def air_density():
Rd = 287. # [J K-1 kg-1]
# density values do not vary significantly
Tv = tm.virtual_temperature(C=stemp, mixing_ratio=smixr / 1000.) + 273.15
air_density1 = (spress * 100.) / (Rd * Tv)
Tv = tm.virtual_temperature(C=mtemp, mixing_ratio=smixr / 1000.) + 273.15
air_density2 = (mpress * 100.) / (Rd * Tv)
def air_density():
Rd = 287. # [J K-1 kg-1]
# density values do not vary significantly
Tv = tm.virtual_temperature(C=stemp, mixing_ratio=smixr/1000.)+273.15
air_density1 = (spress*100.)/(Rd*Tv)
Tv = tm.virtual_temperature(C=mtemp, mixing_ratio=smixr/1000.)+273.15
air_density2 = (mpress*100.)/(Rd*Tv)
def get_meteo(self,start_time, end_time):
start = self.df.index.searchsorted(start_time)
end = self.df.index.searchsorted(end_time)
# print meteo.columns.tolist()
cols=[0,1,2,3,6,7,8,9,13,14,15]
meteo=self.df.ix[start:end,cols].copy()
''' pressure '''
pres = meteo.apres.values
''' add relative humidity '''
temp = meteo.atemp.values
dewp = meteo.dewp.values
relh = thermo.relative_humidity(C=temp,Dewp=dewp) # [%]
meteo.loc[:,'relh'] = pd.Series(relh,index=meteo.index)
''' mixing ratio '''
satmixr=thermo.sat_mix_ratio(C=temp,hPa=pres)
mixr=relh*satmixr/100
''' add theta '''
theta = thermo.theta2(C=temp,hPa=pres,mixing_ratio=mixr)
meteo.loc[:,'theta'] = pd.Series(theta,index=meteo.index)
''' add thetav '''
thetav = thermo.virtual_temperature(theta=theta,mixing_ratio=mixr)
meteo.loc[:,'thetav'] = pd.Series(thetav,index=meteo.index)
''' add thetae '''
thetaeq = thermo.theta_equiv2(C=temp,hPa=pres,
mixing_ratio=mixr,relh=relh)
meteo.loc[:,'thetaeq'] = pd.Series(thetaeq,index=meteo.index)
return meteo
def parse_surface_met(file_met, index_field=None,
name_field=None, locelevation=None):
from scipy import stats
'''
Process surface meteorological files
with .met extension
Output dataframe assumes:
- Each file is a 24h period observation
- Frequency of observation is constant
'''
casename = file_met.split('/')[-2]
station_id = file_met.split('/')[-1][:3]
usr_case = str(int(casename[-2:]))
if usr_case in ['1', '2']:
index_field = {'bby': [3, 4, 10, 5, 6, 11, 13],
'czc': [3, 4, 10, 5, 6, 11, 13]}
elif usr_case in ['3', '4', '5', '6', '7']:
index_field = {'bby': [3, 5, 8, 10, 12, 17, 26],
'czc': [3, 4, 5, 6, 8, 13, 22]}
else:
index_field = {'bby': [3, 4, 5, 6, 8, 13, 15],
'czc': [3, 4, 5, 6, 8, 13, 15],
'frs': [3, 4, 5, 6, 8, 13, 15]}
elev = {'bby': 15, 'czc': 462, 'frs': 45}
name_field = ['press', 'temp', 'rh', 'wspd', 'wdir', 'precip', 'mixr']
index_field = index_field[station_id]
locelevation = elev[station_id]
''' read the csv file '''
raw_dframe = pd.read_csv(file_met, header=None)
''' parse date columns into a single date col '''
dates_col = [0, 1, 2]
raw_dates = raw_dframe.ix[:, dates_col]
raw_dates.columns = ['Y', 'j', 'HHMM']
raw_dates['HHMM'] = raw_dates['HHMM'].apply(lambda x: '{0:0>4}'.format(x))
raw_dates = raw_dates.apply(
lambda x: '%s %s %s' % (x['Y'], x['j'], x['HHMM']), axis=1)
dates_fmt = '%Y %j %H%M'
dates = raw_dates.apply(lambda x: datetime.strptime(x, dates_fmt))
''' make meteo df, assign datetime index, and name columns '''
meteo = raw_dframe.ix[:, index_field]
meteo.index = dates
meteo.columns = name_field
''' create a dataframe with regular and continuous 24h period time index
(this works as a QC for time gaps)
'''
nano = 1000000000
time_diff = np.diff(meteo.index)
sample_freq = (stats.mode(time_diff)[0][0]/nano).astype(int) # [seconds]
fidx = meteo.index[0] # (start date for dataframe)
fidx_str = pd.to_datetime(
fidx.year*10000 + fidx.month*100 + fidx.day, format='%Y%m%d')
periods = (24*60*60)/sample_freq
ts = pd.date_range(fidx_str, periods=periods, freq=str(sample_freq)+'s')
nanarray = np.empty((periods, 1))
nanarray[:] = np.nan
df = pd.DataFrame(nanarray, index=ts)
meteo = df.join(meteo, how='outer')
meteo.drop(0, axis=1, inplace=True)
''' make field with hourly acum precip '''
hour = pd.TimeGrouper('H')
preciph = meteo.precip.groupby(hour).sum()
meteo = meteo.join(preciph, how='outer', rsuffix='h')
''' add thermodynamics '''
theta = tm.theta1(C=meteo.temp, hPa=meteo.press)
thetaeq = tm.theta_equiv1(C=meteo.temp, hPa=meteo.press)
meteo.loc[:, 'theta'] = pd.Series(theta, index=meteo.index)
meteo.loc[:, 'thetaeq'] = pd.Series(thetaeq, index=meteo.index)
''' add sea level pressure '''
Tv = tm.virtual_temperature(C=meteo.temp, mixing_ratio=meteo.mixr/1000.)
