def calc_DWR(datalist):
var = h.lin2z(datalist[0]['var']) - h.lin2z(datalist[1]['var'])
mask = np.logical_or(datalist[0]['mask'], datalist[1]['mask'])
return var, mask
def filter_ghost_echos_RPG94GHz_FMCW(data, **kwargs):
######################################################################
#
# 2nd and 3rd chirp ghost echo filter
if 'clean_spectra' in kwargs and kwargs['clean_spectra']:
sensitivity_limit = kwargs['SL'] if 'SL' in kwargs else sys.exit(
'Error in clean_spectra ghost echo filter :: Sensitivity Limit missing!')
Ze = kwargs['Ze'] if 'Ze' in kwargs else sys.exit(
'Error in clean_spectra ghost echo filter :: Ze values missing!')
for ichirp in [0, 1, 2]:
# threholds for 3rd chrip ghost echo filter
new_ny_vel = data[ichirp]['vel'].max() - 2.5
ic_Ze_max = h.z2lin(-22.5)
idx_left = np.argwhere(-new_ny_vel > data[ichirp]['vel']).max()
idx_right = np.argwhere(new_ny_vel < data[ichirp]['vel']).min()
Ze_lin_left = data[ichirp]['var'][:, :, :idx_left].copy()
Ze_lin_right = data[ichirp]['var'][:, :, idx_right:].copy()
# if noise was already removed by the RPG software, replace the ghost with -999.,
# if noise factor 0 was selected in the RPG software, replace the ghost by the minimum spectrum value,
# to avoid wrong noise estimations (to much signal would be lost otherwise),
idx_ts_nf0 = np.argwhere(data[ichirp]['var'][:, 0, 0] != -999.0)
if idx_ts_nf0.size > 0:
mask_left, mask_right = Ze_lin_left < ic_Ze_max, Ze_lin_right < ic_Ze_max
min_left, min_right = np.amin(Ze_lin_left, axis=2), np.amin(Ze_lin_right, axis=2)
for i_bin in range(mask_left.shape[2]):
Ze_lin_left[mask_left[:, :, i_bin], i_bin] = min_left[mask_left[:, :, i_bin]]
for i_bin in range(mask_right.shape[2]):
Ze_lin_right[mask_right[:, :, i_bin], i_bin] = min_right[mask_right[:, :, i_bin]]
else:
Ze_lin_left[Ze_lin_left < ic_Ze_max] = -999.0
Ze_lin_right[Ze_lin_right < ic_Ze_max] = -999.0
data[ichirp]['var'][:, :, :idx_left] = Ze_lin_left.copy()
data[ichirp]['var'][:, :, idx_right:] = Ze_lin_right.copy()
######################################################################
#
# 2nd and 3rd chirp ghost echo filter
if 'C2C3' in kwargs and kwargs['C2C3']:
for ichirp in [0, 1, 2]:
# threholds for 3rd chrip ghost echo filter
new_ny_vel = data[ichirp]['vel'].max() - 2.5
ic_Ze_max = h.z2lin(-22.5)
idx_left = np.argwhere(-new_ny_vel > data[ichirp]['vel']).max()
idx_right = np.argwhere(new_ny_vel < data[ichirp]['vel']).min()
Ze_lin_left = data[ichirp]['var'][:, :, :idx_left].copy()
Ze_lin_right = data[ichirp]['var'][:, :, idx_right:].copy()
# if noise was already removed by the RPG software, replace the ghost with -999.,
# if noise factor 0 was selected in the RPG software, replace the ghost by the minimum spectrum value,
# to avoid wrong noise estimations (to much signal would be lost otherwise),
idx_ts_nf0 = np.argwhere(data[ichirp]['var'][:, 0, 0] != -999.0)
if idx_ts_nf0.size > 0:
mask_left, mask_right = Ze_lin_left < ic_Ze_max, Ze_lin_right < ic_Ze_max
min_left, min_right = np.amin(Ze_lin_left, axis=2), np.amin(Ze_lin_right, axis=2)
for i_bin in range(mask_left.shape[2]):
Ze_lin_left[mask_left[:, :, i_bin], i_bin] = min_left[mask_left[:, :, i_bin]]
for i_bin in range(mask_right.shape[2]):
Ze_lin_right[mask_right[:, :, i_bin], i_bin] = min_right[mask_right[:, :, i_bin]]
else:
Ze_lin_left[Ze_lin_left < ic_Ze_max] = -999.0
