def H_A_30_test():
dirs = "../result_h/A/A_30_test/"
main_v.mkdir(dirs)
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(i+1, M))
_, _, _, data = main_v.main_data(
start, start,
span=30,
get_columns=["hermert"],
region=None,
accumulate=False
)
save_name = dirs + str(start)[:6] + ".png"
visualize.plot_map_once(
data["A"],
data_type="type_non_wind",
save_name=save_name,
show=False,
vmax=None,
vmin=None,
cmap=plt.cm.jet
)
print("\n")
def H_A_and_theta_by_year():
A_by_year_list = ["01", "02", "03", "04", "05", "06", "07", "08", "09", "10", "11", "12"]
dirs_A = "../result_h/A/A_by_year/"
main_v.mkdir(dirs_A)
dirs_theta = "../result_h/theta/theta_by_year/"
main_v.mkdir(dirs_theta)
for item in A_by_year_list:
fname = "../data/csv_Helmert_by_year/Helmert_by_year_" + item + ".csv"
df_coeffs = pd.read_csv(fname, sep=',', dtype='float32')
save_name_A = dirs_A + "Hermert_" + item + ".png"
visualize.plot_map_once(
df_coeffs["A"],
data_type="type_non_wind",
save_name=save_name_A,
show=False,
vmax=0.025,
vmin=0,
cmap=plt.cm.jet
)
save_name_theta = dirs_theta + "Hermert_" + item + ".png"
visualize.plot_map_once(
df_coeffs["theta"],
data_type="type_non_wind",
save_name=save_name_theta,
show=False,
vmax=180,
vmin=-180,
cmap=cm_angle_2
)
print("\n")
def test_scatter():
mkdir("../result_h/test/test_scatter/")
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(i+1, M))
data_ex_dir = "../data/csv_Helmert_ex/Helmert_ex_" + str(start)[:-2] + ".csv"
data = pd.read_csv(data_ex_dir)
#data_01 = []
try:
#data_01 = data[((data.area_label==16)) & (data.coastal_region_1==False) & (data.ic0_30<99)].dropna()
#data_01 = data[((data.area_label==8)|(data.area_label==10)) & (data.coastal_region_1==False) & (data.ic0_30<99)].dropna()
#data_01 = data[((data.area_label==1)|(data.area_label==5)|(data.area_label==7)|(data.area_label==8)) & (data.coastal_region_1==False) & (data.ic0_30<99)].dropna()
#data_01 = data.query('area_label in [1,4,5,7,10,11]')[(data.coastal_region_1==False) & (data.ic0_30<99)].dropna()
data_01 = data[(data.area_label.isin([1,4,5,7,10,11])) & (data.coastal_region_1==False) & (data.ic0_30<99)].dropna()
except:
data_01 = pd.DataFrame([])
#print(data_01.head())
if len(data_01) >= 4:
sns.jointplot(x="ic0_30", y="A", data=data_01, kind="reg")
save_name = "../result_h/test/test_scatter/" + "A_ic0_" + str(start)[:-2] + ".png"
plt.savefig(save_name, dpi=600)
plt.close()
"""
visualize.visual_non_line(
data,
mode=["scatter", ["sit_30", "A"]],
save_name=save_dir + str(start)[:6] + ".png",
show=False
)
"""
print("\n")
def plot_hist():
mkdir("../result_h/search_anomaly/hist/")
start_list.pop()
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(
i + 1, M - 1))
data_ex_dir = "../data/csv_Helmert_ex/Helmert_ex_" + str(
start)[:-2] + ".csv"
data = pd.read_csv(data_ex_dir)
x = data.R2[((data.area_label == 16))
& ((data.A < 0.003) | (data.ic0_30 < 97))].dropna().values
y = data.R2[((data.area_label == 16)) & (
(data.A >= 0.003) & (data.ic0_30 >= 97))].dropna().values
