def blend_images(self, data): #data must have iterable 0th dimension
#pdb.set_trace()
blended = np.zeros([self.masky, self.maskx, 4])
if np.ndim(data) == 2:
composite = np.zeros([self.masky, self.maskx, 4])
color = cm.hsv(.1)
alpha = np.squeeze(data)
alpha[np.isnan(alpha)] = .001 #zap nans
alpha[alpha == 0] = .001
blended = rgba_blend(composite, color, alpha)
elif np.ndim(data) == 3:
composite = np.zeros([self.masky, self.maskx, 4, data.shape[0]])
composite[:] = .001
color_scale = np.arange(.1, 1, .9 / data.shape[0])
for ii in range(data.shape[0]
): # for each cluster build a unique color channel
toget = color_scale[ii]
color = cm.hsv(toget)
alpha = data[ii, :, :]
alpha[np.isnan(alpha)] = .001 #zap nans
alpha[alpha == 0] = .001
composite[:, :, :, ii] = rgba_blend(composite, color, alpha)
for ii in range(3): # for each color make weighted average
color = composite[:, :, ii, :]
alpha = composite[:, :, 3, :]
blended[:, :, ii] = np.average(color, axis=2, weights=alpha)
blended[:, :, 3] = np.nanmax(composite[:, :, 3, :], axis=2)
return blended
def compareTwoAnnualWeather(df1, lab1, df2, lab2, item, title):
"""
df1: pandas.dataframe
the dataframe of weather data converted from WTH or WTD
df2: pandas.dataframe
the dataframe of weather data converted from WTH or WTD
item: str
the type of measured climate; SRAD, TMAX, TMIN, RAIN is defined
"""
unit = {
'SRAD': 'MJ/m2',
'TMAX': 'celcius',
'TMIN': 'celcius',
'RAIN': 'mm'
}
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
ax.plot(np.arange(len(df1.index)),
df1.loc[:, item],
color=cm.hsv(1 / 2),
label=lab1)
ax.plot(np.arange(len(df2.index)),
df2.loc[:, item],
color=cm.hsv(2 / 2),
label=lab2)
plt.legend(loc='best')
#plt.xticks(np.arange(len(wthdf.index)), np.inspace(1, 365, num=5))
plt.title(title, fontsize=18)
plt.xlabel('Day of year', fontsize=16)
plt.ylabel(item + '(' + unit[item] + ')', fontsize=12)
plt.show()
def plot(x, y, labels, K, title):
fig = plt.figure(figsize=(6, 4))
subplot1 = fig.add_subplot(211)
subplot1.plot(x)
subplot1.set_xlabel("Time Steps", size=7)
subplot1.set_ylabel("Node 1 that sends\nsignals to Node 2", size=7)
subplot1.set_xlim(0, len(x))
plt.xticks(fontsize=7)
plt.yticks(fontsize=7)
plt.title(title, size=7)
subplot2 = fig.add_subplot(212)
subplot2.plot(y, c="g")
subplot2.set_xlabel("Time Steps", size=7)
subplot2.set_xlim(0, len(x))
subplot2.set_ylabel("Node 2 that recieves\nsignals from Node 1", size=7)
plt.xticks(fontsize=7)
plt.yticks(fontsize=7)
plt.subplots_adjust(hspace=0.5)
for i, color_i in enumerate(labels):
subplot1.axvspan(i,
i + 1,
facecolor=cm.hsv(float(color_i) / K),
alpha=0.2,
lw=0)
subplot2.axvspan(i,
i + 1,
facecolor=cm.hsv(float(color_i) / K),
alpha=0.2,
lw=0)
plt.show()
def visualizeWeather(wthdf, title):
"""
reference for xtick -> http://python-remrin.hatenadiary.jp/entry/2017/05/27/114816
wthdf : pandas.DataFrame
the dataframe which contains the information of weather data
title_bn_nn_an : str
the title of the graph which bn_nn_an condition you use.
