def get_ctth_bias_high(self, caObj):
from get_flag_info import get_calipso_high_clouds
high_clouds = get_calipso_high_clouds(caObj)
detected_high = np.logical_and(self.detected_height, high_clouds[self.use])
delta_h = self.height_bias.copy()
delta_h[~detected_high]=0
self.height_bias_high = delta_h.copy()
self.detected_height_high = detected_high
delta_h = self.lapse_bias.copy()
delta_h[~detected_high]=0
self.lapse_bias_high = delta_h
def make_compare(caObj, caObj2, name):
low_clouds = get_calipso_low_clouds(caObj)
high_clouds = get_calipso_high_clouds(caObj)
medium_clouds = get_calipso_medium_clouds(caObj)
height_c = (1000 * caObj.calipso.all_arrays['layer_top_altitude'][:, 0] -
caObj.calipso.all_arrays['elevation'])
height_pps1 = caObj.avhrr.all_arrays['ctth_height']
height_pps2 = caObj2.avhrr.all_arrays['ctth_height']
use = np.logical_and(height_pps1 > -1, height_pps2 > -1)
use = np.logical_and(
use, caObj.calipso.all_arrays['layer_top_altitude'][:, 0] > -1)
low = np.logical_and(low_clouds, use)
medium = np.logical_and(medium_clouds, use)
high = np.logical_and(high_clouds, use)
bias1 = height_pps1 - height_c
bias2 = height_pps2 - height_c
limit = np.percentile(bias2[use], 5)
print limit
fig = plt.figure(figsize=(9, 9))
ax = fig.add_subplot(111)
plt.boxplot(
[
bias1[low], bias1[medium], bias1[high], bias2[low], bias2[medium],
bias2[high]
],
whis=[2, 98],
sym='',
labels=["low1", "medium1", "high1", "low2", "medium2", "high2"],
showmeans=True)
ax.set_ylim(-15000, 15000)
ax.fill_between(np.arange(0, 12),
-1500,
1500,
facecolor='green',
alpha=0.2)
ax.fill_between(np.arange(0, 12), 2000, 15000, facecolor='red', alpha=0.2)
ax.fill_between(np.arange(0, 12),
-2000,
-15000,
facecolor='red',
alpha=0.2)
for y_val in xrange(-5000, 6000, 1000):
plt.plot(np.arange(0, 12), y_val + 0 * np.arange(0, 12), ':k')
plt.title(name)
plt.savefig(
"/home/a001865/PICTURES_FROM_PYTHON/CTTH_LAPSE_RATE_INVESTIGATION/ctth_box_plot_%s.png"
% (name))
plt.show()
"""
def table_21_do_for_atbd(caObj):
cloudtype_conditions = caObj.avhrr.all_arrays['cloudtype_conditions']
cloudtype_status = caObj.avhrr.all_arrays['cloudtype_status']
(no_qflag, night_flag, twilight_flag, day_flag,
all_dnt_flag) = get_day_night_twilight_info_pps2014(cloudtype_conditions)
low_clouds = get_calipso_low_clouds(caObj)
high_clouds = get_calipso_high_clouds(caObj)
medium_clouds = get_calipso_medium_clouds(caObj)
caliop_ok = np.logical_or(low_clouds,
np.logical_or(high_clouds, medium_clouds))
pps_ctype = caObj.avhrr.all_arrays['cloudtype']
use = np.logical_and(pps_ctype > 4, pps_ctype < 23)
use = np.logical_and(
use,
np.logical_and(caliop_ok,
caObj.calipso.all_arrays['cloud_fraction'] > 0.9))
pps_low = [pps_ctype_i in [5, 6, 10] for pps_ctype_i in pps_ctype]
pps_medium = [pps_ctype_i in [7] for pps_ctype_i in pps_ctype]
pps_high = [
pps_ctype_i in [8, 9, 11, 12, 13, 14, 15, 16, 17, 18]
for pps_ctype_i in pps_ctype
]
print "POD low medium high FAR low medium high"
for use_i, name in zip([all_dnt_flag, day_flag, night_flag, twilight_flag],
["all", "day", "night", "twilight"]):
use_this = np.logical_and(use, use_i)
pps_low_i = np.logical_and(pps_low, use_this)
low_clouds_i = np.logical_and(low_clouds, use_this)
n_low_ok = 1.0 * np.sum(np.logical_and(pps_low_i, low_clouds_i))
POD_low = 100 * n_low_ok / np.sum(low_clouds_i)
FAR_low = 100 * (np.sum(pps_low_i) -
n_low_ok) * 1.0 / np.sum(pps_low_i)
print n_low_ok, np.sum(pps_low_i), np.sum(low_clouds_i)
pps_medium_i = np.logical_and(pps_medium, use_this)
medium_clouds_i = np.logical_and(medium_clouds, use_this)
n_medium_ok = 1.0 * np.sum(
np.logical_and(pps_medium_i, medium_clouds_i))
POD_medium = 100 * n_medium_ok / np.sum(medium_clouds_i)
FAR_medium = 100 * (np.sum(pps_medium_i) -
n_medium_ok) / np.sum(pps_medium_i)
pps_high_i = np.logical_and(pps_high, use_this)
high_clouds_i = np.logical_and(high_clouds, use_this)
n_high_ok = 1.0 * np.sum(np.logical_and(pps_high_i, high_clouds_i))
POD_high = 100 * n_high_ok / np.sum(high_clouds_i)
FAR_high = 100 * (np.sum(pps_high_i) - n_high_ok) / np.sum(pps_high_i)
print "%3.1f %3.1f %3.1f %3.1f %3.1f %3.1f" % (
POD_low, POD_medium, POD_high, FAR_low, FAR_medium, FAR_high)
