def _concat(self, arm_data_objs, close_gaps=True):
for att in self._concatable:
first_object = getattr(arm_data_objs[0], att)
which_type = type(first_object).__name__
data_period = first_object._data_period
if which_type == 'TimeSeries_2D':
value = _timeseries.TimeSeries_2D(
_pd.concat([getattr(i, att).data for i in arm_data_objs]))
elif which_type == 'TimeSeries':
value = _timeseries.TimeSeries(
_pd.concat([getattr(i, att).data for i in arm_data_objs]))
elif which_type == 'AMS_Timeseries_lev01':
value = _AMS.AMS_Timeseries_lev01(
_pd.concat([getattr(i, att).data for i in arm_data_objs]))
elif which_type == 'SizeDist_TS':
# value = _AMS.AMS_Timeseries_lev01(pd.concat([getattr(i, att).data for i in arm_data_objs]))
data = _pd.concat(
[getattr(i, att).data for i in arm_data_objs])
value = _sizedistribution.SizeDist_TS(
data,
getattr(arm_data_objs[0], att).bins,
'dNdlogDp',
ignore_data_gap_error=True,
)
elif which_type == 'TimeSeries_3D':
value = _timeseries.TimeSeries_3D(
_pd.concat([getattr(i, att).data for i in arm_data_objs]))
else:
raise TypeError(
'%s is not an allowed type here (TimeSeries_2D, TimeSeries)'
% which_type)
if hasattr(first_object, 'availability'):
try:
avail_concat = _pd.concat([
getattr(i, att).availability.availability
for i in arm_data_objs
])
avail = Data_Quality(None, avail_concat, None,
first_object.flag_info)
value.availability = avail
except:
_warnings.warn(
'availability could not be concatinated make sure you converted it to a pandas frame at some point!'
)
value._data_period = data_period
if close_gaps:
setattr(self, att, value.close_gaps())
else:
setattr(self, att, value)
def _parse_netCDF(self):
"returns a dictionary, with panels in it"
super(ArmDatasetSub,self)._parse_netCDF()
size_bins = self._read_variable('size_bins')['data'] * 1000
df = pd.DataFrame(self._read_variable('RH_interDMA')['data'], index = self.time_stamps, columns=size_bins)
df.columns.name = 'size_bin_center_nm'
self.RH_interDMA = timeseries.TimeSeries(df)
self.RH_interDMA._data_period = self._data_period
data = self._read_variable('hyg_distributions')['data']
growthfactors = self._read_variable('growthfactors')['data']
data = pd.Panel(data, items= self.time_stamps, major_axis = size_bins, minor_axis = growthfactors)
data.major_axis.name = 'size_bin_center_nm'
data.minor_axis.name = 'growthfactors'
self.hyg_distributions = timeseries.TimeSeries_3D(data)
self.hyg_distributions._data_period = self._data_period
def mean_growth_factor(self):
"""Calculates the mean growthfactor of the particular size bin."""
if '__mean_growth_factor' not in dir(self):
def mean_linewise(gf_dist):
growthfactors = self.hyg_distributions.data.minor_axis.values
# meanl = ((gf_dist[~ gf_dist.mask] * np.log10(growthfactors[~ gf_dist.mask])).sum()/gf_dist[~gf_dist.mask].sum())
meanl = ((gf_dist[~ np.isnan(gf_dist)] * np.log10(growthfactors[~ np.isnan(gf_dist)])).sum()/gf_dist[~np.isnan(gf_dist)].sum())
stdl = np.sqrt((gf_dist[~ np.isnan(gf_dist)] * (np.log10(growthfactors[~ np.isnan(gf_dist)]) - meanl)**2).sum()/gf_dist[~np.isnan(gf_dist)].sum())
return np.array([10**meanl,stdl])
data = self.hyg_distributions.data
allmeans = timeseries.TimeSeries_3D(pd.Panel(items=data.items, major_axis=data.major_axis, minor_axis= ['mean', 'std_log']))
for i,time in enumerate(data.values):
for e,size in enumerate(time):
allmeans.data.iloc[i,e] = mean_linewise(size)
self.__mean_growth_factor = allmeans
self.__mean_growth_factor._data_period = self._data_period
return self.__mean_growth_factor