Ze_lin_right[Ze_lin_right < ic_Ze_max] = -999.0
data[ichirp]['var'][:, :, :idx_left] = Ze_lin_left.copy()
data[ichirp]['var'][:, :, idx_right:] = Ze_lin_right.copy()
######################################################################
#
# 1st chirp ghost echo filter
if 'C1' in kwargs and kwargs['C1']:
# invalid mask has to be provided via kwarg for this filter
invalid_mask = kwargs['inv_mask']
rg_offsets = kwargs['offset']
SL = kwargs['SL']
# setting higher threshold if chirp 2 contains high reflectivity values
# sum_over_heightC1 = np.ma.sum(data['Ze'][:rg_offsets[1], :], axis=0)
sum_over_heightC2 = np.ma.sum(data['Ze'][rg_offsets[1]:rg_offsets[2], :], axis=0)
# load sensitivity limits (time, height) and calculate the mean over time
sens_reduction = 15.0 # sensitivity in chirp 1 is reduced by 12.0 dBZ
C2_Ze_threshold = 18.0 # if sum(C2_Ze) > this threshold, ghost echo in C1 is assumed
sens_lim = h.z2lin(h.lin2z(np.mean(SL, axis=0)) + sens_reduction)
ts_to_mask = np.argwhere(h.lin2z(sum_over_heightC2) > C2_Ze_threshold)[:, 0]
m1 = invalid_mask[:rg_offsets[1], :].copy()
for idx_ts in ts_to_mask:
m1[:, idx_ts] = data['Ze'][:rg_offsets[1], idx_ts] < sens_lim
# a = data['Ze'][:rg_offsets[1], idx_ts] < sens_lim
# b = data['VEL'][:rg_offsets[1], idx_ts] < -0.5
# m1[:, idx_ts] = a == b
invalid_mask[:rg_offsets[1], :] = m1.copy()
for mom in ['Ze', 'VEL', 'sw', 'skew', 'kurt']:
data[mom][:rg_offsets[1], :] = np.ma.masked_where(m1, data[mom][:rg_offsets[1], :])
return invalid_mask
fig, ax = pyLARDA.Transformations.plot_timeheight(LIMRAD94_Z_interp,
range_interval=plot_range,
z_converter='lin2z')
fig.savefig('limrad_Z_interp.png', dpi=250)
def calc_DWR(datalist):
var = h.lin2z(datalist[0]['var']) - h.lin2z(datalist[1]['var'])
mask = np.logical_or(datalist[0]['mask'], datalist[1]['mask'])
return var, mask
new_keys = {
'system': '',
'name': 'DWR',
'colormap': 'gist_rainbow',
'rg_unit': 'm',
'var_lims': [-7, 2]
}
DWR = pyLARDA.Transformations.combine(calc_DWR, [MIRA_Zg, LIMRAD94_Z_interp],
new_keys)
fig, ax = pyLARDA.Transformations.plot_timeheight(DWR,
range_interval=plot_range)
fig.savefig('DWR_MIRA_LIMRAD.png', dpi=250)
MRR_Z = larda.read("MRRPRO", "Ze", [begin_dt, end_dt], [0, 'max'])
MRR_Z['var'] = h.lin2z(MRR_Z['var'])
fig, ax = pyLARDA.Transformations.plot_timeheight(MRR_Z,
range_interval=plot_range)
fig.savefig('mrr_Z.png', dpi=250)
range_interval=plot_range,
z_converter='lin2z')
fig.savefig(case_prefix + 'limrad_Z.png', dpi=250)
LIMRAD94_Z_interp = pyLARDA.Transformations.interpolate2d(
LIMRAD94_Z, new_time=MIRA_Zg['ts'], new_range=MIRA_Zg['rg'])
fig, ax = pyLARDA.Transformations.plot2d(LIMRAD94_Z_interp,
range_interval=plot_range,
z_converter='lin2z')
fig.savefig(case_prefix + 'limrad_Z_interp.png', dpi=250)
# Z scatterplot
combined_mask = np.logical_or(MIRA_Zg['mask'], LIMRAD94_Z_interp['mask'])
Mira_Z_scatter = h.lin2z(MIRA_Zg['var'][~combined_mask].ravel()) #+4.5
Limrad_Z_scatter = h.lin2z(LIMRAD94_Z_interp['var'][~combined_mask].ravel())
s, i, r, p, std_err = stats.linregress(Mira_Z_scatter, Limrad_Z_scatter)
H, xedges, yedges = np.histogram2d(Mira_Z_scatter,
Limrad_Z_scatter,
bins=120,
range=[[-75, 30], [-75, 30]])
X, Y = np.meshgrid(xedges, yedges)
fig, ax = plt.subplots(1, figsize=(5.7, 5.7))
ax.pcolormesh(
X,
Y,
np.transpose(H),
norm=matplotlib.colors.LogNorm(),
)