
min(t_final)).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(
max(t_final)).strftime('%Y-%m-%dT%H:%M:%S')
deploy = list(np.unique(deploy_final))
deployments = [int(dd) for dd in deploy]
if len(data_final) > 1:
[num_outliers, mean, vmin, vmax, sd, n_stats
] = cf.variable_statistics(data_final, sd_calc)
else:
mean = None
vmin = None
vmax = None
sd = None
n_stats = None
note = 'restricted stats calculation to data points where pressure is within defined ranges' \
' (average of all pressure data +/- 1 SD)'
rows.append([
m,
list(np.unique(pms)), deployments, sv, lunits, t0, t1,
fv_lst, [g_min, g_max], n_all,
[round(ipress_min, 2),
round(ipress_max,
2)], n_excluded, n_nan, n_fv, n_grange, sd_calc,
num_outliers, n_stats, mean, vmin, vmax, sd, note
])
# plot CTDMO data used for stats
psave_dir = os.path.join(plotting_sDir, array, subsite, r,
'timeseries_plots_stats')
cf.create_dir(psave_dir)
dr_data = cf.refdes_datareview_json(r)
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = cf.get_deployment_information(
dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(
pd.to_datetime(deploy_info['stop_date']))
sname = '-'.join((r, sv))
fig, ax = pf.plot_timeseries_all(t_final,
data_final,
sv,
lunits[0],
stdev=None)
ax.set_title(
(r + '\nDeployments: ' + str(sorted(deployments)) +
'\n' + t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes,
color='k',
linestyle='--',
linewidth=.6)
pf.save_fig(psave_dir, sname)
if sd_calc:
sname = '-'.join((r, sv, 'rmoutliers'))
fig, ax = pf.plot_timeseries_all(t_final,
data_final,
sv,
lunits[0],
stdev=sd_calc)
ax.set_title(
(r + '\nDeployments: ' + str(sorted(deployments)) +
'\n' + t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes,
color='k',
linestyle='--',
linewidth=.6)
pf.save_fig(psave_dir, sname)
else:
if not sd_calc:
sdcalc = None
print('\nPARAMETERS: ')
for sv, vinfo in n['vars'].items():
print(sv)
fv_lst = np.unique(vinfo['fv']).tolist()
if len(fv_lst) == 1:
fill_value = fv_lst[0]
else:
print(fv_lst)
print('No unique fill value for {}'.format(sv))
lunits = np.unique(vinfo['units']).tolist()
t = vinfo['t']
if len(t) > 1:
data = vinfo['values']
n_all = len(t)
if 'SPKIR' in r or 'presf_abc_wave_burst' in m:
if 'SPKIR' in r:
[dd_data, g_min, g_max, n_nan, n_fv,
n_grange] = index_dataset_2d(
r, 'spkir_abj_cspp_downwelling_vector',
data, fill_value)
else:
[dd_data, g_min, g_max, n_nan, n_fv,
n_grange] = index_dataset_2d(
r, 'presf_wave_burst_pressure', data,
fill_value)
t_final = t
t0 = pd.to_datetime(
min(t_final)).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(
max(t_final)).strftime('%Y-%m-%dT%H:%M:%S')
deploy_final = vinfo['deployments']
deploy = list(np.unique(deploy_final))
deployments = [int(dd) for dd in deploy]
num_outliers = []
mean = []
vmin = []
vmax = []
sd = []
n_stats = []
for i in range(len(dd_data)):
dd = data[i]
# drop nans before calculating stats
dd = dd[~np.isnan(dd)]
[
num_outliersi, meani, vmini, vmaxi, sdi,
n_statsi
] = cf.variable_statistics(dd, sd_calc)
num_outliers.append(num_outliersi)
mean.append(meani)
vmin.append(vmini)
vmax.append(vmaxi)
sd.append(sdi)
n_stats.append(n_statsi)
else:
[dataind, g_min, g_max, n_nan, n_fv,
n_grange] = index_dataset(r, vinfo['var_name'],
data, fill_value)
t_final = t[dataind]
if len(t_final) > 0:
t0 = pd.to_datetime(
min(t_final)).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(
max(t_final)).strftime('%Y-%m-%dT%H:%M:%S')
data_final = data[dataind]
# if sv == 'Dissolved Oxygen Concentration':
# xx = (data_final > 0) & (data_final < 400)
# data_final = data_final[xx]
# t_final = t_final[xx]
# if sv == 'Seawater Conductivity':
# xx = (data_final > 1) & (data_final < 400)
# data_final = data_final[xx]
# t_final = t_final[xx]
deploy_final = vinfo['deployments'][dataind]
deploy = list(np.unique(deploy_final))
deployments = [int(dd) for dd in deploy]
if len(data_final) > 1:
[
num_outliers, mean, vmin, vmax, sd,
n_stats