def make_boxplot(caObj, name):
low_clouds = get_calipso_low_clouds(caObj)
high_clouds = get_calipso_high_clouds(caObj)
medium_clouds = get_calipso_medium_clouds(caObj)
height_c = (1000 * caObj.calipso.all_arrays['layer_top_altitude'][:, 0] -
caObj.calipso.all_arrays['elevation'])
height_pps = caObj.avhrr.all_arrays['ctth_height']
use = np.logical_and(
height_pps > -1,
caObj.calipso.all_arrays['layer_top_altitude'][:, 0] > -1)
low = np.logical_and(low_clouds, use)
medium = np.logical_and(medium_clouds, use)
high = np.logical_and(high_clouds, use)
bias = height_pps - height_c
limit = np.percentile(bias[use], 5)
print limit
from matplotlib import rcParams
rcParams.update({'figure.autolayout': True})
fig = plt.figure(figsize=(6, 9))
ax = fig.add_subplot(111)
#plt.tight_layout()
#plt.subplots_adjust(left=0.2)
#plt.subplots_adjust(left=10, bottom=10, right=10, top=10, wspace=0, hspace=0)
plt.boxplot([bias[low], bias[medium], bias[high]],
whis=[2, 98],
sym='',
labels=["low", "medium", "high"],
showmeans=True)
ax.set_ylim(-14000, 8000)
ax.fill_between(np.arange(0, 6), -500, 500, facecolor='green', alpha=0.6)
ax.fill_between(np.arange(0, 6), -1000, 1000, facecolor='green', alpha=0.4)
ax.fill_between(np.arange(0, 6), -1500, 1500, facecolor='green', alpha=0.2)
ax.fill_between(np.arange(0, 6), 2000, 15000, facecolor='red', alpha=0.2)
ax.fill_between(np.arange(0, 6), -2000, -15000, facecolor='red', alpha=0.2)
for y_val in [-5, -4, -3, -2, 2, 3, 4, 5]:
plt.plot(np.arange(0, 6), y_val * 1000 + 0 * np.arange(0, 6), ':k')
plt.plot(np.arange(0, 6), -10 * 1000 + 0 * np.arange(0, 6), ':k')
plt.plot(np.arange(0, 6), 0 + 0 * np.arange(0, 6), 'k')
plt.title(name)
plt.savefig(
"/home/a001865/PICTURES_FROM_PYTHON/CTTH_LAPSE_RATE_INVESTIGATION/ctth_box_plot_%s.png"
% (name))
def prototype_ctth_alg_lapse_rate(caObj, filename):
from get_flag_info import (get_semi_opaque_info_pps2014,
get_calipso_high_clouds,
get_calipso_medium_clouds,
get_calipso_low_clouds)
low_clouds = get_calipso_low_clouds(caObj)
high_clouds = get_calipso_high_clouds(caObj)
medium_clouds = get_calipso_medium_clouds(caObj)
semi, opaq = get_semi_opaque_info_pps2014(caObj.avhrr.ctth_status)
print "prototype new ctth height"
height_c = (1000*caObj.calipso.all_arrays['layer_top_altitude'][:,0] -
caObj.calipso.all_arrays['elevation'])
height_pps = caObj.avhrr.all_arrays['ctth_height']
bias = height_pps - height_c
tsur = caObj.avhrr.all_arrays['surftemp']
tsur = caObj.avhrr.all_arrays['segment_nwp_temp'][:,0]
d45 = (caObj.avhrr.all_arrays['bt11micron'] -
caObj.avhrr.all_arrays['bt12micron'])
temperature_pps = caObj.avhrr.all_arrays['bt12micron'].copy()#
temp_diff = temperature_pps - tsur
new_pps_h = calculate_lapse_rate_h(caObj)
bias_l_rate = new_pps_h - height_c
use = np.logical_and(caObj.avhrr.all_arrays['ctth_height']>-8,
caObj.calipso.all_arrays['layer_top_altitude'][:,0]>-999)
#use = np.logical_and(use,caObj.calipso.all_arrays['elevation']<10)
#use = np.logical_and(use,semi)
#use = np.logical_and(use,low_clouds)
#print "bias", np.mean(bias_nona[use]), "bias", np.mean(bias_l_rate_nona[use])
#caObj.calipso.all_arrays['number_layers_found'][:]>0)
print "low, medium, high", filename
for clouds in [low_clouds, medium_clouds, high_clouds, use]:
use_this_type = np.logical_and(use,clouds)
if not use_this_type.any():
continue
print "N %d", len(bias[use_this_type])
print "bias old %d"%(np.mean(bias[use_this_type])),
print "bias %d"%(np.mean(bias_l_rate[use_this_type]))
print "std old %d"%(np.std(bias[use_this_type])),
print "std %d"%(np.std(bias_l_rate[use_this_type]))
print "MAE old %d"%(np.mean(np.abs(bias[use_this_type]))),
print "MAE %d"%(np.mean(np.abs(bias_l_rate[use_this_type])))
print "RMS old %d"%(np.sqrt(np.mean((bias[use_this_type])**2))),
print "RMS %d"%(np.sqrt(np.mean((bias_l_rate[use_this_type])**2)))
print "N %d", len(bias[use_this_type])
bias_p =bias/height_c
bias_l_rate_p =bias_l_rate/height_c
print "percent bias old %1.2f"%(np.mean(bias_p[use_this_type])),
print "percent bias %1.2f"%(np.mean(bias_l_rate_p[use_this_type]))
print "percent std old %1.2f"%(np.std(bias_p[use_this_type])),
print "percent std %1.2f"%(np.std(bias_l_rate_p[use_this_type]))
