def plot_matrix(self, detector_nr):
plt.rcParams.update({'font.size': 22})
detector_view = self.geo_matrix[detector_nr].reshape(
(self.pre_pixel_nr, self.pre_pixel_nr))
print(detector_view[0, 0].shape)
print(detector_view.flatten()[25])
# plt.imshow(detector_view, cmap=get_cmap())
plt.imshow(detector_view,
norm=matplotlib.colors.SymLogNorm(1e-14),
cmap=get_cmap())
cbar = plt.colorbar() # format='%.0e')
cbar.ax.set_ylabel('Influence ($m^3$)', rotation=90, fontsize=22)
plt.xlabel("#pixel x")
plt.ylabel("#pixel z")
import numpy as np
import sys
import grid_from_lat_lon
from mpl_toolkits.basemap import Basemap
import matplotlib.pyplot as plt
import cmap
brcmap = cmap.get_cmap()
warm_dict = np.load('PIAI_12.npy', allow_pickle=True, encoding='bytes').item()
model_colors = ['#20ab2e', '#4042c7', '#bc87e8', '#ab2020']
models = list(warm_dict.keys())
model_names = ['Random Forest', 'SGB', 'MVLR', 'XGBoost']
regions = warm_dict[models[0]].keys()
areas = [str(b)[2:-1] for b in regions]
differences = {}
for model in models:
differences[model] = []
for model in models:
for area in regions:
differences[model].append(
np.nanmean(warm_dict[model][area][b'values']) * 100)
for i, model in enumerate(differences.keys()):
grid, lats, lons = grid_from_lat_lon.make_grid(differences[model], areas)
m = Basemap(projection='cyl',
llcrnrlat=-90,
urcrnrlat=90,
llcrnrlon=-180,
yslice = 0
xx,zz = np.meshgrid(x1,(z1-zhub),indexing='ij')
j = np.argmin(np.abs(yslice-y1))
u0=u0[:,j,:]
u = u[:,j,:]
w = w[:,j,:]
return xx,zz,u0,u,w
# ---
fig,ax = plt.subplots(nrows=2, ncols=2,
sharex=True, sharey=True,
figsize=(16,8))
#cmap = 'magma'
cmap,valuesOri=get_cmap(0,1.05)
levelsLines = np.sort([1.05,1.0,0.99,0.98,0.95,0.9,0.8,0.7,0.6,0.5,0.4,0.3,0.2,0.1,0])
for irow,suffix in enumerate(suffixlist):
for icol,CT in enumerate(CTlist):
xAD,zAD,uAD,wAD = loadAD(CT=CT)
xGP,zGP,u0,uGP,wGP = loadGP(CT=CT,suffix=suffix)
cm = ax[irow,icol].contourf(xAD,zAD,uAD,levels=levelsLines,cmap=cmap)
cnt = ax[irow,icol].contour(xAD,zAD,uAD,levels=levelsLines,
colors='k',linewidths=0.5)
cnt2 = ax[irow,icol].contour(xGP,zGP,uGP,levels=levelsLines,
colors='k',linewidths=0.5,linestyles='--')
txt = ax[irow,icol].text(0.025,0.95,suffix,
horizontalalignment='left',
verticalalignment='top',
transform=ax[irow,icol].transAxes)
def plot_year(predictor, X_vars, y_var, year, GCM=False):
global year_dir
import datetime
if GCM:
year_data = hadgem_comp.get_hadgem()
y_var = 'cf'
year = 1850
else:
year_data = np.load(year_dir + str(year) + '.npy',
allow_pickle=True).item()
pred_y = np.zeros(year_data[y_var][0].shape)
ssims = []
mean_diffs = []
for month in range(len(year_data['sst']))[10:]:
date = datetime.date(month=month + 1, year=year, day=1)
print(date.strftime('%B'), 'date')
for lat_row in range(len(year_data['sst'][month])):
for lon_tile in range(len(year_data['sst'][month][lat_row])):
xai = []
for var in X_vars:
xai.append(year_data[var][month][lat_row][lon_tile])
if True in np.isnan(xai):
pred_y[lat_row][lon_tile] = np.nan
else:
pred_y[lat_row][lon_tile] = predictor.predict([xai])[0]
print(xai)
land = np.isnan(pred_y)
year_data[y_var][month][land] = np.nan
over = (pred_y > np.max(year_data[y_var][month]))
pred_y[over] = np.max(year_data[y_var][month])
print(np.nanmean(pred_y), 'predicted mean')
print(np.nanmean(year_data[y_var][month]), 'actual mean')
mean_diffs.append(
np.nanmean(pred_y) - np.nanmean(year_data['cf'][month]))
plot_max = max([np.nanmax(pred_y), np.nanmax(year_data[y_var][month])])
plot_min = min([np.nanmin(pred_y), np.nanmin(year_data[y_var][month])])
plt.pcolormesh(pred_y)
plt.colorbar()
plt.show()
plt.subplot(121)
plt.pcolormesh(year_data['longitudes'],
year_data['latitudes'],
pred_y,
cmap='cividis',
vmin=plot_min,
vmax=plot_max)
plt.title('Predicted CF')
plt.colorbar()
plt.subplot(122)
plt.pcolormesh(year_data['longitudes'],
year_data['latitudes'],
pred_y - year_data[y_var][month],
cmap=cmap.get_cmap())
plt.title('Differenced CF')
plt.colorbar()
plt.suptitle('{} {}'.format(date.strftime('%B'), date.strftime('%Y')))
plt.savefig('GCM Test')
plt.show()
pred_y[np.isnan(pred_y)] = 0.
year_data[y_var][month][np.isnan(year_data[y_var][month])] = 0.
sys.exit()
similarity = ssim(
pred_y,
year_data[y_var][month],
data_range=max([np.amax(pred_y),
np.amax(year_data[y_var][month])]))
print(similarity)
ssims.append(similarity)
return ssims, mean_diffs