] = cf.variable_statistics(
data_final, sd_calc)
else:
sdcalc = None
num_outliers = None
mean = None
vmin = None
vmax = None
sd = None
n_stats = None
else:
sdcalc = None
num_outliers = None
mean = None
vmin = None
vmax = None
sd = None
n_stats = None
deployments = None
t0 = None
t1 = None
else:
sdcalc = None
num_outliers = None
mean = None
vmin = None
vmax = None
sd = None
n_stats = None
deployments = None
t0 = None
t1 = None
t_final = []
if sd_calc:
print_sd = sd_calc
else:
print_sd = sdcalc
rows.append([
m,
list(np.unique(pms)), deployments, sv, lunits, t0, t1,
fv_lst, [g_min, g_max], n_all, n_nan, n_fv, n_grange,
print_sd, num_outliers, n_stats, mean, vmin, vmax, sd
])
if len(t_final) > 0:
# plot data used for stats
psave_dir = os.path.join(
plotting_sDir, array, subsite, r,
'timeseries_reviewed_datarange')
cf.create_dir(psave_dir)
dr_data = cf.refdes_datareview_json(r)
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = cf.get_deployment_information(
dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(
pd.to_datetime(deploy_info['stop_date']))
sname = '-'.join((r, sv))
# plot hourly averages for streaming data
if 'streamed' in sci_vars_dict[list(
sci_vars_dict.keys())[0]]['ms'][0]:
sname = '-'.join((sname, 'hourlyavg'))
df = pd.DataFrame({
'dfx': t_final,
'dfy': data_final
})
dfr = df.resample('H', on='dfx').mean()
# Plot all data
fig, ax = pf.plot_timeseries_all(dfr.index,
dfr['dfy'],
sv,
lunits[0],
stdev=None)
ax.set_title((r + '\nDeployments: ' +
str(sorted(deployments)) + '\n' +
t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes,
color='k',
linestyle='--',
linewidth=.6)
pf.save_fig(psave_dir, sname)
if sd_calc:
sname = '-'.join(
(sname, 'hourlyavg_rmoutliers'))
fig, ax = pf.plot_timeseries_all(dfr.index,
dfr['dfy'],
sv,
lunits[0],
stdev=sd_calc)
ax.set_title((r + '\nDeployments: ' +
str(sorted(deployments)) + '\n' +
t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes,
color='k',
linestyle='--',
linewidth=.6)
pf.save_fig(psave_dir, sname)
elif 'SPKIR' in r:
fig, ax = pf.plot_spkir(t_final, dd_data, sv,
lunits[0])
ax.set_title((r + '\nDeployments: ' +
str(sorted(deployments)) + '\n' +
t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes,
color='k',
linestyle='--',
linewidth=.6)
pf.save_fig(psave_dir, sname)
# plot each wavelength
wavelengths = [
'412nm', '443nm', '490nm', '510nm', '555nm',
'620nm', '683nm'
]
for wvi in range(len(dd_data)):
fig, ax = pf.plot_spkir_wv(
t_final, dd_data[wvi], sv, lunits[0], wvi)
ax.set_title((r + '\nDeployments: ' +
str(sorted(deployments)) + '\n' +
t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes,
color='k',
linestyle='--',
linewidth=.6)
snamewvi = '-'.join((sname, wavelengths[wvi]))
pf.save_fig(psave_dir, snamewvi)
elif 'presf_abc_wave_burst' in m:
fig, ax = pf.plot_presf_2d(t_final, dd_data, sv,
lunits[0])
ax.set_title((r + '\nDeployments: ' +
str(sorted(deployments)) + '\n' +
t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes,
color='k',
linestyle='--',
linewidth=.6)
snamewave = '-'.join((sname, m))
pf.save_fig(psave_dir, snamewave)
else: # plot all data if not streamed
fig, ax = pf.plot_timeseries_all(t_final,
data_final,
sv,
lunits[0],
stdev=None)
ax.set_title((r + '\nDeployments: ' +
str(sorted(deployments)) + '\n' +
t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes,
color='k',
linestyle='--',
linewidth=.6)
pf.save_fig(psave_dir, sname)
if sd_calc:
sname = '-'.join((r, sv, 'rmoutliers'))
fig, ax = pf.plot_timeseries_all(t_final,
data_final,
sv,
lunits[0],
stdev=sd_calc)
ax.set_title((r + '\nDeployments: ' +
str(sorted(deployments)) + '\n' +
t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes,
color='k',
linestyle='--',
linewidth=.6)
pf.save_fig(psave_dir, sname)
fsum = pd.DataFrame(rows, columns=headers)
fsum.to_csv('{}/{}_data_ranges.csv'.format(save_dir, r), index=False)
def main(sDir, ncdir):
rd_list = [ncdir.split('/')[-2]]
for r in rd_list:
