Skip to content

Commit

Permalink
update figures
Browse files Browse the repository at this point in the history
  • Loading branch information
@Lkruitwagen
Lkruitwagen committed Jun 30, 2021
1 parent 2619674 commit f4ec421
Show file tree
Hide file tree
Showing 2 changed files with 86 additions and 5 deletions.
91 changes: 86 additions & 5 deletions makefigs/fig-A8_band_perturbation.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -29,12 +29,9 @@ def hex2rgb(h):
df_S2 = pd.read_csv(os.path.join(os.getcwd(),'data','band_perturbation','band_perturbation_S2.csv')).sort_values('records').set_index(['records','bands'])
df_SPOT = pd.read_csv(os.path.join(os.getcwd(),'data','band_perturbation','band_perturbation_SPOT.csv'))

root = os.getcwd()

# prep SPOT df




idx = pd.IndexSlice

#all_bands = list(df.index.get_level_values(1).unique())
Expand Down Expand Up @@ -94,10 +91,25 @@ def hex2rgb(h):
axs['SPOT'] = {'bdo':fig.add_subplot(gs[0,7:]),
'additive':fig.add_subplot(gs[1,7:]),
'multiplicative':fig.add_subplot(gs[2,7:])}

data_out = {}

# do band dropout
data = []
for ii_b, band in enumerate(S2_bands):
data.append(df_S2.loc[idx[:,band['name']],'band_dropout'].values.clip(-1,1))
el = df_S2.loc[idx[:,band['name']],'band_dropout'].values.clip(-1,1)
print ('band')
print (band)
data_out[band['name']] = {
'mean':el.mean(),
'Q20':np.percentile(el,20),
'Q80':np.percentile(el,80),
'min':el.min(),
'max':el.max(),
}
df_out = pd.DataFrame(data_out)
df_out.to_csv(os.path.join(root,'makefigs','data','fig-A8_S2_bands.csv'))

bplot0S2 = axs['S2']['bdo'].boxplot(data, whis='range',patch_artist=True, medianprops = dict(linestyle='-', linewidth=1, color='firebrick'))
axs['S2']['bdo'].set_ylabel('IoU Impairment')
Expand All @@ -107,8 +119,19 @@ def hex2rgb(h):
axs['S2']['bdo'].yaxis.set_major_formatter(mtick.PercentFormatter(1))

data = []
data_out = {}
for ii_b, band in enumerate(SPOT_bands):
data.append(df_SPOT[band['name']+'_bdo'].values.clip(-1,1))
el = df_SPOT[band['name']+'_bdo'].values.clip(-1,1)
data_out[band['name']] = {
'mean':el.mean(),
'Q20':np.percentile(el,20),
'Q80':np.percentile(el,80),
'min':el.min(),
'max':el.max()
}
df_out = pd.DataFrame(data_out)
df_out.to_csv(os.path.join(root,'makefigs','data','fig-A8_SPOT_bands.csv'))

bplot0SPOT = axs['SPOT']['bdo'].boxplot(data, whis='range',patch_artist=True, medianprops = dict(linestyle='-', linewidth=1, color='firebrick'))
#axs['SPOT']['bdo'].set_ylabel('IoU Impairment')
Expand All @@ -121,11 +144,25 @@ def hex2rgb(h):
# do additive 1,2,3
data = []
positions=[]
data_out = {}
for ii_b, band in enumerate(S2_bands):
data.append(df_S2.loc[idx[:,band['name']],'additive_0.1'].values.clip(-1,1))
data.append(df_S2.loc[idx[:,band['name']],'additive_0.2'].values.clip(-1,1))
data.append(df_S2.loc[idx[:,band['name']],'additive_0.3'].values.clip(-1,1))
positions += [(ii_b+1 + (ii_p-1)/4) for ii_p in range(3)]
el1 = df_S2.loc[idx[:,band['name']],'additive_0.1'].values.clip(-1,1)
el2 = df_S2.loc[idx[:,band['name']],'additive_0.2'].values.clip(-1,1)
el3 = df_S2.loc[idx[:,band['name']],'additive_0.3'].values.clip(-1,1)
for el, label in zip([el1,el2,el3],['_additive_0.1','_additive_0.2','_additive_0.3']):
data_out[band['name']+label] = {
'mean':el.mean(),
'Q20':np.percentile(el,20),
'Q80':np.percentile(el,80),
'min':el.min(),
'max':el.max()
}
df_out = pd.DataFrame(data_out)
df_out.to_csv(os.path.join(root,'makefigs','data','fig-A8_S2_additive.csv'))

bplot1S2 = axs['S2']['additive'].boxplot(data, positions=positions, whis='range', widths=0.15, patch_artist=True, medianprops = dict(linestyle='-', linewidth=1, color='firebrick'))
axs['S2']['additive'].set_ylabel('IoU Impairment')
Expand All @@ -137,11 +174,25 @@ def hex2rgb(h):