sns.distplot(x, color="skyblue")
sns.distplot(y, color="red")
"""
fig, ax = plt.subplots()
for a in [x, y]:
sns.distplot(a, ax=ax, kde=True)
ax.set_xlim([0, 0.03])
"""
save_name = "../result_h/search_anomaly/hist/" + "A_ic0_" + str(
start)[:-2] + ".png"
plt.savefig(save_name, dpi=600)
plt.close()
print("\n")
def H_vec_mean_ocean_currents():
dirs = "../result_h/mean_vector/mean_ocean_currents/"
main_v.mkdir(dirs)
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(i+1, M))
_, _, _, data = main_v.main_data(
start, start,
span=30,
get_columns=["hermert"],
region=None,
accumulate=False
)
save_name = dirs + "mean_ocean_currents_" + str(start)[:6] + ".png"
visualize.plot_map_once(
data.loc[:, ["ocean_u", "ocean_v"]],
data_type="type_wind",
save_name=save_name,
show=False,
vmax=None,
vmin=None,
cmap=None
)
print("\n")
def test_scatter():
mkdir("../result_h/test/test_scatter/")
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(
i + 1, M))
data_ex_dir = "../data/csv_Helmert_ex/Helmert_ex_" + str(
start)[:-2] + ".csv"
data = pd.read_csv(data_ex_dir)
#data01 = data.loc[data.coastal_region_1==0, :].dropna()
data_01 = data[((data.area_label == 4) | (data.area_label == 5))
& (data.coastal_region_1 == False) &
(data.ic0_30 < 99)].dropna()
sns.jointplot(x="ic0_30", y="A", data=data_01, kind="reg")
save_name = "../result_h/test/test_scatter/" + "A_ic0_" + str(
start)[:-2] + ".png"
plt.savefig(save_name, dpi=600)
plt.close()
"""
visualize.visual_non_line(
data,
mode=["scatter", ["sit_30", "A"]],
save_name=save_dir + str(start)[:6] + ".png",
show=False
)
"""
print("\n")
def H_A_30_with_coef():
dirs = "../result_h/A/A_30_with_coef/"
main_v.mkdir(dirs)
df_latlon = pd.read_csv("../data/latlon_ex.csv")
lon = df_latlon.Lon
lat = df_latlon.Lat
lon = np.array(lon)
lat = np.array(lat)
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(i+1, M))
_, _, _, data = main_v.main_data(
start, start,
span=30,
get_columns=["hermert"],
region=None,
accumulate=False
)
idx_low_coef = data[data.R2<threshold_R2].index
data_low_coef = np.array([np.nan]*(145**2))
data_low_coef[idx_low_coef] = 1
data_low_coef1 = np.reshape(data_low_coef, (145,145), order="F")
data_low_coef1 = np.ma.masked_invalid(data_low_coef1)
data.A.loc[data.R2<threshold_R2] = np.nan
data = np.array(data.A)
data1 = np.reshape(data, (145,145), order="F")
data1 = np.ma.masked_invalid(data1)
save_name = dirs + str(start)[:6] + ".png"
m = Basemap(lon_0=180, boundinglat=50, resolution='i', projection='npstere')
fig = plt.figure(figsize=(6.5, 6.5))
x, y = m(lon, lat)
x1 = np.reshape(x, (145,145), order="F")
y1 = np.reshape(y, (145,145), order="F")
dx1 = (x1[1,0]-x1[0,0])/2
dy1 = (y1[0,1]-y1[0,0])/2
x2 = np.linspace(x1[0,0], x1[144,0], 145)
y2 = np.linspace(y1[0,0], y1[0,144], 145)
xx, yy = np.meshgrid(x2, y2)
xx, yy = xx.T, yy.T
xx = np.hstack([xx, xx[:,0].reshape(145,1)])
xx_ex = np.vstack([xx, (xx[144,:] + (xx[1,0]-xx[0,0]))])
yy = np.vstack([yy, yy[0,:]])
yy_ex = np.hstack([(yy[:,0].reshape(146,1) + (yy[0,0]-yy[0,1])), yy])
