default='SONDE_,',
help='Remove,replace on sonde files')
parser.add_argument('sondes', nargs='+', help='Sonde files')
parser.add_argument('figpath', help='Path for output figure')
args = parser.parse_args()
modellabel = args.label
args.season = seasons[args.season]
mf = None
sf = None
sids = []
lats = []
tk = None
for path in args.sondes:
modpath = path.replace(*args.modreplace.split(','))
modf = pncopen(modpath, format='netcdf', addcf=False)
mt = modf.getTimes()
ismine = np.array([i for i, t in enumerate(mt) if t.month in args.season])
if ismine.size == 0:
continue
if tk is None:
if 'TSTEP' in modf.dimensions:
tk = 'TSTEP'
else:
tk = 'time'
modf = modf.sliceDimensions(**{tk: ismine})\
.applyAlongDimensions(**{tk: 'mean'})
if mf is None:
mf = modf
else:
mf = mf.stack(modf, tk)
type=lambda x: tuple([eval(s) for s in x.split(',')]),
help='Label for model in figure')
parser.add_argument('--modreplace',
default='SONDE_,',
help='Remove,replace on sonde files')
parser.add_argument('sondes', nargs='+', help='Sonde files')
parser.add_argument('figpath', help='Path for output figure')
args = parser.parse_args()
modellabel = args.label
mf = None
sf = None
tk = None
for path in args.sondes:
modpath = path.replace(*args.modreplace.split(','))
modf = pncopen(modpath, format='netcdf', addcf=False)
if tk is None:
if 'TSTEP' in modf.dimensions:
tk = 'TSTEP'
else:
tk = 'time'
if mf is None:
mf = modf
else:
mf = mf.stack(modf, tk)
sonf = pncopen(path, format='netcdf', addcf=False)
if sf is None:
sf = sonf
else:
sf = sf.stack(sonf, tk)
def __init__(self, *args, **kwds):
"""
GCNC is defined here for the sole purpose of evalwoudc
It copies variables and attributes from the new netCDF
formatted GEOS-Chem and renames them to be consistent
with the PseudoNetCDF bpch reader output. This allows
the PseudoNetCDF bpch_base class to provide coordinate
manipulation avgSigma and ll2ij
"""
# Read the GEOS-Chem netCDF file
ds = pncopen(*args, format='netcdf', **kwds)
# Make quick references to object dictionaries
gvs = ds.variables
gds = ds.dimensions
# Copy all global attributes
self.setncatts(ds.getncatts())
# Make a list of coordinate variables to copy
coordkeys = 'time lat lon lev ilev hyai hybi'.split()
for coordk in coordkeys:
if coordk in gds:
self.copyDimension(gds[coordk], key=coordk)
self.copyVariable(gvs[coordk], key=coordk)
# Copy SpeciesConc_O3 as O3
self.copyVariable(gvs['SpeciesConc_O3'], 'O3')
# Rename lat/lon dimensions and coordinate variables
self.renameDimensions(lat='latitude', lon='longitude', inplace=True)
self.renameVariables(lat='latitude', lon='longitude', inplace=True)
# Get lat/lon for calculating edges
lat = self.variables['latitude'][:]
lon = self.variables['longitude'][:]
dlat = np.abs(np.median(np.diff(lat))) / 2.
dlon = np.abs(np.median(np.diff(lon))) / 2.
# Create variables for holding lat/lon boundaries
self.createDimension('nv', 2)
latb = self.createVariable('latitude_bounds', lat.dtype,
('latitude', 'nv'))
latb[:, 0] = np.maximum(lat - dlat, -90)
latb[:, 1] = np.minimum(lat + dlat, 90)
lonb = self.createVariable('longitude_bounds', lat.dtype,
('longitude', 'nv'))
lonb[:, 0] = lon - dlon
lonb[:, 1] = lon + dlon
# Convert the units of O3 to ppb
ov = self.variables['O3']
ov[:] *= 1e9
ov.units = 'ppb'