print('\n{}'.format(r))
subsite = r.split('-')[0]
array = subsite[0:2]
ps_df, n_streams = cf.get_preferred_stream_info(r)
# get end times of deployments
dr_data = cf.refdes_datareview_json(r)
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = get_deployment_information(dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
# filter datasets
fdatasets = []
for root, dirs, files in os.walk(ncdir):
for f in files:
if f.endswith('.nc'):
fdatasets.append(f)
# for u in url_list:
# splitter = u.split('/')[-2].split('-')
# rd_check = '-'.join((splitter[1], splitter[2], splitter[3], splitter[4]))
# if rd_check == r:
# udatasets = cf.get_nc_urls([u])
# datasets.append(udatasets)
# datasets = list(itertools.chain(*datasets))
# main_sensor = r.split('-')[-1]
# fdatasets = cf.filter_collocated_instruments(main_sensor, datasets)
methodstream = []
for f in fdatasets:
strm = '_'.join((f.split('-')[-2].split('_')[0], f.split('-')[-2].split('_')[1]))
methodstream.append('-'.join((f.split('-')[-3], strm)))
for ms in np.unique(methodstream):
fdatasets_sel = [x for x in fdatasets if ms in x]
save_dir = os.path.join(sDir, array, subsite, r, 'timeseries_plots_all')
cf.create_dir(save_dir)
stream_sci_vars_dict = dict()
for x in dr_data['instrument']['data_streams']:
dr_ms = '-'.join((x['method'], x['stream_name']))
if ms == dr_ms:
stream_sci_vars_dict[dr_ms] = dict(vars=dict())
sci_vars = dict()
for y in x['stream']['parameters']:
if y['data_product_type'] == 'Science Data':
sci_vars.update({y['name']: dict(db_units=y['unit'])})
if len(sci_vars) > 0:
stream_sci_vars_dict[dr_ms]['vars'] = sci_vars
sci_vars_dict = cd.initialize_empty_arrays(stream_sci_vars_dict, ms)
print('\nAppending data from files: {}'.format(ms))
for fd in fdatasets_sel:
ds = xr.open_dataset(os.path.join(ncdir, fd), mask_and_scale=False)
for var in list(sci_vars_dict[ms]['vars'].keys()):
sh = sci_vars_dict[ms]['vars'][var]
if ds[var].units == sh['db_units']:
if ds[var]._FillValue not in sh['fv']:
sh['fv'].append(ds[var]._FillValue)
if ds[var].units not in sh['units']:
sh['units'].append(ds[var].units)
tD = ds['time'].values
varD = ds[var].values
sh['t'] = np.append(sh['t'], tD)
sh['values'] = np.append(sh['values'], varD)
print('\nPlotting data')
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print(sv)
if len(vinfo['t']) < 1:
print('no variable data to plot')
else:
sv_units = vinfo['units'][0]
t0 = pd.to_datetime(min(vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
x = vinfo['t']
y = vinfo['values']
# reject NaNs
nan_ind = ~np.isnan(y)
x_nonan = x[nan_ind]
y_nonan = y[nan_ind]
# reject fill values
fv_ind = y_nonan != vinfo['fv'][0]
x_nonan_nofv = x_nonan[fv_ind]
y_nonan_nofv = y_nonan[fv_ind]
# reject extreme values
Ev_ind = cf.reject_extreme_values(y_nonan_nofv)
y_nonan_nofv_nE = y_nonan_nofv[Ev_ind]
x_nonan_nofv_nE = x_nonan_nofv[Ev_ind]
# reject values outside global ranges:
global_min, global_max = cf.get_global_ranges(r, sv)
if global_min is not None and global_max is not None:
gr_ind = cf.reject_global_ranges(y_nonan_nofv_nE, global_min, global_max)
y_nonan_nofv_nE_nogr = y_nonan_nofv_nE[gr_ind]
x_nonan_nofv_nE_nogr = x_nonan_nofv_nE[gr_ind]
else:
y_nonan_nofv_nE_nogr = y_nonan_nofv_nE
x_nonan_nofv_nE_nogr = x_nonan_nofv_nE
title = ' '.join((r, ms.split('-')[0]))
if len(y_nonan_nofv) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((r, sv))
else:
sname = '-'.join((r, m, sv))
# Plot all data
fig, ax = pf.plot_timeseries_all(x_nonan_nofv, y_nonan_nofv, sv, sv_units, stdev=None)
ax.set_title((title + '\nDeployments: ' + str(sorted(deployments)) + '\n' + t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes, color='b', linestyle='--', linewidth=.6)
# if global_min is not None and global_max is not None:
# ax.axhline(y=global_min, color='r', linestyle='--', linewidth=.6)
# ax.axhline(y=global_max, color='r', linestyle='--', linewidth=.6)
pf.save_fig(save_dir, sname)