"""
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
for i in range(3):
ax.plot(np.arange(len(wthdf.index)),
wthdf.iloc[:, i],
color=cm.hsv(i / 5),
label=wthdf.columns[i])
ax.bar(np.arange(len(wthdf.index)),
wthdf.iloc[:, 3],
color=cm.hsv(4 / 5),
label=wthdf.columns[3])
print('SRAD : ' + repr(mean(wthdf.loc[:, 'SRAD'])) + ', TMAX : ' +
repr(mean(wthdf.loc[:, 'TMAX'])) + ', TMIN : ' +
repr(mean(wthdf.loc[:, 'TMIN'])) + ', RAIN : ' +
repr(sum(wthdf.loc[:, 'RAIN'])))
plt.legend(loc='best')
#plt.xticks(np.arange(len(wthdf.index)), np.inspace(1, 365, num=5))
plt.title(title, fontsize=18)
plt.xlabel('Day of year', fontsize=16)
plt.ylabel(
'temperature(celcius), Solar radiation(MJ/m2), precipitation(mm)',
fontsize=12)
plt.show()
def draw_spectrum(power, psd, spectrum, shape, name, sym=2, inv=1):
#colors=[np.array(cm.jet(i)[:3])*255 for i in range(256)]
t0 = time.time()
psdmax = psd.max()
if psdmax <= 0.0:
im = Image.new('RGB', shape)
return np.array(im).astype(np.float32), (0, 0, 0, 0, 0)
data = normalize_spectrum(spectrum, psdmax, inv=inv)
colors = [np.array(cm.hsv(i)[:3]) * 255 for i in range(256)]
colordata = data.copy()
colordata = (1.0 - colordata)**(1 / 3.0)
psdmax = 10 * np.log10(psdmax)
if psdmax > -20:
colordata = np.choose(spectrum < -30, (colordata, colordata / 10))
else:
colordata = np.choose(spectrum < psdmax - 10,
(colordata, colordata / 10))
colordata -= colordata.min()
colordata /= colordata.max()
t0 = time.time() - t0
t1 = time.time()
#colordata*=255
xs, ys = data_to_circle(data, psd, sym=sym)
xs, ys = center_circle(xs, ys, shape)
t1 = time.time() - t1
t2 = time.time()
nsemi = xs.shape[0] / sym
colorads = get_radians(nsemi, 2 * pi * np.log2(nsemi))
t2 = time.time() - t2
t3 = time.time()
#colors=[colors[i] for i in colordata]
colors = get_colors(colors, colorads, colordata)
t3 = time.time() - t3
#print colordata.max(),colordata.min(),name
t4 = time.time()
im = Image.new('RGB', shape)
#draw=ImageDraw.Draw(im)
im = np.array(im).astype(np.float32)
for i in range(sym):
draw_points_interp(xs[i * nsemi:(i + 1) * nsemi],
ys[i * nsemi:(i + 1) * nsemi], colors, im)
t4 = time.time() - t4
#imsave(name,im.astype(np.uint8))
#im.save(name)
return np.array(im).astype(np.float32), (t0, t1, t2, t3, t4
) #.astype(np.float32)
def plot_all(self, dat1, dat2, dat3, dat4):
#pdb.set_trace()
fig = plt.figure(figsize=(12, 4))
fig.subplots_adjust(left=.05,
right=.95,
bottom=.05,
top=.9,
hspace=.05,
wspace=.05)
color_scale = np.arange(.1, 1, .9 / len(self.labels_l))
ax1 = fig.add_subplot(1, 4, 1, xticks=[], yticks=[])
for ii in range(
len(dat1)
): # dat1 is one longer tha dat2 because it contains unused centroids
dat1_vals = np.squeeze(np.asarray(dat1[ii].values()))
if ii == 0: # first entry is unused centroids