print "percent MAE old %1.2f"%(np.mean(np.abs(bias_p[use_this_type]))),
print "percent MAE %1.2f"%(np.mean(np.abs(bias_l_rate_p[use_this_type])))
print "percent RMS old %1.2f"%(np.sqrt(np.mean((bias_p[use_this_type])**2))),
print "percent RMS %1.2f"%(np.sqrt(np.mean((bias_l_rate_p[use_this_type])**2)))
if len(bias[use_this_type])>1 and 1==2:
print "n", len(bias[use_this_type]), "%d/"%(np.mean(bias[use_this_type])),
print "%d"%(np.mean(bias_l_rate[use_this_type])),
print "%d/"%(np.mean(bias[np.logical_and(temp_diff>0, use_this_type)])),
print "%d"%(np.mean(bias_l_rate[np.logical_and(temp_diff>0, use_this_type)])),
print "%0.2f"%(np.mean(d45[use_this_type])),
print "%d %d %d %d"%(np.percentile(new_pps_h[use_this_type],10),
np.percentile(new_pps_h[use_this_type],90),
np.percentile(height_pps[use_this_type],10),
np.percentile(height_pps[use_this_type],90))
from cloudsat_calipso_avhrr_plot import (drawCalClsatGEOPROFAvhrrPlot,
drawCalPPSHeightPlot_PrototypePPSHeight )
"""
drawCalClsatGEOPROFAvhrrPlot(None,
caObj.calipso,
None,
use,
None,
caObj.calipso.layer_base_altitude*1000,
caObj.calipso.layer_top_altitude*1000,
use,
new_pps_h + caObj.calipso.all_arrays['elevation'],
"/home/a001865/CTTH_LAPSE_RATE_INVESTIGATION/",
"test_plot_file_orig",
'BASIC',
emissfilt_calipso_ok=None,
file_type='png',
instrument='modis')
"""
"""
def make_boxplot_pressure(caObj, name, modis_lvl2=False):
low_clouds = get_calipso_low_clouds(caObj)
high_clouds = get_calipso_high_clouds(caObj)
medium_clouds = get_calipso_medium_clouds(caObj)
pressure_c = caObj.calipso.all_arrays['layer_top_pressure'][:, 0]
if modis_lvl2:
pressure_pps = caObj.modis.all_arrays['pressure']
else:
pressure_pps = 0.01 * caObj.avhrr.all_arrays['ctth_pressure']
if modis_lvl2:
height_pps = caObj.modis.all_arrays['height']
else:
height_pps = caObj.avhrr.all_arrays['ctth_height']
thin = np.logical_and(
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] <
0.30,
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] >
0)
very_thin = np.logical_and(
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] <
0.10,
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] >
0)
thin_top = np.logical_and(
caObj.calipso.all_arrays['number_layers_found'] > 1, thin)
thin_1_lay = np.logical_and(
caObj.calipso.all_arrays['number_layers_found'] == 1, thin)
use = np.logical_and(
pressure_pps > 0,
caObj.calipso.all_arrays['layer_top_altitude'][:, 0] >= 0)
low = np.logical_and(low_clouds, use)
medium = np.logical_and(medium_clouds, use)
high = np.logical_and(high_clouds, use)
c_all = np.logical_or(high, np.logical_or(low, medium))
high_very_thin = np.logical_and(high, very_thin)
high_thin = np.logical_and(high, np.logical_and(~very_thin, thin))
high_thick = np.logical_and(high, ~thin)
#print "thin, thick high", np.sum(high_thin), np.sum(high_thick)
bias = pressure_pps - pressure_c
abias = np.abs(bias)
#abias[abias>2000]=2000
print name.ljust(30, " "), "%3.1f" % (np.mean(
abias[c_all])), "%3.1f" % (np.mean(abias[low])), "%3.1f" % (np.mean(
abias[medium])), "%3.1f" % (np.mean(abias[high]))
c_all = np.logical_or(np.logical_and(~very_thin, high),
np.logical_or(low, medium))
number_of = np.sum(c_all)
#print name.ljust(30, " "), "%3.1f"%(np.sum(abias[c_all]<250)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<500)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<1000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<1500)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<2000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<3000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<4000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<5000)*100.0/number_of)
from matplotlib import rcParams
rcParams.update({'figure.autolayout': True})
fig = plt.figure(figsize=(6, 9))
ax = fig.add_subplot(111)
plt.xticks(rotation=70)
ax.fill_between(np.arange(0, 8), -50, 50, facecolor='green', alpha=0.6)
ax.fill_between(np.arange(0, 8), -100, 100, facecolor='green', alpha=0.4)
ax.fill_between(np.arange(0, 8), -150, 150, facecolor='green', alpha=0.2)
ax.fill_between(np.arange(0, 8), 200, 2000, facecolor='red', alpha=0.2)
ax.fill_between(np.arange(0, 8), -2000, -200, facecolor='red', alpha=0.2)