# Plot data with extreme values, data outside global ranges and outliers removed
fig, ax = pf.plot_timeseries_all(x_nonan_nofv_nE_nogr, y_nonan_nofv_nE_nogr, sv, sv_units, stdev=5)
ax.set_title((title + '\nDeployments: ' + str(sorted(deployments)) + '\n' + t0 + ' - ' + t1),
fontsize=8)
for etimes in end_times:
ax.axvline(x=etimes, color='b', linestyle='--', linewidth=.6)
# if global_min is not None and global_max is not None:
# ax.axhline(y=global_min, color='r', linestyle='--', linewidth=.6)
# ax.axhline(y=global_max, color='r', linestyle='--', linewidth=.6)
sfile = '_'.join((sname, 'rmoutliers'))
pf.save_fig(save_dir, sfile)
def main(sDir, url_list, preferred_only):
rd_list = []
ms_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
ms = uu.split(rd + '-')[1].split('/')[0]
if rd not in rd_list:
rd_list.append(rd)
if ms not in ms_list:
ms_list.append(ms)
for r in rd_list:
print('\n{}'.format(r))
subsite = r.split('-')[0]
array = subsite[0:2]
# filter datasets
datasets = []
for u in url_list:
print(u)
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join(
(splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
fdatasets = []
if preferred_only == 'yes':
# get the preferred stream information
ps_df, n_streams = cf.get_preferred_stream_info(r)
for index, row in ps_df.iterrows():
for ii in range(n_streams):
try:
rms = '-'.join((r, row[ii]))
except TypeError:
continue
for dd in datasets:
spl = dd.split('/')[-2].split('-')
catalog_rms = '-'.join(
(spl[1], spl[2], spl[3], spl[4], spl[5], spl[6]))
fdeploy = dd.split('/')[-1].split('_')[0]
if rms == catalog_rms and fdeploy == row['deployment']:
fdatasets.append(dd)
else:
fdatasets = datasets
main_sensor = r.split('-')[-1]
fdatasets = cf.filter_collocated_instruments(main_sensor, fdatasets)
# ps_df, n_streams = cf.get_preferred_stream_info(r)
# get end times of deployments
dr_data = cf.refdes_datareview_json(r)
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = get_deployment_information(dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
# # filter datasets
# datasets = []
# for u in url_list:
# print(u)
# splitter = u.split('/')[-2].split('-')
# rd_check = '-'.join((splitter[1], splitter[2], splitter[3], splitter[4]))
# if rd_check == r:
# udatasets = cf.get_nc_urls([u])
# datasets.append(udatasets)
# datasets = list(itertools.chain(*datasets))
# main_sensor = r.split('-')[-1]
# fdatasets = cf.filter_collocated_instruments(main_sensor, datasets)
# fdatasets = cf.filter_other_streams(r, ms_list, fdatasets)
methodstream = []
for f in fdatasets:
methodstream.append('-'.join((f.split('/')[-2].split('-')[-2],
f.split('/')[-2].split('-')[-1])))
ms_dict = save_dir_path(ms_list)
for ms in np.unique(methodstream):
fdatasets_sel = [x for x in fdatasets if ms in x]
check_ms = ms.split('-')[1]
if 'recovered' in check_ms:
check_ms = check_ms.split('_recovered')[0]
if ms_dict['ms_count'][ms_dict['ms_unique'] == ms.split('-')
[0]] == 1:
save_dir = os.path.join(sDir, array, subsite, r,
'timeseries_yearly_plot',
ms.split('-')[0])
else:
save_dir = os.path.join(sDir, array, subsite, r,
'timeseries_yearly_plot',
ms.split('-')[0], check_ms)
cf.create_dir(save_dir)
stream_sci_vars_dict = dict()
for x in dr_data['instrument']['data_streams']:
dr_ms = '-'.join((x['method'], x['stream_name']))
if ms == dr_ms:
stream_sci_vars_dict[dr_ms] = dict(vars=dict())
sci_vars = dict()
for y in x['stream']['parameters']:
if y['data_product_type'] == 'Science Data':
sci_vars.update(
{y['name']: dict(db_units=y['unit'])})
if len(sci_vars) > 0:
stream_sci_vars_dict[dr_ms]['vars'] = sci_vars
sci_vars_dict = cd.initialize_empty_arrays(stream_sci_vars_dict,
ms)
print('\nAppending data from files: {}'.format(ms))
for fd in fdatasets_sel:
ds = xr.open_dataset(fd, mask_and_scale=False)
print(fd)
for var in list(sci_vars_dict[ms]['vars'].keys()):
sh = sci_vars_dict[ms]['vars'][var]
try:
ds[var]
print(var)