ccc.plot_class_centroids(self, dat1_vals, cm.gray_r(.3), ax1)
else:
ccc.plot_class_centroids(self, dat1_vals,
cm.hsv(color_scale[ii - 1]), ax1)
ax1.set_title('clustered centroids')
ax2 = fig.add_subplot(1, 4, 2, xticks=[], yticks=[])
composite = ccc.blend_images(self, dat2)
ax2.imshow(self.mask, cmap='gray_r', vmax=10, interpolation='nearest')
ax2.imshow(composite, interpolation='nearest')
ax2.set_title('clustered seed-pixels')
ax3 = fig.add_subplot(1, 4, 3, xticks=[], yticks=[])
ax3.imshow(dat3, interpolation='nearest')
ax3.set_title('projected correlation maps')
ax4 = fig.add_subplot(1, 4, 4, xticks=[], yticks=[])
ax4.imshow(dat4, interpolation='nearest')
ax4.set_title('correlation masks')
arr = np.asarray(arr)
arr = np.concatenate((arr, np.unique(adat["PREF"]).reshape(-1,1)), axis=1)
adat3 = pd.DataFrame(arr, index=np.arange(1, arr.shape[0]+1), columns=adat.columns)
#visualize ADAT
adat2 = adat3.iloc[:, [0,1,3]]
fig = plt.figure(figsize=(10, 6))
ax = fig.add_subplot(1,1,1)
name = ["observed", "Koshihikari", "SD1"]
n = 0
for i in range(len(adat2.columns)):
ax.bar(np.arange(len(adat2.index))+n, adat2.iloc[:, i], color=cm.hsv(i/3),
label=name[i], width=0.8/3)
n = n + 0.8/3
plt.legend(loc="best",fontsize=14)
plt.xticks(np.arange(len(adat2.index)+0.55), adat3["PREF"].values, rotation=45, fontsize=14)
plt.title("Comparison of anthesis date among observed, koshi_param and SD1_param", fontsize=18)
plt.xlabel("field name", fontsize=15)
plt.ylabel("Anthesis date", fontsize=15)
plt.ylim([min(adat2.iloc[:, 1]-10), max(adat2.iloc[:, 0]+10)])
ax.set_yticklabels(doylist2Date(ax.get_yticks() -2018000))
plt.setp(ax.get_yticklabels(), fontsize=14, rotation=45, visible=True)
plt.savefig("190313_rice_gencalc_png/190316_ADAT_validation_SD1.png", bbox_inches="tight")
plt.show()
"""
fig = plt.figure(figsize=(12, 6))
ax = fig.add_subplot(1, 1, 1)
order = sum_df.sort_values("HWAH").index.values.astype(np.int32)
#wnum = order[[1,50,80]]
wnum = [83, 97, 94]
for i in range(len(wnum)):
prep = searchElementforX(wlis[wnum[i] - 1],
"^18")["RAIN"].values.astype(np.float32)
hwah = sum_df.loc[repr(wnum[i]), 'HWAH']
ax.plot(np.arange(0, len(prep)),
prep,
color=cm.hsv(i / len(wnum)),
label="No" + repr(wnum[i]) + "(Yield=" + repr(hwah) + "kg/ha)")
plt.legend(fontsize=14, loc="best")
ax.set_ylabel('daily rainfall(mm)', fontsize=15)
plt.title("Daily rainfall of the scenario", fontsize=18)
plt.xlabel('Day of year', fontsize=15)
ax.set_xlim(180, 320)
plt.setp(ax.get_xticklabels(), fontsize=13, visible=True)
plt.setp(ax.get_yticklabels(), fontsize=13, visible=True)
plt.setp(ax2.get_yticklabels(), fontsize=13, visible=True)
#ax.set_ylim(0, 30)
#plt.savefig('190108_Reiton_RIX_png/190105_N_flow_in_1993_SSKT9605.png', bbox_inches='tight')
plt.show()
"""
190201 single year's plant N and precipitation.