m.drawcoastlines(color = '0.15')
# m.plot(xx[144,0], yy[144,0], "bo")
m.pcolormesh(xx_ex-dx1, yy_ex+dy1, data1, cmap=plt.cm.jet, vmax=0.025, vmin=0)
m.colorbar(location='bottom')
cm_brown = visualize.generate_cmap(["burlywood", "burlywood"])
m.pcolormesh(xx_ex-dx1, yy_ex+dy1, data_low_coef1, cmap=cm_brown)
#plt.show()
plt.savefig(save_name, dpi=900)
plt.close()
print("\n")
def heatmap():
dirs = "../result_h/heatmap/"
mkdir(dirs)
start_list.pop()
for i, start in enumerate(start_list):
print("******************* {}/{} *******************".format(
i + 1, M - 1))
#csv_Helmert_90の当該月のcsvを読み込む
#hermert_file_name_90 = "../data/csv_Helmert_both_90/Helmert_both_90_" + str(start)[:6] + ".csv"
#helmert_90_data = pd.read_csv(hermert_file_name_90)
#csv_Helmert_30の当該月のcsvを読み込む
hermert_file_name_30 = "../data/csv_Helmert_both_30/Helmert_both_30_" + str(
start)[:6] + ".csv"
helmert_30_data = pd.read_csv(hermert_file_name_30)
#木村のcsvの読み込み(90)
#coeff_file_name_90 = "../data/csv_A_90/ssc_amsr_ads" + str(start)[2:6] + "_90_fin.csv"
#coeff_90_data = calc_data.get_1month_coeff_data(coeff_file_name_90)
#木村のcsvの読み込み(30)
#coeff_file_name_30 = "../data/csv_A_30/ssc_amsr_ads" + str(start)[2:6] + "_30_fin.csv"
#coeff_30_data = calc_data.get_1month_coeff_data(coeff_file_name_30)
helmert_30_data["cross_ic0_sit"] = helmert_30_data[
"ic0_30"] * helmert_30_data["sit_30"] / 100
helmert_30_data["ocean_speed"] = np.sqrt(
helmert_30_data["ocean_u"]**2 + helmert_30_data["ocean_v"]**2)
cr = [
'A', 'theta', 'R2', 'epsilon2', 'ic0_30', 'sit_30',
'cross_ic0_sit', 'ocean_speed'
]
cr_matrix = helmert_30_data.loc[helmert_30_data.area_label == 7,
cr].corr()
mask = np.zeros_like(cr_matrix, dtype=np.bool)
mask[np.triu_indices_from(mask)] = True
try:
sns.set(font_scale=0.7)
sns.heatmap(cr_matrix,
mask=mask,
cbar=True,
annot=True,
square=True,
fmt='.2f',
annot_kws={'size': 10},
yticklabels=cr,
xticklabels=cr,
cmap="bwr",
vmin=-1,
vmax=1)
plt.savefig(dirs + "area_7_" + str(start)[:6] + ".png", dpi=450)
plt.close()
except:
continue
def test_ocean_mean_iw_plot():
dirs = "../result_h/test/test_ocean/"
mkdir(dirs)
start_list_plus_1month = start_list + [20170901]
for i, start in enumerate(start_list):
print("******************* {}/{} *******************".format(
i + 1, M))
month_end = start_list_plus_1month[i + 1]
month_end = date(month_end // 10000, (month_end % 10000) // 100,
(month_end % 10000) % 100) - timedelta(days=1)
end = start + month_end.day - 1
#csv_Helmert_90の当該月のcsvを読み込む
hermert_file_name_90 = "../data/csv_Helmert_both_90/Helmert_both_90_" + str(
start)[:6] + ".csv"
helmert_90_data = pd.read_csv(hermert_file_name_90).dropna()
#csv_Helmert_30の当該月のcsvを読み込む
hermert_file_name_30 = "../data/csv_Helmert_30/Helmert_30_" + str(
start)[:6] + ".csv"
helmert_30_data = pd.read_csv(hermert_file_name_30).dropna()
#木村のcsvの読み込み(90)
coeff_file_name_90 = "../data/csv_A_90/ssc_amsr_ads" + str(
start)[2:6] + "_90_fin.csv"
coeff_90_data = calc_data.get_1month_coeff_data(