deployment_num = np.unique(ds['deployment'].values)[0]
sh['deployments'] = np.append(sh['deployments'],
deployment_num)
if ds[var].units == sh['db_units']:
if ds[var]._FillValue not in sh['fv']:
sh['fv'].append(ds[var]._FillValue)
if ds[var].units not in sh['units']:
sh['units'].append(ds[var].units)
tD = ds['time'].values
varD = ds[var].values
sh['t'] = np.append(sh['t'], tD)
sh['values'] = np.append(sh['values'], varD)
except KeyError:
print('KeyError: ', var)
print('\nPlotting data')
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print(sv)
if len(vinfo['t']) < 1:
print('no variable data to plot')
else:
sv_units = vinfo['units'][0]
deployments_num = vinfo['deployments']
fv = vinfo['fv'][0]
t0 = pd.to_datetime(min(
vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(
vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
x = vinfo['t']
y = vinfo['values']
# reject NaNs
nan_ind = ~np.isnan(y)
x_nonan = x[nan_ind]
y_nonan = y[nan_ind]
# reject fill values
fv_ind = y_nonan != vinfo['fv'][0]
x_nonan_nofv = x_nonan[fv_ind]
y_nonan_nofv = y_nonan[fv_ind]
# reject extreme values
Ev_ind = cf.reject_extreme_values(y_nonan_nofv)
y_nonan_nofv_nE = y_nonan_nofv[Ev_ind]
x_nonan_nofv_nE = x_nonan_nofv[Ev_ind]
# reject values outside global ranges:
global_min, global_max = cf.get_global_ranges(r, sv)
print('global ranges: ', global_min, global_max)
if global_min and global_max:
gr_ind = cf.reject_global_ranges(
y_nonan_nofv_nE, global_min, global_max)
y_nonan_nofv_nE_nogr = y_nonan_nofv_nE[gr_ind]
x_nonan_nofv_nE_nogr = x_nonan_nofv_nE[gr_ind]
else:
y_nonan_nofv_nE_nogr = y_nonan_nofv_nE
x_nonan_nofv_nE_nogr = x_nonan_nofv_nE
# check array length
if len(y_nonan_nofv_nE_nogr) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((r, sv))
print(var, 'empty array')
else:
sname = '-'.join((r, m, sv))
# group data by year
groups, g_data = gt.group_by_time_range(
x_nonan_nofv_nE_nogr, y_nonan_nofv_nE_nogr,
'A')
# create bins
# groups_min = min(groups.describe()['DO']['min'])
# lower_bound = int(round(groups_min))
# groups_max = max(groups.describe()['DO']['max'])
# if groups_max < 1:
# upper_bound = 1
# step_bound = 1
# else:
# upper_bound = int(round(groups_max + (groups_max / 50)))
# step_bound = int(round((groups_max - groups_min) / 10))
#
# if step_bound == 0:
# step_bound += 1
#
# if (upper_bound - lower_bound) == step_bound:
# lower_bound -= 1
# upper_bound += 1
# if (upper_bound - lower_bound) < step_bound:
# print('<')
# step_bound = int(round(step_bound / 10))
# print(lower_bound, upper_bound, step_bound)
# bin_range = list(range(lower_bound, upper_bound, step_bound))
# print(bin_range)
# preparing color palette
colors = color_names[:len(groups)]
# colors = [color['color'] for color in
# list(pyplot.rcParams['axes.prop_cycle'][:len(groups)])]
fig0, ax0 = pyplot.subplots(nrows=2, ncols=1)
# subplot for histogram and basic statistics table
ax0[1].axis('off')
ax0[1].axis('tight')
the_table = ax0[1].table(
cellText=groups.describe().round(2).values,
rowLabels=groups.describe().index.year,
rowColours=colors,
colLabels=groups.describe().columns.levels[1],
loc='center')
the_table.set_fontsize(5)
# subplot for data
fig, ax = pyplot.subplots(nrows=len(groups),
ncols=1,
sharey=True)
if len(groups) == 1:
ax = [ax]
t = 1
for ny in range(len(groups)):
# prepare data for plotting
y_data = g_data[ny + (t + 1)].dropna(axis=0)
x_time = g_data[ny + t].dropna(axis=0)
t += 1
if len(y_data) != 0 and len(x_time) != 0:
n_year = x_time[0].year
col_name = str(n_year)
serie_n = pd.DataFrame(columns=[col_name],
index=x_time)
serie_n[col_name] = list(y_data[:])
# plot histogram
# serie_n.plot.hist(ax=ax0[0], bins=bin_range,
# histtype='bar', color=colors[ny], stacked=True)
if len(serie_n) != 1:
serie_n.plot.kde(ax=ax0[0],
color=colors[ny])
ax0[0].legend(fontsize=8,
bbox_to_anchor=(0., 1.12,
1.,
.102),
loc=3,
ncol=len(groups),
mode="expand",
borderaxespad=0.)