# Convert times to have an absolute reference
# much more convenient stacking later.
times = self.getTimes()
tv = self.variables['time']
tv.units = 'seconds since 1970-01-01 00:00:00+0000'
tv[:] = np.array([t.timestamp() for t in times])
default=False,
action='store_true',
help='Print bias on plot for each bin (every fifth between 1000-500hPa)')
parser.add_argument('--label', default='Mod', help='Label for model in figure')
parser.add_argument('--season',
default='ALL',
choices=list(seasons),
help='Choose a season, default ALL')
parser.add_argument('obs', help='Sonde observations')
parser.add_argument('mod', help='Mod for sondes')
parser.add_argument('figpath', help='Path for output figure')
args = parser.parse_args()
modellabel = args.label
args.season = seasons[args.season]
sf = pncopen(args.obs, format='ioapi', addcf=False)
mf = pncopen(args.mod, format='ioapi', addcf=False)
st = sf.getTimes()
sti = np.array([ti for ti, t in enumerate(st) if t.month in args.season])
mt = mf.getTimes()
mti = np.array([ti for ti, t in enumerate(mt) if t.month in args.season])
if 'TSTEP' in sf.dimensions:
tk = 'TSTEP'
else:
tk = 'time'
# sf = sf.sliceDimensions(**{tk: sti})
# mf = mf.sliceDimensions(**{tk: mti})
if 'Press' in sf.variables:
psrf = sf.variables['Press'][:].mean()
# Get altitude edges
altedges = self.variables['altitude_edges']
pr = self.variables['blview_backscatter_profile']
plot_kw.setdefault('norm', plt.matplotlib.colors.LogNorm())
patches = ax.pcolormesh(time_edges[:], altedges[:], pr[:].T, **plot_kw)
plt.colorbar(patches, cax=cax, label=pr.units.strip())
return fig
class test_vaisala(unittest.TestCase):
def setUp(self):
from PseudoNetCDF.testcase import ceilometerfiles_paths
self.vaisalapath = ceilometerfiles_paths['vaisala']
def testVAISALA2NCF(self):
import os
vfile = ceilometerl2(self.vaisalapath)
outpath = self.vaisalapath + '.nc'
ncfile = vfile.save(outpath)
for k, ncv in ncfile.variables.items():
vpv = vfile.variables[k]
np.testing.assert_allclose(ncv[...], vpv[...])
os.remove(outpath)
if __name__ == '__main__':
from PseudoNetCDF import pncopen
testdir = '/work/ROMO/users/kpc/flintfx/convceilometer/'
testpath = testdir + 'CEILOMETER_1_LEVEL_2_20.his'
f = pncopen(testpath, format='ceilometerl2')
print('Skipping', ', '.join(notmydatestrs))
if len(mydatestrs) == 0:
print('Skipping', prefix, 'no files match filter')
continue
# Make a list of model paths for the sonde files in teh season of interest
modpaths = [
dates[datestr].strftime(args.intemplate) for datestr in mydatestrs
]
# Make a list of paths for the sonde files in the season of interest
sondepaths = [sondepathdict[datestr] for datestr in mydatestrs]
# Open an example file for metadata
tmpf = pncopen(modpaths[0], **modformat, addcf=False)
# Use temp file to calculate i, j indices and ensure they are not arrays
nrow = len(tmpf.dimensions[dim2dim['latitude']])
ncol = len(tmpf.dimensions[dim2dim['longitude']])
try:
modi, modj = tmpf.ll2ij(lon, lat)
modi = int(modi)
modj = int(modj)
if ((modi < 0) or (modj < 0) or (modi >= ncol) or (modj >= nrow)):
raise ValueError()
except Exception:
print('Skipping', prefix, 'not in domain')
continue
# For CMAQ construct a "hybrid" sigma eta
from glob import glob
from PseudoNetCDF import pncopen
from collections import OrderedDict
import json
import sys
dataroot = sys.argv[1]
paths = glob(dataroot + '/*.l100')
out = {}
for path in paths:
print(path)
try:
f = pncopen(path, format='l100', addcf=False)
stype = 'NOAA'
sid = f.Flight_Number[:2]
key = '%s%s' % (stype, sid)
print(key)
name = f.Station
lat = f.variables['latitude'][0]
lon = f.variables['longitude'][0]
hght = f.Station_Height
hght_unit = hght.split(' ')[1]
outdate = f.getTimes()[0]
datestr = outdate.strftime('%F %H:%M:%S%z')
print(outdate)
if key not in out:
print('addedkey', key)