coeff_file_name_90).dropna()
#木村のcsvの読み込み(30)
coeff_file_name_30 = "../data/csv_A_30/ssc_amsr_ads" + str(
start)[2:6] + "_30_fin.csv"
coeff_30_data = calc_data.get_1month_coeff_data(
coeff_file_name_30).dropna()
ocean_90_h_u = helmert_90_data["ocean_u"].values
ocean_90_h_v = helmert_90_data["ocean_v"].values
ocean_90_h_speed = np.sqrt(ocean_90_h_u**2 + ocean_90_h_v**2)
ocean_30_h_u = helmert_30_data["ocean_u"].values
ocean_30_h_v = helmert_30_data["ocean_v"].values
ocean_30_h_speed = np.sqrt(ocean_30_h_u**2 + ocean_30_h_v**2)
ocean_90_c_u = coeff_90_data["mean_ocean_u"].values
ocean_90_c_v = coeff_90_data["mean_ocean_v"].values
ocean_90_c_speed = np.sqrt(ocean_90_c_u**2 + ocean_90_c_v**2)
ocean_30_c_u = coeff_30_data["mean_ocean_u"].values
ocean_30_c_v = coeff_30_data["mean_ocean_v"].values
ocean_30_c_speed = np.sqrt(ocean_30_c_u**2 + ocean_30_c_v**2)
fig, axes = plt.subplots(1, 2)
axes[0].hist(ocean_90_c_speed / ocean_90_h_speed)
axes[1].hist(ocean_30_c_speed / ocean_30_h_speed)
plt.savefig(dirs + "hist_ocean_" + str(start)[:6] + ".png", dpi=450)
plt.close()
def H_angle_30_and_wind():
dirs = "../result_h/angle/angle_30_and_wind/"
main_v.mkdir(dirs)
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(i+1, M))
month_end = start_list_plus_1month[i+1]
month_end = date(month_end//10000, (month_end%10000)//100, (month_end%10000)%100) - timedelta(days=1)
end = start + month_end.day - 1
#angleデータの取得
_, _, _, data = main_v.main_data(
start, start,
span=30,
get_columns=["hermert"],
region=None,
accumulate=False
)
data.theta[data.R2<threshold_R2] = np.nan
#地衡風平均の出力
_, _, _, data_w = main_v.main_data(
start, end,
span=30,
get_columns=["w"],
region=None,
accumulate=True
)
data_array = np.array(data_w)
data_ave = np.nanmean(data_array, axis=0)
data_ave = pd.DataFrame(data_ave)
data_ave.columns = ["w_speed", "w_u", "w_v"]
save_name = dirs + "angle_30_and_wind_" + str(start)[:6] + ".png"
visualize.plot_map_multi(
data_ave.loc[:, ["w_u", "w_v"]],
data["theta"],
data_type="type_non_wind",
save_name=save_name,
show=False,
vmax=180,
vmin=-180,
cmap=cm_angle_2
)
print("\n")
def get_helmert_both_30_modified():
dirs = "../data/csv_Helmert_both_30/"
mkdir(dirs)
for i, start in enumerate(start_list):
print("******************* {}/{} *******************".format(i+1, M))
#csv_Helmert_30の当該月のcsvを読み込む
hermert_file_name_30 = "../data/csv_Helmert_30/Helmert_30_" + str(start)[:6] + ".csv"
helmert_30_data = pd.read_csv(hermert_file_name_30)
data_ex_dir = "../data/csv_Helmert_ex/Helmert_ex_" + str(start)[:-2] + ".csv"
data_ex = pd.read_csv(data_ex_dir)
data = pd.concat([helmert_30_data, data_ex.loc[:, ["coastal_region_1", "coastal_region_2", "area_label", "ic0_30", "ic0_30_median", "sit_30", "sit_30_median"]]], axis=1)
save_name = dirs + "Helmert_both_30_" + str(start)[:6] + ".csv"
print(save_name)
data.to_csv(save_name, index=False)
def H_angle_30():
dirs_original = "../result_h/angle/angle_30/"
main_v.mkdir(dirs_original)
dirs_h_coef = "../result_h/angle/angle_30_high_coef/"