# ax0[0].set_xticks(bin_range)
ax0[0].set_xlabel('Observation Ranges',
fontsize=8)
ax0[0].set_ylabel(
'Density', fontsize=8
) #'Number of Observations'
ax0[0].set_title(
ms.split('-')[0] + ' (' + sv +
', ' + sv_units + ')' +
' Kernel Density Estimates',
fontsize=8)
# plot data
serie_n.plot(ax=ax[ny],
linestyle='None',
marker='.',
markersize=0.5,
color=colors[ny])
ax[ny].legend().set_visible(False)
# plot Mean and Standard deviation
ma = serie_n.rolling('86400s').mean()
mstd = serie_n.rolling('86400s').std()
ax[ny].plot(ma.index,
ma[col_name].values,
'k',
linewidth=0.15)
ax[ny].fill_between(
mstd.index,
ma[col_name].values -
2 * mstd[col_name].values,
ma[col_name].values +
2 * mstd[col_name].values,
color='b',
alpha=0.2)
# prepare the time axis parameters
datemin = datetime.date(n_year, 1, 1)
datemax = datetime.date(n_year, 12, 31)
ax[ny].set_xlim(datemin, datemax)
xlocator = mdates.MonthLocator(
) # every month
myFmt = mdates.DateFormatter('%m')
ax[ny].xaxis.set_minor_locator(
xlocator)
ax[ny].xaxis.set_major_formatter(myFmt)
# prepare the time axis parameters
# ax[ny].set_yticks(bin_range)
ylocator = MaxNLocator(prune='both',
nbins=3)
ax[ny].yaxis.set_major_locator(
ylocator)
# format figure
ax[ny].tick_params(axis='both',
color='r',
labelsize=7,
labelcolor='m')
if ny < len(groups) - 1:
ax[ny].tick_params(
which='both',
pad=0.1,
length=1,
labelbottom=False)
ax[ny].set_xlabel(' ')
else:
ax[ny].tick_params(which='both',
color='r',
labelsize=7,
labelcolor='m',
pad=0.1,
length=1,
rotation=0)
ax[ny].set_xlabel('Months',
rotation=0,
fontsize=8,
color='b')
ax[ny].set_ylabel(n_year,
rotation=0,
fontsize=8,
color='b',
labelpad=20)
ax[ny].yaxis.set_label_position(
"right")
if ny == 0:
if global_min and global_max:
ax[ny].set_title(
sv + '( ' + sv_units +
') -- Global Range: [' +
str(int(global_min)) +
',' +
str(int(global_max)) +
'] \n'
'Plotted: Data, Mean and 2STD (Method: One day rolling window calculations) \n',
fontsize=8)
else:
ax[ny].set_title(
sv + '( ' + sv_units +
') -- Global Range: [] \n'
'Plotted: Data, Mean and 2STD (Method: One day rolling window calculations) \n',
fontsize=8)
# plot global ranges
# ax[ny].axhline(y=global_min, color='r', linestyle='--', linewidth=.6)
# ax[ny].axhline(y=global_max, color='r', linestyle='--', linewidth=.6)
# mark deployment end times on figure
ymin, ymax = ax[ny].get_ylim()
#dep = 1
for etimes in range(len(end_times)):
if end_times[
etimes].year == n_year:
ax[ny].axvline(
x=end_times[etimes],
color='b',
linestyle='--',
linewidth=.6)
ax[ny].text(
end_times[etimes],
ymin,
'End' +
str(deployments_num[etimes]
),
fontsize=6,
style='italic',
bbox=dict(boxstyle='round',
ec=(0., 0.5,
0.5),
fc=(1., 1., 1.)))
# dep += 1
# ax[ny].set_ylim(5, 12)
# save figure to a file
sfile = '_'.join(('all', sname))
save_file = os.path.join(save_dir, sfile)
fig.savefig(str(save_file), dpi=150)
sfile = '_'.join(('Statistics', sname))
save_file = os.path.join(save_dir, sfile)
fig0.savefig(str(save_file), dpi=150)
pyplot.close()
def main(url_list, sDir, plot_type, deployment_num, start_time, end_time):
"""""
URL : path to instrument data by methods
sDir : path to the directory on your machine to save files
plot_type: folder name for a plot type
""" ""
rd_list = []
ms_list = []
for uu in url_list:
elements = uu.split('/')[-2].split('-')
rd = '-'.join((elements[1], elements[2], elements[3], elements[4]))
ms = uu.split(rd + '-')[1].split('/')[0]
if rd not in rd_list:
rd_list.append(rd)
if ms not in ms_list:
ms_list.append(ms)
'''
separate different instruments
'''
for r in rd_list:
print('\n{}'.format(r))
subsite = r.split('-')[0]
array = subsite[0:2]
main_sensor = r.split('-')[-1]
ps_df, n_streams = cf.get_preferred_stream_info(r)
# read in the analysis file
dr_data = cf.refdes_datareview_json(r)
# get end times of deployments
deployments = []
end_times = []
for index, row in ps_df.iterrows():
deploy = row['deployment']
deploy_info = get_deployment_information(dr_data, int(deploy[-4:]))
deployments.append(int(deploy[-4:]))
end_times.append(pd.to_datetime(deploy_info['stop_date']))