for y_val in [-6, -4, -2, 2, 4, 6, 8, -8]:
plt.plot(np.arange(0, 8),
y_val * 100 + 0 * np.arange(0, 8),
':k',
alpha=0.4)
plt.plot(np.arange(0, 8), 0 + 0 * np.arange(0, 8), ':k', alpha=0.4)
bplot = ax.boxplot([
bias[low], bias[medium], bias[high], bias[high_thick], bias[high_thin],
bias[high_very_thin]
],
whis=[5, 95],
sym='',
labels=[
"low", "medium", "high-all", "high-thick\n od>0.4",
"high-thin \n 0.1<od<0.4", "high-vthin\n od<0.1"
],
showmeans=True,
patch_artist=True)
ax.set_ylim(-1000, 800)
for box in bplot['boxes']:
box.set_facecolor('0.9')
plt.title(name)
plt.savefig(
"/home/a001865/PICTURES_FROM_PYTHON/CTTH_BOX/ctth_box_plot_pressure_%s_5_95_filt.png"
% (name))
def make_boxplot(caObj, name, month="xx", modis_lvl2=False, use_m2_pix=True):
low_clouds = get_calipso_low_clouds(caObj)
high_clouds = get_calipso_high_clouds(caObj)
medium_clouds = get_calipso_medium_clouds(caObj)
height_c = 1000 * caObj.calipso.all_arrays['layer_top_altitude'][:, 0]
cloud_elevation = 1000 * caObj.calipso.all_arrays[
'layer_top_altitude'][:, 0] - caObj.calipso.all_arrays['elevation']
if modis_lvl2:
height_imager = caObj.modis.all_arrays['height']
else:
height_imager = caObj.avhrr.all_arrays[
'imager_ctth_m_above_seasurface']
if height_imager is None:
height_imager = caObj.avhrr.all_arrays[
'ctth_height'] + caObj.calipso.all_arrays['elevation']
use = np.logical_and(height_imager > -1, height_c >= 0)
use = np.logical_and(height_imager < 45000, use)
USE_ONLY_PIXELS_WHERE_PPS_AND_MODIS_C6_HAVE_VALUES = use_m2_pix
if USE_ONLY_PIXELS_WHERE_PPS_AND_MODIS_C6_HAVE_VALUES:
height_mlvl2 = caObj.modis.all_arrays['height']
height_pps = caObj.avhrr.all_arrays['imager_ctth_m_above_seasurface']
use = np.logical_and(use, height_mlvl2 > -1)
use = np.logical_and(use, height_mlvl2 < 45000)
use = np.logical_and(use, height_pps > -1)
use = np.logical_and(use, height_pps < 45000)
thin = np.logical_and(
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] <
0.30,
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] >
0)
very_thin = np.logical_and(
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] <
0.10,
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] >
0)
thin_top = np.logical_and(
caObj.calipso.all_arrays['number_layers_found'] > 1, thin)
thin_1_lay = np.logical_and(
caObj.calipso.all_arrays['number_layers_found'] == 1, thin)
low = np.logical_and(low_clouds, use)
medium = np.logical_and(medium_clouds, use)
high = np.logical_and(high_clouds, use)
c_all = np.logical_or(high, np.logical_or(low, medium))
high_very_thin = np.logical_and(high, very_thin)
high_thin = np.logical_and(high, np.logical_and(~very_thin, thin))
high_thick = np.logical_and(high, ~thin)
#print "thin, thick high", np.sum(high_thin), np.sum(high_thick)
bias = height_imager - height_c
abias = np.abs(bias)
#abias[abias>2000]=2000
print name.ljust(30, " "), "%3.1f" % (np.mean(
abias[c_all])), "%3.1f" % (np.mean(abias[low])), "%3.1f" % (np.mean(
abias[medium])), "%3.1f" % (np.mean(abias[high]))
c_all = np.logical_or(np.logical_and(~very_thin, high),
np.logical_or(low, medium))
number_of = np.sum(c_all)
MAE = np.mean(abias[c_all])
#print name.ljust(30, " "), "%3.1f"%(np.sum(abias[c_all]<250)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<500)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<1000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<1500)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<2000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<3000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<4000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<5000)*100.0/number_of)
from matplotlib import rcParams
rcParams.update({'figure.autolayout': True})
fig = plt.figure(figsize=(6, 9))
ax = fig.add_subplot(111)
plt.xticks(rotation=70)
ax.fill_between(np.arange(0, 8), -500, 500, facecolor='green', alpha=0.6)
ax.fill_between(np.arange(0, 8), -1000, 1000, facecolor='green', alpha=0.4)
ax.fill_between(np.arange(0, 8), -1500, 1500, facecolor='green', alpha=0.2)
ax.fill_between(np.arange(0, 8), 2000, 15000, facecolor='red', alpha=0.2)
ax.fill_between(np.arange(0, 8), -2000, -15000, facecolor='red', alpha=0.2)