slp = tm.sea_level_press(K=Tv+273.15, Pa=meteo.press*100, m=locelevation)
meteo.loc[:, 'sea_levp'] = slp
''' assign metadata (prototype, not really used) '''
units = {'press': 'mb', 'temp': 'C', 'rh': '%', 'wspd': 'm s-1',
'wdir': 'deg', 'precip': 'mm', 'mixr': 'g kg-1'}
agl = {'press': 'NaN', 'temp': '10 m', 'rh': '10 m',
'wspd': 'NaN', 'wdir': 'NaN', 'precip': 'NaN', 'mixr': 'NaN'}
for n in name_field:
meteo[n].units = units[n]
meteo[n].agl = agl[n]
meteo[n].nan = -9999.999
meteo[n].sampling_freq = '2 minute'
return meteo
def parse_surface_met(file_met,
index_field=None,
name_field=None,
locelevation=None):
from scipy import stats
'''
Process surface meteorological files
with .met extension
Output dataframe assumes:
- Each file is a 24h period observation
- Frequency of observation is constant
'''
casename = file_met.split('/')[-2]
station_id = file_met.split('/')[-1][:3]
usr_case = str(int(casename[-2:]))
if usr_case in ['1', '2']:
index_field = {
'bby': [3, 4, 10, 5, 6, 11, 13],
'czc': [3, 4, 10, 5, 6, 11, 13]
}
elif usr_case in ['3', '4', '5', '6', '7']:
index_field = {
'bby': [3, 5, 8, 10, 12, 17, 26],
'czc': [3, 4, 5, 6, 8, 13, 22]
}
else:
index_field = {
'bby': [3, 4, 5, 6, 8, 13, 15],
'czc': [3, 4, 5, 6, 8, 13, 15],
'frs': [3, 4, 5, 6, 8, 13, 15]
}
elev = {'bby': 15, 'czc': 462, 'frs': 45}
name_field = ['press', 'temp', 'rh', 'wspd', 'wdir', 'precip', 'mixr']
index_field = index_field[station_id]
locelevation = elev[station_id]
''' read the csv file '''
raw_dframe = pd.read_csv(file_met, header=None)
''' parse date columns into a single date col '''
dates_col = [0, 1, 2]
raw_dates = raw_dframe.ix[:, dates_col]
raw_dates.columns = ['Y', 'j', 'HHMM']
raw_dates['HHMM'] = raw_dates['HHMM'].apply(lambda x: '{0:0>4}'.format(x))
raw_dates = raw_dates.apply(lambda x: '%s %s %s' %
(x['Y'], x['j'], x['HHMM']),
axis=1)
dates_fmt = '%Y %j %H%M'
dates = raw_dates.apply(lambda x: datetime.strptime(x, dates_fmt))
''' make meteo df, assign datetime index, and name columns '''
meteo = raw_dframe.ix[:, index_field]
meteo.index = dates
meteo.columns = name_field
''' create a dataframe with regular and continuous 24h period time index
(this works as a QC for time gaps)
'''
nano = 1000000000
time_diff = np.diff(meteo.index)
sample_freq = (stats.mode(time_diff)[0][0] / nano).astype(int) # [seconds]
fidx = meteo.index[0] # (start date for dataframe)
fidx_str = pd.to_datetime(fidx.year * 10000 + fidx.month * 100 + fidx.day,
format='%Y%m%d')
periods = (24 * 60 * 60) / sample_freq
ts = pd.date_range(fidx_str, periods=periods, freq=str(sample_freq) + 's')
nanarray = np.empty((periods, 1))
nanarray[:] = np.nan
df = pd.DataFrame(nanarray, index=ts)
meteo = df.join(meteo, how='outer')
meteo.drop(0, axis=1, inplace=True)
''' make field with hourly acum precip '''
hour = pd.TimeGrouper('H')
preciph = meteo.precip.groupby(hour).sum()
meteo = meteo.join(preciph, how='outer', rsuffix='h')
''' add thermodynamics '''
theta = tm.theta1(C=meteo.temp, hPa=meteo.press)
thetaeq = tm.theta_equiv1(C=meteo.temp, hPa=meteo.press)
meteo.loc[:, 'theta'] = pd.Series(theta, index=meteo.index)
meteo.loc[:, 'thetaeq'] = pd.Series(thetaeq, index=meteo.index)
''' add sea level pressure '''
Tv = tm.virtual_temperature(C=meteo.temp, mixing_ratio=meteo.mixr / 1000.)
slp = tm.sea_level_press(K=Tv + 273.15,
Pa=meteo.press * 100,
m=locelevation)
meteo.loc[:, 'sea_levp'] = slp
''' assign metadata (prototype, not really used) '''
units = {
'press': 'mb',
'temp': 'C',
'rh': '%',
'wspd': 'm s-1',
'wdir': 'deg',
'precip': 'mm',
'mixr': 'g kg-1'
}
agl = {
'press': 'NaN',
'temp': '10 m',
'rh': '10 m',
'wspd': 'NaN',
'wdir': 'NaN',
'precip': 'NaN',
'mixr': 'NaN'
}
for n in name_field:
meteo[n].units = units[n]
meteo[n].agl = agl[n]
meteo[n].nan = -9999.999
meteo[n].sampling_freq = '2 minute'
return meteo