# get the list of data files and filter out collocated instruments and other streams chat
datasets = []
for u in url_list:
print(u)
splitter = u.split('/')[-2].split('-')
rd_check = '-'.join(
(splitter[1], splitter[2], splitter[3], splitter[4]))
if rd_check == r:
udatasets = cf.get_nc_urls([u])
datasets.append(udatasets)
datasets = list(itertools.chain(*datasets))
fdatasets = cf.filter_collocated_instruments(main_sensor, datasets)
fdatasets = cf.filter_other_streams(r, ms_list, fdatasets)
'''
separate the data files by methods
'''
for ms in ms_list:
fdatasets_sel = [x for x in fdatasets if ms in x]
# create a dictionary for science variables from analysis file
stream_sci_vars_dict = dict()
for x in dr_data['instrument']['data_streams']:
dr_ms = '-'.join((x['method'], x['stream_name']))
if ms == dr_ms:
stream_sci_vars_dict[dr_ms] = dict(vars=dict())
sci_vars = dict()
for y in x['stream']['parameters']:
if y['data_product_type'] == 'Science Data':
sci_vars.update(
{y['name']: dict(db_units=y['unit'])})
if len(sci_vars) > 0:
stream_sci_vars_dict[dr_ms]['vars'] = sci_vars
# initialize an empty data array for science variables in dictionary
sci_vars_dict = cd.initialize_empty_arrays(stream_sci_vars_dict,
ms)
print('\nAppending data from files: {}'.format(ms))
y_unit = []
y_name = []
for fd in fdatasets_sel:
ds = xr.open_dataset(fd, mask_and_scale=False)
print(fd)
if start_time is not None and end_time is not None:
ds = ds.sel(time=slice(start_time, end_time))
if len(ds['time'].values) == 0:
print(
'No data to plot for specified time range: ({} to {})'
.format(start_time, end_time))
continue
fname, subsite, refdes, method, stream, deployment = cf.nc_attributes(
fd)
if deployment_num is not None:
if int(deployment.split('0')[-1]) is not deployment_num:
print(type(int(deployment.split('0')[-1])),
type(deployment_num))
continue
save_dir = os.path.join(sDir, array, subsite, refdes,
plot_type,
ms.split('-')[0], deployment)
cf.create_dir(save_dir)
for var in list(sci_vars_dict[ms]['vars'].keys()):
sh = sci_vars_dict[ms]['vars'][var]
if ds[var].units == sh['db_units']:
if ds[var]._FillValue not in sh['fv']:
sh['fv'].append(ds[var]._FillValue)
if ds[var].units not in sh['units']:
sh['units'].append(ds[var].units)
# time
t = ds['time'].values
t0 = pd.to_datetime(
t.min()).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(
t.max()).strftime('%Y-%m-%dT%H:%M:%S')
# sci variable
z = ds[var].values
sh['t'] = np.append(sh['t'], t)
sh['values'] = np.append(sh['values'], z)
# add pressure to dictionary of sci vars
if 'MOAS' in subsite:
if 'CTD' in main_sensor: # for glider CTDs, pressure is a coordinate
pressure = 'sci_water_pressure_dbar'
y = ds[pressure].values
if ds[pressure].units not in y_unit:
y_unit.append(ds[pressure].units)
if ds[pressure].long_name not in y_name:
y_name.append(ds[pressure].long_name)
else:
pressure = 'int_ctd_pressure'
y = ds[pressure].values
if ds[pressure].units not in y_unit:
y_unit.append(ds[pressure].units)
if ds[pressure].long_name not in y_name:
y_name.append(ds[pressure].long_name)
else:
pressure = pf.pressure_var(ds, ds.data_vars.keys())
y = ds[pressure].values
if ds[pressure].units not in y_unit:
y_unit.append(ds[pressure].units)
if ds[pressure].long_name not in y_name:
y_name.append(ds[pressure].long_name)
sh['pressure'] = np.append(sh['pressure'], y)
if len(y_unit) != 1:
print('pressure unit varies UHHHHHHHHH')
else:
y_unit = y_unit[0]
if len(y_name) != 1:
print('pressure long name varies UHHHHHHHHH')
else:
y_name = y_name[0]
for m, n in sci_vars_dict.items():
for sv, vinfo in n['vars'].items():
print(sv)
if len(vinfo['t']) < 1:
print('no variable data to plot')
else:
sv_units = vinfo['units'][0]
fv = vinfo['fv'][0]
t0 = pd.to_datetime(min(
vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t1 = pd.to_datetime(max(
vinfo['t'])).strftime('%Y-%m-%dT%H:%M:%S')
t = vinfo['t']
z = vinfo['values']
y = vinfo['pressure']
title = ' '.join((r, ms.split('-')[1]))
# Check if the array is all NaNs
if sum(np.isnan(z)) == len(z):
print('Array of all NaNs - skipping plot.')