for y_val in [-5, -4, -3, -2, 2, 3, 4, 5]:
plt.plot(np.arange(0, 8),
y_val * 1000 + 0 * np.arange(0, 8),
':k',
alpha=0.4)
plt.plot(np.arange(0, 8),
-10 * 1000 + 0 * np.arange(0, 8),
':k',
alpha=0.4)
plt.plot(np.arange(0, 8), 0 + 0 * np.arange(0, 8), ':k', alpha=0.4)
bplot = ax.boxplot([
bias[low], bias[medium], bias[high], bias[high_thick], bias[high_thin],
bias[high_very_thin]
],
whis=[5, 95],
sym='',
labels=[
"low", "medium", "high-all", "high-thick\n od>0.4",
"high-thin \n 0.1<od<0.4", "high-vthin\n od<0.1"
],
showmeans=True,
patch_artist=True)
ax.set_ylim(-14000, 8000)
for box in bplot['boxes']:
box.set_facecolor('0.9')
plt.title("%s MAE = %3.0f" % (name, MAE))
plt.savefig(
"/home/a001865/PICTURES_FROM_PYTHON/CTTH_BOX/ctth_box_plot_%s_5_95_filt.png"
% (name))
elevation_zero = np.logical_and(
use, caObj.calipso.all_arrays['elevation'] > 5000)
low_clouds = height_c < 2500
medium_clouds = np.logical_and(height_c >= 2500, height_c <= 5000)
high_clouds = height_c > 5000
low = np.logical_and(low_clouds, use)
medium = np.logical_and(medium_clouds, use)
high = np.logical_and(high_clouds, use)
fig = plt.figure(figsize=(6, 9))
ax = fig.add_subplot(111)
plt.xticks(rotation=50)
ax.fill_between(np.arange(0, 8), -500, 500, facecolor='green', alpha=0.6)
ax.fill_between(np.arange(0, 8), -1000, 1000, facecolor='green', alpha=0.4)
ax.fill_between(np.arange(0, 8), -1500, 1500, facecolor='green', alpha=0.2)
ax.fill_between(np.arange(0, 8), 2000, 15000, facecolor='red', alpha=0.2)
ax.fill_between(np.arange(0, 8), -2000, -15000, facecolor='red', alpha=0.2)
for y_val in [-5, -4, -3, -2, 2, 3, 4, 5]:
plt.plot(np.arange(0, 8),
y_val * 1000 + 0 * np.arange(0, 8),
':k',
alpha=0.4)
plt.plot(np.arange(0, 8),
-10 * 1000 + 0 * np.arange(0, 8),
':k',
alpha=0.4)
plt.plot(np.arange(0, 8), 0 + 0 * np.arange(0, 8), ':k', alpha=0.4)
bplot = ax.boxplot(
[bias[low], bias[medium], bias[high], bias[elevation_zero]],
whis=[5, 95],
sym='',
labels=["low <2.5km", "medium", "high>5km", "ground>5km"],
showmeans=True,
patch_artist=True)
ax.set_ylim(-8000, 8000)
for box in bplot['boxes']:
box.set_facecolor('0.9')
plt.title("Calipso %s \nHeight bias comparison MAE= %3.0f" % (name, MAE))
plt.savefig(
"/home/a001865/PICTURES_FROM_PYTHON/CTTH_BOX/ctth_box_plot_hkm_%s_5_95_filt.png"
% (name))
def extract_data(cObj, sat='cloudsat'):
pltObj = PlotAndDataObject()
if sat.lower() in 'cloudsat':
pltObj.height_c2 = cObj.cloudsat.all_arrays['clsat_max_height']
clsat_max_height = -9 + 0 * np.zeros(cObj.cloudsat.latitude.shape)
for i in range(125):
height = cObj.cloudsat.Height[:, i]
cmask_ok = cObj.cloudsat.CPR_Cloud_mask[:, i]
top_height = height + 120
#top_height[height<240*4] = -9999 #Do not use not sure why these are not used Nina 20170317
is_cloudy = cmask_ok > 30
top_height[~is_cloudy] = -9999
clsat_max_height[clsat_max_height < top_height] = top_height[
clsat_max_height < top_height]
height_c = clsat_max_height
pltObj.low_clouds = np.logical_and(
height_c < cObj.avhrr.all_arrays['segment_nwp_h680'],
height_c > -9)
pltObj.medium_clouds = np.logical_and(
height_c >= cObj.avhrr.all_arrays['segment_nwp_h680'],
height_c <= cObj.avhrr.all_arrays['segment_nwp_h440'])
pltObj.high_clouds = np.logical_and(
height_c > cObj.avhrr.all_arrays['segment_nwp_h440'],
height_c > -9)
elevation = cObj.cloudsat.all_arrays['elevation']
elevation[elevation < 0] = 0
elif sat.lower() in 'calipso':
height_c = 1000 * cObj.calipso.all_arrays['layer_top_altitude'][:, 0]
pltObj.low_clouds = get_calipso_low_clouds(cObj)
pltObj.high_clouds = get_calipso_high_clouds(cObj)
pltObj.medium_clouds = get_calipso_medium_clouds(cObj)
elevation = cObj.calipso.all_arrays['elevation']
elevation[elevation < 0] = 0
use_part = False
part = "all"
if use_part:
part = "single_layer_sea_below_60_in5km_as_art"
use_part = np.logical_and(
use_part, cObj.calipso.all_arrays['number_layers_found'] == 1)
use_part = np.logical_and(
use_part,
np.abs(cObj.calipso.all_arrays['latitude']) < 60)
use = np.logical_and(use_part,
np.equal(cObj.calipso.igbp_surface_type, 17))
#use = np.logical_and(use_part,height_c<3000+cObj.calipso.all_arrays['elevation'])
use_part = np.logical_and(
use_part, cObj.calipso.