plt.setp(ax.get_xticklabels(), fontsize=15, visible=True)
plt.setp(ax.get_yticklabels(), fontsize=15, visible=True)
#plt.savefig('181224_Dssat_png/190105_NH4_surface_in_DAS0_VS_yield_SSKT9601.png')
plt.show()
#plot rainfall and soil N content
fig = plt.figure(figsize=(8, 6))
ax = fig.add_subplot(1, 1, 1)
ax2 = ax.twinx()
for i in range(1, 5):
weath = searchElementforX(wlis[i - 1], '^18')
ax.plot(sn_df2.loc[:, 'DOY'][sn_df2.index == i].values.astype(np.int32),
sn_df2.loc[:, 'NI1D'][sn_df2.index == i].values.astype(np.float64),
color=cm.hsv((i - 1) / 4),
label="soil" + repr(i) + '(y=' + repr(hwah[i - 1]) + ')')
ax2.plot(np.arange(1,
len(weath.index) + 1, 1),
weath.loc[:, 'RAIN'].values.astype(np.float64),
color=cm.hsv((i - 1) / 4))
ax.legend(loc='best', ncol=2)
#plt.xticks(np.arange(len(wthdf.index)), np.inspace(1, 365, num=5))
plt.title("Total soil N content(TPDOY=213)", fontsize=18)
ax.set_xlabel('Day of year', fontsize=15)
ax.set_ylabel('Total soil N in profile(kg/ha)', fontsize=15)
ax2.set_ylabel('Daily rainfall(mm)', fontsize=15)
ax.set_ylim(0, 105)
ax.set_xlim(200, 250)
plt.setp(ax.get_xticklabels(), fontsize=13, visible=True)
lis5 = []
for i in range(1986, 2017):
df = searchElementforX(nitro_df5, repr(i))
nucm = sum(df["NUCM"])
lis5.append(nucm)
#visualize the result
fig = plt.figure(figsize = (12, 6))
ax = fig.add_subplot(1,1,1)
lis = [lis1, lis4, lis5]
for i in range(len(lis)):
ax.plot(np.arange(len(lis[i])), lis[i], color=cm.hsv(i/len(lis)),
label="pattern"+repr(i+1))
plt.legend(loc='best', fontsize=15)
plt.title("amount of N uptake in each simulation", fontsize=18)
plt.xticks(np.arange(0, len(lis[i]), 5), np.arange(1986, 1986+len(lis[i]), 5))
plt.xlabel('Year', fontsize=16)
plt.ylabel("N uptake (kg/ha)", fontsize=16)
plt.setp(ax.get_xticklabels(), fontsize=15, visible=True)
plt.setp(ax.get_yticklabels(), fontsize=15, visible=True)
plt.savefig('181224_Dssat_png/181227_Amount_of_N_uptake.png')
plt.show()
def genDaysInEachYear(year):
comparing calh vs real vs sim
visualize the 8 point of data by bar plot
"""
#HWAH
hwah2 = hwah.iloc[:, [0, 1, 3]]
fig = plt.figure(figsize=(10, 6))
ax = fig.add_subplot(1, 1, 1)
name = ["Koshihikari", "observed", "SD2"]
n = 0
for i in range(len(hwah2.columns)):
ax.bar(np.arange(len(hwah2.index)) + n,
hwah2.iloc[:, i],
color=cm.hsv(i / 3),
label=name[i],
width=0.8 / 3)
n = n + 0.8 / 3
plt.legend(loc="best", fontsize=14)
plt.xticks(np.arange(len(hwah2.index) + 0.55), [
"Kasai_q2_low", "Kasai_q2_high", "Kasai_d4_low", "Kasai_d4_high",
"Makabe_low", "Makabe_high", "Akita_low", "Akita_high"
],
rotation=30,
fontsize=14)
plt.title("Comparison of yield among observed, koshi_param and SD2_param",
fontsize=18)
plt.xlabel("field name", fontsize=15)