main_v.mkdir(dirs_h_coef)
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(i+1, M))
_, _, _, data = main_v.main_data(
start, start,
span=30,
get_columns=["hermert"],
region=None,
accumulate=False
)
#angle_30
save_name_original = dirs_original + "angle_30_" + str(start)[:6] + ".png"
visualize.plot_map_once(
data["theta"],
data_type="type_non_wind",
save_name=save_name_original,
show=False,
vmax=180,
vmin=-180,
cmap=cm_angle_2
)
#angle_30_high_coef
data.theta[data.R2<threshold_R2] = np.nan
save_name = dirs_h_coef + "angle_30_high_coef_" + str(start)[:6] + ".png"
visualize.plot_map_once(
data["theta"],
data_type="type_non_wind",
save_name=save_name,
show=False,
vmax=180,
vmin=-180,
cmap=cm_angle_2
)
print("\n")
def H_A_by_day_30():
"""
基本的に古い方のA_by_day_30と変わらないが，こっちの方がちょっとだけ正確かも
"""
dirs = "../result_h/A/A_by_day_30/"
main_v.mkdir(dirs)
for i, start in enumerate(start_list):
print("******************* {}/{} *******************".format(i+1, M))
month_end = start_list_plus_1month[i+1]
month_end = date(month_end//10000, (month_end%10000)//100, (month_end%10000)%100) - timedelta(days=1)
end = start + month_end.day - 1
date_ax, date_ax_str, skipping_date_str, data = main_v.main_data(
start, end,
span=30,
get_columns=["ex_2"],
region=None,
accumulate=True
)
data_array = np.array(data)
data_ave = np.nanmean(data_array, axis=0)
#A_by_dayなので0列目
data_ave = pd.DataFrame(data_ave[:, 0])
data_ave.columns = ["A_by_day"]
save_name = dirs + "A_by_day_30_" + str(start)[:6] + ".png"
visualize.plot_map_once(
data_ave["A_by_day"],
data_type="type_non_wind",
save_name=save_name,
show=False,
vmax=0.025,
vmin=None,
cmap=plt.cm.jet
)
print("\n")
def compare_3csv():
dirs = "../result_h/compare_3csv/"
dirs_A = "../result_h/compare_3csv/A/"
dirs_theta = "../result_h/compare_3csv/theta/"
dirs_R2 = "../result_h/compare_3csv/R2/"
dirs_e2 = "../result_h/compare_3csv/e2/"
mkdir(dirs_A)
mkdir(dirs_theta)
mkdir(dirs_R2)
mkdir(dirs_e2)
kw_list = [["A", "A_30_90", "A_30_w_iw"],
["theta", "theta_30_90", "theta_w_iw"],
["R2", "R2_30_90", "R2_w_iw"],
["epsilon2", "epsilon2_30_90", "epsilon2_w_iw"]]
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(
i + 1, M))
helmert_30_30_fname = "../data/csv_Helmert_30/Helmert_30_" + str(
start)[:6] + ".csv"
data_30 = pd.read_csv(helmert_30_30_fname)
helmert_30_90_fname = "../data/csv_Helmert_ocean_90/Helmert_30_90_" + str(
start)[:6] + ".csv"
data_90 = pd.read_csv(helmert_30_90_fname)
helmert_30_30_w_iw_fname = "../data/csv_Helmert_both_30_w_iw/Helmert_both_30_w_iw_" + str(
start)[:6] + ".csv"
data_30_w_iw = pd.read_csv(helmert_30_30_w_iw_fname)
save_name_list = ["A", "theta", "R2", "e2"]
for j in range(4):
data_30_plot = np.array(data_30[kw_list[j][0]])
data_90_plot = np.array(data_90[kw_list[j][1]])
data_30_w_iw_plot = np.array(data_30_w_iw[kw_list[j][2]])
data_list = [data_30_plot, data_90_plot, data_30_w_iw_plot]
v_lim_list = [[0.025, 0], [180, -180], [1, 0], [None, 0]]
fig, axes = plt.subplots(1, 3)
for k in range(3):