all_arrays['feature_optical_depth_532_top_layer_5km'] > 0)
use_part = np.logical_and(
use_part, cObj.calipso.
all_arrays['feature_optical_depth_532_top_layer_5km'] ==
cObj.calipso.all_arrays['total_optical_depth_5km'])
part = "single_layer_sea_below_60_in5km_all_od_top"
low = np.logical_and(low_clouds, use_part)
#low = np.logical_and(height_c<low_clouds,use_part)
low = np.logical_and(
use_part, height_c <
(3000 + cObj.calipso.all_arrays['elevation']))
medium = np.logical_and(medium_clouds, use_part)
high = np.logical_and(height_c > 8000, use_part)
pltObj.height_c = height_c
pltObj.height_mlvl2 = cObj.modis.all_arrays['height'] #+elevation #??
#height_pps = cObj.avhrr.all_arrays['imager_ctth_m_above_seasurface']
pltObj.height_pps = cObj.avhrr.all_arrays['ctthnnant_height'] + elevation
pltObj.height_old = cObj.avhrr.all_arrays['ctthold_height'] + elevation
pltObj.pps_bias = pltObj.height_pps - height_c
pltObj.old_bias = pltObj.height_old - height_c
pltObj.mlvl2_bias = pltObj.height_mlvl2 - height_c
use = height_c >= 0
use = np.logical_and(use, pltObj.height_mlvl2 > -1)
use = np.logical_and(use,
cObj.modis.all_arrays['pressure'] >= 70) #no_effekt
use = np.logical_and(use, cObj.avhrr.all_arrays['ctthnnant_height'] > -1)
use = np.logical_and(use, cObj.avhrr.all_arrays['ctthold_height'] > -1)
#use = np.logical_and(use,cObj.avhrr.all_arrays['ctthnnant_height']<55000)no_effekt
use = np.logical_and(use,
cObj.avhrr.all_arrays['ctthold_pressure'] >= 70 * 100)
use = np.logical_and(use, cObj.modis.all_arrays['cloud_emissivity'] <= 100)
pltObj.use = use
use_all = use.copy()
for var in [
'ctthnnant_height', 'ctthnna1nt_height', 'ctthnnvnt_height',
'ctthnnm2nt_height', 'ctthnnmintnco2_height', 'ctthnnmint_height',
'ctthold_height'
]:
name = var.replace('_height', '').replace('ctth', 'bias_')
pltObj.all_arrays[
name] = cObj.avhrr.all_arrays[var] + elevation - height_c
#ok_ctth = np.logical_and(cObj.avhrr.all_arrays[var]<550000, cObj.avhrr.all_arrays[var]>-1)
ok_ctth = cObj.avhrr.all_arrays[var] > -1
use_all = np.logical_and(use_all, ok_ctth)
if 'nnmi' in var:
ok_ctth = np.logical_and(ok_ctth,
cObj.avhrr.all_arrays['modis_27'] > -1)
if 'nna1' in var:
ok_ctth = np.logical_and(ok_ctth,
cObj.avhrr.all_arrays['bt37micron'] > -1)
if 'nnv' in var or 'nnm' in var:
ok_ctth = np.logical_and(ok_ctth,
cObj.avhrr.all_arrays['bt86micron'] > -1)
if 'nna1' not in var:
ok_ctth = np.logical_and(ok_ctth,
cObj.avhrr.all_arrays['bt12micron'] > -1)
if 'nnm' in var:
ok_ctth = np.logical_and(ok_ctth,
cObj.avhrr.all_arrays['modis_28'] > -1)
if 'ctthnnmint_height' == var:
ok_ctth = np.logical_and(ok_ctth,
cObj.avhrr.all_arrays['modis_33'] > -1)
ok_ctth = np.logical_and(ok_ctth,
cObj.avhrr.all_arrays['bt11micron'] > -1)
use_all = np.logical_and(use_all, ok_ctth)
use_all = np.logical_and(use_all,
cObj.avhrr.all_arrays['modis_27'] > -1)
use_all = np.logical_and(use_all,
cObj.avhrr.all_arrays['modis_28'] > -1)
use_all = np.logical_and(use_all,
cObj.avhrr.all_arrays['modis_33'] > -1)
use_all = np.logical_and(use_all,
cObj.avhrr.all_arrays['text_t37'] > -9)
use_all = np.logical_and(use_all,
cObj.avhrr.all_arrays['warmest_t37'] > -9)
use_all = np.logical_and(use_all,
cObj.avhrr.all_arrays['coldest_t37'] > -9)
use_all = np.logical_and(use_all,
cObj.avhrr.all_arrays['bt11micron'] > -1)
use_all = np.logical_and(use_all,
cObj.avhrr.all_arrays['bt12micron'] > -1)
use_all = np.logical_and(use_all,
cObj.avhrr.all_arrays['bt37micron'] > -1)
use_all = np.logical_and(use_all,
cObj.avhrr.all_arrays['bt86micron'] > -1)
use_all = np.logical_and(use_all, cObj.avhrr.all_arrays['ciwv'] > -9)
pltObj.all_arrays[name.replace('bias_', 'ok_')] = ok_ctth
pltObj.use_all = use_all
pltObj.ok_cma_mlvl2 = cObj.modis.all_arrays['cloud_emissivity'] <= 100
pltObj.ok_cma_pps = np.logical_or(cObj.avhrr.all_arrays['cloudmask'] == 1,
cObj.avhrr.all_arrays['cloudmask'] == 2)
pltObj.ok_cma_pps = np.logical_and(pltObj.ok_cma_pps, height_c >= 0)
pltObj.ok_cma_mlvl2 = np.logical_and(pltObj.ok_cma_mlvl2, height_c >= 0)