data = []
positions=[]
data_out = {}
for ii_b, band in enumerate(SPOT_bands):
data.append(df_SPOT[band['name']+'_additive_0.1'].values.clip(-1,1))
data.append(df_SPOT[band['name']+'_additive_0.2'].values.clip(-1,1))
data.append(df_SPOT[band['name']+'_additive_0.3'].values.clip(-1,1))
positions += [(ii_b+1 + (ii_p-1)/4) for ii_p in range(3)]
el1 = df_SPOT[band['name']+'_additive_0.1'].values.clip(-1,1)
el2 = df_SPOT[band['name']+'_additive_0.2'].values.clip(-1,1)
el3 = df_SPOT[band['name']+'_additive_0.3'].values.clip(-1,1)
for el, label in zip([el1,el2,el3],['_additive_0.1','_additive_0.2','_additive_0.3']):
data_out[band['name']+label] = {
'mean':el.mean(),
'Q20':np.percentile(el,20),
'Q80':np.percentile(el,80),
'min':el.min(),
'max':el.max()
}
df_out = pd.DataFrame(data_out)
df_out.to_csv(os.path.join(root,'makefigs','data','fig-A8_SPOT_additive.csv'))

bplot1SPOT = axs['SPOT']['additive'].boxplot(data, positions=positions, whis='range', widths=0.15, patch_artist=True, medianprops = dict(linestyle='-', linewidth=1, color='firebrick'))
axs['SPOT']['additive'].set_title('SPOT6/7')
Expand All @@ -156,11 +207,27 @@ def hex2rgb(h):
# do multiplicative 1,2,3
data = []
positions=[]
data_out = {}
for ii_b, band in enumerate(S2_bands):
data.append(df_S2.loc[idx[:,band['name']],'multiplicative_0.1'].values.clip(-1,1))
data.append(df_S2.loc[idx[:,band['name']],'multiplicative_0.2'].values.clip(-1,1))
data.append(df_S2.loc[idx[:,band['name']],'multiplicative_0.3'].values.clip(-1,1))
positions += [(ii_b+1 + (ii_p-1)/4) for ii_p in range(3)]
el1 = df_S2.loc[idx[:,band['name']],'multiplicative_0.1'].values.clip(-1,1)
el2 = df_S2.loc[idx[:,band['name']],'multiplicative_0.2'].values.clip(-1,1)
el3 = df_S2.loc[idx[:,band['name']],'multiplicative_0.3'].values.clip(-1,1)
for el, label in zip([el1,el2,el3],['_multiplicative_0.1','_multiplicative_0.2','_multiplicative_0.3']):
data_out[band['name']+label] = {
'mean':el.mean(),
'Q20':np.percentile(el,20),
'Q80':np.percentile(el,80),
'min':el.min(),
'max':el.max()
}
df_out = pd.DataFrame(data_out)
df_out.to_csv(os.path.join(root,'makefigs','data','fig-A8_S2_multiplicative.csv'))



bplot2S2 = axs['S2']['multiplicative'].boxplot(data, positions=positions, whis='range', widths=0.15,patch_artist=True, medianprops = dict(linestyle='-', linewidth=1, color='firebrick'))
axs['S2']['multiplicative'].set_xticks(range(1,15))
Expand All @@ -173,12 +240,26 @@ def hex2rgb(h):

data = []
positions=[]

data_out = {}
for ii_b, band in enumerate(SPOT_bands):
data.append(df_SPOT[band['name']+'_multiplicative_0.1'].values.clip(-1,1))
data.append(df_SPOT[band['name']+'_multiplicative_0.2'].values.clip(-1,1))
data.append(df_SPOT[band['name']+'_multiplicative_0.3'].values.clip(-1,1))
positions += [(ii_b+1 + (ii_p-1)/4) for ii_p in range(3)]
el1 = df_SPOT[band['name']+'_multiplicative_0.1'].values.clip(-1,1)
el2 = df_SPOT[band['name']+'_multiplicative_0.2'].values.clip(-1,1)
el3 = df_SPOT[band['name']+'_multiplicative_0.3'].values.clip(-1,1)
for el, label in zip([el1,el2,el3],['_multiplicative_0.1','_multiplicative_0.2','_multiplicative_0.3']):
data_out[band['name']+label] = {
'mean':el.mean(),
'Q20':np.percentile(el,20),
'Q80':np.percentile(el,80),
'min':el.min(),
'max':el.max()
}
df_out = pd.DataFrame(data_out)
df_out.to_csv(os.path.join(root,'makefigs','data','fig-A8_SPOT_multiplicative.csv'))


bplot2SPOT = axs['SPOT']['multiplicative'].boxplot(data, positions=positions, whis='range', widths=0.15,patch_artist=True, medianprops = dict(linestyle='-', linewidth=1, color='firebrick'))
axs['SPOT']['multiplicative'].set_xticks(range(1,5))
Expand Down
Binary file modified makefigs/figures/fig-3_land_cover_global.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.

0 comments on commit f4ec421

Please sign in to comment.