# Check if the array is all fill values
elif len(z[z != fv]) == 0:
print('Array of all fill values - skipping plot.')
else:
# reject fill values
fv_ind = z != fv
y_nofv = y[fv_ind]
t_nofv = t[fv_ind]
z_nofv = z[fv_ind]
print(len(z) - len(fv_ind), ' fill values')
# reject NaNs
nan_ind = ~np.isnan(z)
t_nofv_nonan = t_nofv[nan_ind]
y_nofv_nonan = y_nofv[nan_ind]
z_nofv_nonan = z_nofv[nan_ind]
print(len(z) - len(nan_ind), ' NaNs')
# reject extreme values
ev_ind = cf.reject_extreme_values(z_nofv_nonan)
t_nofv_nonan_noev = t_nofv_nonan[ev_ind]
colors = cm.rainbow(
np.linspace(0, 1, len(t_nofv_nonan_noev)))
y_nofv_nonan_noev = y_nofv_nonan[ev_ind]
z_nofv_nonan_noev = z_nofv_nonan[ev_ind]
print(
len(z) - len(ev_ind), ' Extreme Values',
'|1e7|')
if len(y_nofv_nonan_noev) > 0:
if m == 'common_stream_placeholder':
sname = '-'.join((r, sv))
else:
sname = '-'.join((r, m, sv))
# Plot all data
ylabel = y_name + " (" + y_unit + ")"
xlabel = sv + " (" + sv_units + ")"
clabel = 'Time'
clabel = sv + " (" + sv_units + ")"
fig, ax = pf.plot_profiles(z_nofv_nonan_noev,
y_nofv_nonan_noev,
colors,
xlabel,
ylabel,
stdev=None)
ax.set_title((
title + '\n' + str(deployment_num) + ': ' + t0 +
' - ' + t1 + '\n' +
'used bin = 2 dbar to calculate an average profile (black line) and 3-STD envelope (shaded area)'
),
fontsize=9)
# group by depth range
columns = ['time', 'pressure', str(sv)]
# ranges = [0, 50, 100, 200, 400, 600]
ranges = list(
range(int(round(min(y_nofv_nonan_noev))),
int(round(max(y_nofv_nonan_noev))), 1))
groups, d_groups = gt.group_by_depth_range(
t_nofv_nonan_noev, y_nofv_nonan_noev,
z_nofv_nonan_noev, columns, ranges)
# describe_file = '_'.join((sname, 'statistics.csv'))
# # groups.describe().to_csv(save_dir + '/' + describe_file)
ind = groups.describe()[sv]['mean'].notnull()
groups.describe()[sv][ind].to_csv(
'{}/{}_statistics.csv'.format(save_dir, sname),
index=True)
tm = 1
fig, ax = pyplot.subplots(nrows=2, ncols=1)
pyplot.margins(y=.08, x=.02)
pyplot.grid()
y_avg, n_avg, n_min, n_max, n0_std, n1_std, l_arr = [], [], [], [], [], [], []
for ii in range(len(groups)):
nan_ind = d_groups[ii + tm].notnull()
xtime = d_groups[ii + tm][nan_ind]
colors = cm.rainbow(np.linspace(0, 1, len(xtime)))
ypres = d_groups[ii + tm + 1][nan_ind]
nval = d_groups[ii + tm + 2][nan_ind]
tm += 2
# fig, ax = pf.plot_xsection(subsite, xtime, ypres, nval, clabel, ylabel, stdev=None)
# ax.set_title((title + '\n' + t0 + ' - ' + t1), fontsize=9)
# pf.plot_profiles(nval, ypres, colors, ylabel, clabel, stdev=None)
# ax.set_title((title + '\n' + t0 + ' - ' + t1), fontsize=9)
ind2 = cf.reject_outliers(nval, 5)
xD = nval[ind2]
yD = ypres[ind2]
nZ = colors[ind2]
outliers = str(len(nval) - len(xD))
leg_text = ('removed {} outliers (SD={})'.format(
outliers, stdev), )
ax.scatter(xD, yD, c=nZ, s=2, edgecolor='None')
ax.invert_yaxis()
ax.set_xlabel(clabel, fontsize=9)
ax.set_ylabel(ylabel, fontsize=9)
ax.legend(leg_text, loc='best', fontsize=6)
ax.set_title((title + '\n' + t0 + ' - ' + t1),
fontsize=9)
l_arr.append(
len(nval)
) # count of data to filter out small groups
y_avg.append(ypres.mean())
n_avg.append(nval.mean())
n_min.append(nval.min())
n_max.append(nval.max())
n0_std.append(nval.mean() + 3 * nval.std())
n1_std.append(nval.mean() - 3 * nval.std())
ax.plot(n_avg, y_avg, '-k')
# ax.plot(n_min, y_avg, '-b')
# ax.plot(n_max, y_avg, '-b')
ax.fill_betweenx(y_avg,
n0_std,
n1_std,
color='m',
alpha=0.2)
sfile = '_'.join((sname, 'statistics'))
pf.save_fig(save_dir, sfile)