pltObj.ok_modis = np.logical_and(pltObj.height_mlvl2 > -1,
pltObj.height_mlvl2 < 45000)
pltObj.ok_cma = np.logical_and(pltObj.ok_modis, pltObj.ok_cma_pps)
return pltObj
def make_boxplot(caObj, name):
low_clouds = get_calipso_low_clouds(caObj)
high_clouds = get_calipso_high_clouds(caObj)
medium_clouds = get_calipso_medium_clouds(caObj)
height_c = (1000*caObj.calipso.all_arrays['layer_top_altitude'][:,0] -
caObj.calipso.all_arrays['elevation'])
height_c1 = (1000*caObj.calipso.all_arrays['layer_top_altitude'][:,0] -
caObj.calipso.all_arrays['elevation'])
height_c2 = (1000*caObj.calipso.all_arrays['layer_top_altitude'][:,1] -
caObj.calipso.all_arrays['elevation'])
height_c3 = (1000*caObj.calipso.all_arrays['layer_top_altitude'][:,2] -
caObj.calipso.all_arrays['elevation'])
height_c4 = (1000*caObj.calipso.all_arrays['layer_top_altitude'][:,3] -
caObj.calipso.all_arrays['elevation'])
height_pps = caObj.avhrr.all_arrays['ctth_height']
bias_1 = height_pps - height_c1
bias_2 = height_pps - height_c2
bias_3 = height_pps - height_c3
bias_4 = height_pps - height_c4
thin = np.logical_and(caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km']<0.30,
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km']>0)
very_thin = np.logical_and(caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km']<0.10,
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km']>0)
thin_top = np.logical_and(caObj.calipso.all_arrays['number_layers_found']>1, thin)
thin_1_lay = np.logical_and(caObj.calipso.all_arrays['number_layers_found']==1, thin)
#height_c[thin_top] = height_c2[thin_top]
#height_c[np.abs(bias_1)<np.abs(bias_2)] = height_c1[np.abs(bias_1)<np.abs(bias_2)]
#height_c[np.abs(bias_2)<np.abs(bias_1)] = height_c2[np.abs(bias_2)<np.abs(bias_1)]
#bias = height_pps - height_c
#height_c[np.abs(bias_3)<np.abs(bias)] = height_c3[np.abs(bias_3)<np.abs(bias)]
#height_c[~thin_top] = height_c1[~thin_top]
#height_c[thin_top] = height_c2[thin_top]
use = np.logical_and(height_pps >-1,
caObj.calipso.all_arrays['layer_top_altitude'][:,0]>=0)
use = np.logical_and(height_pps <45000,use)
low = np.logical_and(low_clouds,use)
medium = np.logical_and(medium_clouds,use)
high = np.logical_and(high_clouds,use)
c_all = np.logical_or(high,np.logical_or(low,medium))
high_very_thin = np.logical_and(high, very_thin)
high_thin = np.logical_and(high, np.logical_and(~very_thin,thin))
high_thick = np.logical_and(high, ~thin)
#print "thin, thick high", np.sum(high_thin), np.sum(high_thick)
bias = height_pps - height_c
limit = np.percentile(bias[use],5)
#print limit
abias = np.abs(bias)
MAE = np.mean(abias[c_all])
#abias[abias>2000]=2000
print name.ljust(30, " "), "%3.1f"%(np.mean(abias[c_all])), "%3.1f"%(np.mean(abias[low])),"%3.1f"%(np.mean(abias[medium])),"%3.1f"%(np.mean(abias[high])), "%3.1f"%(limit)
c_all = np.logical_or(np.logical_and(~very_thin,high),np.logical_or(low,medium))
number_of = np.sum(c_all)
#print name.ljust(30, " "), "%3.1f"%(np.sum(abias[c_all]<250)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<500)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<1000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<1500)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<2000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<3000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<4000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<5000)*100.0/number_of)
from matplotlib import rcParams
rcParams.update({'figure.autolayout': True})
fig = plt.figure(figsize = (6,9))
ax = fig.add_subplot(111)
plt.xticks(rotation=70)
#plt.tight_layout()
#plt.subplots_adjust(left=0.2)
#plt.subplots_adjust(left=10, bottom=10, right=10, top=10, wspace=0, hspace=0)
ax.fill_between(np.arange(0,8),-500,500, facecolor='green', alpha=0.6)
ax.fill_between(np.arange(0,8),-1000,1000, facecolor='green', alpha=0.4)
ax.fill_between(np.arange(0,8),-1500,1500, facecolor='green', alpha=0.2)
ax.fill_between(np.arange(0,8),2000,15000, facecolor='red', alpha=0.2)