plt.ylabel("Yield (kg/ha)", fontsize=15)
'NI1D', 'NI2D', 'NI3D', 'NI4D', 'NI5D', 'NI6D', 'NI7D', 'NI8D',
'NH1D', 'NH2D', 'NH3D', 'NH4D', 'NH5D', 'NH6D', 'NH7D', 'NH8D',
]
"""
#compare the NO3 contents in soil surface layer
fig = plt.figure(figsize=(12, 6))
ax = fig.add_subplot(1, 1, 1)
lis = [nitro_df1, nitro_df2, nitro_df3, nitro_df4, nitro_df5]
for i in range(len(lis)):
ax.plot(np.arange(0, len(lis[i].index)),
lis[i]["NI1D"],
color=cm.hsv(i / len(lis)),
label="SSKT860" + repr(i + 1))
plt.legend(loc="best")
plt.xticks(np.linspace(0, len(lis[i].index), 31), np.arange(1986, 2017, 1))
plt.title("Soil surface NO3 content in each situation", fontsize=18)
plt.xlabel('Year', fontsize=15)
plt.ylabel('Soil surface NO3(ppm)', fontsize=15)
ax.set_xlim(0, 30)
#ax.set_ylim(0, 100)
#plt.savefig('181224_Dssat_png/181224_comparison_of_yield_5pattern.png', bbox_inches='tight')
plt.show()
def genDaysInEachYear(year):
if year % 4 == 0:
def plot_all(self, dat1, dat2, dat3, dat4):
font = {
'family': 'sans-serif',
'fontname': 'Arial',
'color': 'black',
'weight': 'normal',
'size': 10,
}
text_ycords = self.masky * .05
text_xcords = self.maskx * .04
fig = plt.figure(figsize=(16, 4))
fig.subplots_adjust(left=.05,
right=.95,
bottom=.05,
top=.9,
hspace=.1,
wspace=.05)
color_scale = np.arange(.1, 1, .9 / len(self.labels))
ax1 = fig.add_subplot(1, 4, 1, xticks=[], yticks=[])
ccc.plot_class_centroids(self, dat1, cm.gist_heat_r, ax1, hexbin=True)
ax1.set_title('all centroids', fontdict=font, fontsize=14)
ax1.text(text_xcords,
text_ycords * 19,
'threshold: ' + str(self.corr_thresh),
fontdict=font)
ax1.text(text_xcords,
text_ycords * 18,
'density: ' + str(self.eps),
fontdict=font)
ax1.text(text_xcords,
text_ycords * 17,
'min. size: ' + str(self.min_samples),
fontdict=font)
ax2 = fig.add_subplot(1, 4, 2, xticks=[], yticks=[])
for ii in range(
len(dat2)
): # dat2 is one longer tha dat1 because it contains unused centroids
dat2_vals = np.squeeze(np.asarray(dat2[ii].values()))
if ii == 0: # first entry is unused centroids
ccc.plot_class_centroids(self, dat2_vals, cm.gray_r(.3), ax2)
else:
ccc.plot_class_centroids(self, dat2_vals,
cm.hsv(color_scale[ii - 1]), ax2)
ax2.set_title('clustered centroids', fontdict=font, fontsize=14)
ax2.text(text_xcords,
text_ycords,
'threshold: ' + str(self.corr_thresh),
fontdict=font)
ax2.text(text_xcords,
text_ycords * 2,
'density: ' + str(self.eps),
fontdict=font)
ax2.text(text_xcords,
text_ycords * 3,
'min.size: ' + str(self.min_samples),
fontdict=font)
ax3 = fig.add_subplot(1, 4, 3, xticks=[], yticks=[])
ax3.imshow(dat3, interpolation='nearest')
ax3.set_title('projected correlation maps', fontdict=font, fontsize=14)
ax3.text(text_xcords,
text_ycords,
'threshold: ' + str(self.corr_thresh),
fontdict=font)
ax3.text(text_xcords,