mine = out[key] = OrderedDict()
mine['Station'] = name
mine['Station Height'] = '{} {}'.format(hght, hght_unit)
from glob import glob
from PseudoNetCDF import pncopen
from collections import OrderedDict
import json
import sys
dataroot = sys.argv[1]
paths = glob(dataroot + '/*')
out = {}
for path in paths:
print(path)
try:
f = pncopen(path, format='woudcsonde', addcf=False)
stype = f.variables['PLATFORM_Type'].view('S64')[0, 0].decode().strip()
sid = f.variables['PLATFORM_ID'][:].array()[0]
key = '%s%03d' % (stype, sid)
print(key)
name = f.variables['PLATFORM_Name'].view('S64')[0, 0].decode()
lat = f.variables['LOCATION_Latitude'][0]
lon = f.variables['LOCATION_Longitude'][0]
hghtv = f.variables['LOCATION_Height']
hght = hghtv[0]
hght_unit = hghtv.units.strip().replace('unknown', 'meters')
outdate = f.getTimes()[0]
datestr = outdate.strftime('%F %H:%M:%S%z')
print(outdate)
if key not in out:
print('addedkey', key)
mine = out[key] = OrderedDict()
mine['Station'] = name
import matplotlib.pyplot as plt
from warnings import warn
from scipy.stats import pearsonr, spearmanr
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('--seasonkeys',
default='ALL DJF MAM JJA SON'.split(),
type=lambda x: x.split(','),
help='comma delimited seasons (ALL,DJF,MAM,SON)')
parser.add_argument('--obsname', default='OBS')
parser.add_argument('--modname', default='MOD')
parser.add_argument('obspath')
parser.add_argument('modpath')
args = parser.parse_args()
cmaqf = pncopen(args.modpath, addcf=False, format='netcdf')
cnetf = pncopen(args.obspath, addcf=False, format='netcdf')
times = cmaqf.getTimes()
season = dict(JULY=[7],
DJF=[12, 1, 2],
MAM=[3, 4, 5],
JJA=[6, 7, 8],
SON=[9, 10, 11],
ALL=range(1, 13))
season = dict([(seasonk, season[seasonk]) for seasonk in args.seasonkeys])
sitenames = cnetf.SITENAMES.split('|')
for monthk, monthi in season.items():
tslice = np.array([t.month in monthi for t in times])
INT*4 Number of particles
INT*4 Number of pollutants
INT*4 Time of particle dump (YEAR, MONTH, DAY, HOUR, MINUTES)
"""
rec1 = np.fromfile(self._path, dtype='>i,>7i,>i', count=1)
self._file.seek(0, 0)
hdr = rec1['f1'][0]
self.NPARTICLES = hdr[0]
self.NPOLLUTANTS = hdr[1]
self.YEAR = hdr[2]
self.MONTH = hdr[3]
self.DAY = hdr[4]
self.HOUR = hdr[5]
self.MINUTES = hdr[6]
assert (rec1['f0'] == rec1['f2'])
datfmt = '>i,>({},)f,>i,>i,>6f,>i,>i,>5i,>i'.format(self.NPOLLUTANTS)
self._fmt = np.dtype([('header', '>i,>7i,>i'),
('data', datfmt, (self.NPARTICLES, ))])
data = np.fromfile(self._path, dtype=self._fmt)
assert (data.shape[0] == 1)
self._data = data[0]
db = self._data['data']
assert ((db['f0'] == db['f2']).all())
assert ((db['f3'] == db['f5']).all())
assert ((db['f6'] == db['f8']).all())
if __name__ == '__main__':
from PseudoNetCDF import pncopen
f = pncopen('PARDUMP_143', format='arlpardump')
ax.plot([0, vmax], [0, vmax], color='k')
pr = pearsonr(xc, yc)
sr = spearmanr(xc, yc)
ax.set_title(
'Pearson (r={:0.2f}, p={:0.2f}); Spearman (r={:0.2f}, p={:0.2f})'.
format(*(pr + sr)))
ax.set_xlabel(
args.obsname +
' O3 ppb ({:.0f}, {:.0f}, {:.0f})'.format(cnetmin, cnetmean, cnetmax))
ax.set_ylabel(
args.modname +
' O3 ppb ({:.0f}, {:.0f}, {:.0f})'.format(cmaqmin, cmaqmean, cmaqmax))
return ax.figure
cmaqf = pncopen(args.modpath, addcf=False,
format='netcdf').copy().removeSingleton()
cnetf = pncopen(args.obspath, addcf=False, format='netcdf').copy()
times = cmaqf.getTimes()
for monthk, monthi in season.items():
tslice = np.array([t.month in monthi for t in times])
print(monthk, monthi, tslice.sum())
fig = plotslice(cmaqf, cnetf, tslice)
fig.savefig('hist2d/hist2d_O3_' + monthk + '.png')
lidx = (np.arange(24,
len(times) - 24)[:, None] -
cnetf.variables['TIME_OFFSET']).astype('i')
lidx.dimensions = ('time', 'site')
uidx, sidx = np.indices(lidx[:].shape)
sidx = lidx * 0 + sidx