ax.fill_between(np.arange(0,8),-2000,-15000, facecolor='red', alpha=0.2)
for y_val in [-5,-4,-3,-2,2,3,4,5]:
plt.plot(np.arange(0,8), y_val*1000 + 0*np.arange(0,8),':k')
plt.plot(np.arange(0,8), -10*1000 + 0*np.arange(0,8),':k')
plt.plot(np.arange(0,8), 0 + 0*np.arange(0,8),'k')
plt.boxplot([bias[low],bias[medium],bias[high],bias[high_thick],bias[high_thin],bias[high_very_thin]],whis=[5, 95],sym='',
labels=["low","medium","high-all","high-thick\n od>0.4","high-thin \n 0.1<od<0.4","high-vthin\n od<0.1"],showmeans=True)
ax.set_ylim(-14000,8000)
plt.title("%s MAE = %3.0f"%(name,MAE))
plt.savefig("/home/a001865/PICTURES_FROM_PYTHON/CTTH_LAPSE_RATE_INVESTIGATION/ctth_box_plot_%s_5_95_filt.png"%(name))
def make_violinplot(caObj, name, modis_lvl2=False):
low_clouds = get_calipso_low_clouds(caObj)
high_clouds = get_calipso_high_clouds(caObj)
medium_clouds = get_calipso_medium_clouds(caObj)
height_c = (1000 * caObj.calipso.all_arrays['layer_top_altitude'][:, 0] -
caObj.calipso.all_arrays['elevation'])
if modis_lvl2:
height_pps = caObj.modis.all_arrays['height']
else:
height_pps = caObj.avhrr.all_arrays['ctth_height']
print "min/max height", np.min(height_pps), np.max(height_pps)
thin = np.logical_and(
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] <
0.30,
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] >
0)
very_thin = np.logical_and(
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] <
0.10,
caObj.calipso.all_arrays['feature_optical_depth_532_top_layer_5km'] >
0)
thin_top = np.logical_and(
caObj.calipso.all_arrays['number_layers_found'] > 1, thin)
thin_1_lay = np.logical_and(
caObj.calipso.all_arrays['number_layers_found'] == 1, thin)
use = np.logical_and(
height_pps > -1,
caObj.calipso.all_arrays['layer_top_altitude'][:, 0] >= 0)
use = np.logical_and(height_pps < 45000, use)
low = np.logical_and(low_clouds, use)
medium = np.logical_and(medium_clouds, use)
high = np.logical_and(high_clouds, use)
c_all = np.logical_or(high, np.logical_or(low, medium))
high_very_thin = np.logical_and(high, very_thin)
high_thin = np.logical_and(high, np.logical_and(~very_thin, thin))
high_thick = np.logical_and(high, ~thin)
#print "thin, thick high", np.sum(high_thin), np.sum(high_thick)
bias = height_pps - height_c
abias = np.abs(bias)
#abias[abias>2000]=2000
print name.ljust(30, " "), "%3.1f" % (np.mean(
abias[c_all])), "%3.1f" % (np.mean(abias[low])), "%3.1f" % (np.mean(
abias[medium])), "%3.1f" % (np.mean(abias[high]))
c_all = np.logical_or(np.logical_and(~very_thin, high),
np.logical_or(low, medium))
number_of = np.sum(c_all)
#print name.ljust(30, " "), "%3.1f"%(np.sum(abias[c_all]<250)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<500)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<1000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<1500)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<2000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<3000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<4000)*100.0/number_of), "%3.1f"%(np.sum(abias[c_all]<5000)*100.0/number_of)
from matplotlib import rcParams
rcParams.update({'figure.autolayout': True})
fig = plt.figure(figsize=(6, 9))
ax = fig.add_subplot(111)
plt.xticks(rotation=70)
ax.fill_between(np.arange(0, 8), -1500, 1500, facecolor='green', alpha=0.2)
ax.fill_between(np.arange(0, 8), 2000, 20000, facecolor='red', alpha=0.2)
ax.fill_between(np.arange(0, 8), -2000, -20000, facecolor='red', alpha=0.2)
for y_val in [-5, -4, -3, -2, 2, 3, 4, 5]:
plt.plot(np.arange(0, 8),
y_val * 1000 + 0 * np.arange(0, 8),
':k',
alpha=0.4)
plt.plot(np.arange(0, 8),
-10 * 1000 + 0 * np.arange(0, 8),
':k',
alpha=0.4)
plt.plot(np.arange(0, 8), 0 + 0 * np.arange(0, 8), ':k', alpha=0.4)
bplot = ax.violinplot([
bias[low], bias[medium], bias[high], bias[high_thick], bias[high_thin],
bias[high_very_thin]
],
widths=0.9,
showextrema=False,
showmedians=True)
ax.set_ylim(-20000, 20000)
plt.title(name)
plt.savefig(
"/home/a001865/PICTURES_FROM_PYTHON/CTTH_BOX/ctth_violin_%s_5_95_filt.png"
% (name))