text_ycords * 2,
'density: ' + str(self.eps),
fontdict=font)
ax3.text(text_xcords,
text_ycords * 3,
'min.size: ' + str(self.min_samples),
fontdict=font)
ax4 = fig.add_subplot(1, 4, 4, xticks=[], yticks=[])
ax4.imshow(dat4, interpolation='nearest')
ax4.set_title('seed-pixel masks', fontdict=font, fontsize=14)
ax4.text(text_xcords,
text_ycords,
'threshold: ' + str(self.corr_thresh),
fontdict=font)
ax4.text(text_xcords,
text_ycords * 2,
'density: ' + str(self.eps),
fontdict=font)
ax4.text(text_xcords,
text_ycords * 3,
'min.size: ' + str(self.min_samples),
fontdict=font)
#timestamp_str = time.strftime("%Y%m%d-%H%M")
if self.save_out:
savename = self.basepath + '_ccc.png'
fig.savefig(savename, dpi=300, format='png')
plt.setp(ax.get_yticklabels(), fontsize=15, visible=True)
#plt.savefig('181224_Dssat_png/190105_NH4_surface_in_DAS0_VS_yield_SSKT9601.png')
plt.show()
#plot rainfall and soil N content
fig = plt.figure(figsize = (8, 6))
ax = fig.add_subplot(1,1,1)
ax2 = ax.twinx()
for i in range(1, 5):
weath = searchElementforX(wlis[i-1], '^18')
ax.plot(sn_df.loc[:, 'DOY'][sn_df.index == i].values.astype(np.int32),
sn_df.loc[:, 'NI1D'][sn_df.index == i].values.astype(np.float64),
color=cm.hsv((i-1)/4), label="soil"+repr(i)+'(y='+repr(hwah[i-1])+')')
ax2.plot(np.arange(1, len(weath.index)+1, 1),
weath.loc[:, 'RAIN'].values.astype(np.float64),
color=cm.hsv((i-1)/4))
ax.legend(loc='best', ncol=2)
#plt.xticks(np.arange(len(wthdf.index)), np.inspace(1, 365, num=5))
plt.title("Total soil N content(TPDOY=213)", fontsize=18)
ax.set_xlabel('Day of year', fontsize=15)
ax.set_ylabel('Total soil N in profile(kg/ha)', fontsize=15)
ax2.set_ylabel('Daily rainfall(mm)', fontsize=15)
ax.set_ylim(5, 105)
ax.set_xlim(200, 250)
plt.setp(ax.get_xticklabels(), fontsize=13, visible=True)
plt.setp(ax.get_yticklabels(), fontsize=13, visible=True)
plt.setp(ax2.get_yticklabels(), fontsize=13, visible=True)
y_value = pd.concat(
[sum1["HWAH"], sum2["HWAH"], sum3["HWAH"], sum4["HWAH"], sum5["HWAH"]],
axis=1).values.astype(np.int32)
y_df = pd.DataFrame(y_value,
index=sum1.index,
columns=["sum1", "sum2", "sum3", "sum4", "sum5"])
fig = plt.figure(figsize=(12, 6))
ax = fig.add_subplot(1, 1, 1)
for i in range(len(y_df.columns)):
ax.plot(np.arange(1,
len(y_df.index) + 1),
y_df.iloc[:, i],
color=cm.hsv(i / len(y_df.columns)))
sum_1, = plt.plot((1, 1), (1, 1),
color=cm.hsv(0 / len(y_df.columns)),
linewidth=10)
sum_2, = plt.plot((1, 1), (1, 1),
color=cm.hsv(1 / len(y_df.columns)),
linewidth=10)
sum_3, = plt.plot((1, 1), (1, 1),
color=cm.hsv(2 / len(y_df.columns)),
linewidth=10)
sum_4, = plt.plot((1, 1), (1, 1),
color=cm.hsv(3 / len(y_df.columns)),
linewidth=10)
sum_5, = plt.plot((1, 1), (1, 1),
color=cm.hsv(4 / len(y_df.columns)),