ax = axes[k]
ax.set_title(kw_list[j][k])
m = Basemap(lon_0=180,
boundinglat=50,
resolution='l',
projection='npstere')
m.ax = ax
m.drawcoastlines(color='0.15')
m.fillcontinents(color='#555555')
lon = np.array(latlon_ex.Lon)
lat = np.array(latlon_ex.Lat)
x, y = m(lon, lat)
x1 = np.reshape(x, (145, 145), order='F')
y1 = np.reshape(y, (145, 145), order='F')
dx1 = (x1[1, 0] - x1[0, 0]) / 2
dy1 = (y1[0, 1] - y1[0, 0]) / 2
x2 = np.linspace(x1[0, 0], x1[144, 0], 145)
y2 = np.linspace(y1[0, 0], y1[0, 144], 145)
xx, yy = np.meshgrid(x2, y2)
xx, yy = xx.T, yy.T
xx = np.hstack([xx, xx[:, 0].reshape(145, 1)])
xx_ex = np.vstack([xx, (xx[144, :] + (xx[1, 0] - xx[0, 0]))])
yy = np.vstack([yy, yy[0, :]])
yy_ex = np.hstack([
(yy[:, 0].reshape(146, 1) + (yy[0, 0] - yy[0, 1])), yy
])
data = np.ma.masked_invalid(data_list[k])
data1 = np.reshape(data, (145, 145), order='F')
im = m.pcolormesh(xx_ex - dx1,
yy_ex + dy1,
data1,
cmap=plt.cm.jet,
vmax=v_lim_list[j][0],
vmin=v_lim_list[j][1])
fig.colorbar(im, ax=ax)
#m.colorbar(location='bottom')
save_name = dirs + save_name_list[j] + "/" + str(
start)[:6] + ".png"
print(save_name)
plt.savefig(save_name, dpi=750)
plt.close()
def scatter_ic0_by_csv():
csv_kind_list = ["csv_30_30", "csv_30_90", "csv_30_30_w_iw"]
plot_y_list = ["A", "theta"]
#plot_y_list = ["e2"]
area_list = ["area_" + str(i) for i in range(17)]
for csv_item in csv_kind_list:
for plot_item in plot_y_list:
for area_item in area_list:
dirs = "../result_h/scatter/sit_by_csv/" + csv_item + "/" + plot_item + "/" + area_item + "/"
mkdir(dirs)
kw_y_list = [["A", "A_30_90", "A_30_w_iw"],
["theta", "theta_30_90", "theta_w_iw"]]
#kw_y_list = [["epsilon2", "epsilon2_30_90", "epsilon2_30_w_iw"]]
y_lim_list = [(0, 0.025), (-90, 90)]
sns.set_style("darkgrid")
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(
i + 1, M))
helmert_30_30_fname = "../data/csv_Helmert_both_30/Helmert_both_30_" + str(
start)[:6] + ".csv"
data_30 = pd.read_csv(helmert_30_30_fname)
helmert_30_90_fname = "../data/csv_Helmert_ocean_90/Helmert_30_90_" + str(
start)[:6] + ".csv"
data_90 = pd.read_csv(helmert_30_90_fname)
helmert_30_30_w_iw_fname = "../data/csv_Helmert_both_30_w_iw/Helmert_both_30_w_iw_" + str(
start)[:6] + ".csv"
data_30_w_iw = pd.read_csv(helmert_30_30_w_iw_fname)
for j, kw in enumerate(kw_y_list):
for k, area_item in enumerate(area_list):
try:
save_name_30_30 = "../result_h/scatter/sit_by_csv/" + csv_kind_list[
0] + "/" + plot_y_list[
j] + "/" + area_item + "/" + "lmplot_coastal_2_" + str(
start)[:6] + ".png"
lm = sns.lmplot(x="sit_30",
y=kw[0],
col="coastal_region_2",
data=data_30[data_30["area_label"] == k],
size=3,
fit_reg=True)
lm.set(ylim=y_lim_list[j])
plt.savefig(save_name_30_30, dpi=500)
plt.close()
except:
print("Skipped plot: {}".format(csv_kind_list[0] + "/" +
plot_y_list[j] + "/" +
area_item))
plt.close()
try:
save_name_30_90 = "../result_h/scatter/sit_by_csv/" + csv_kind_list[
1] + "/" + plot_y_list[
j] + "/" + area_item + "/" + "lmplot_coastal_2_" + str(
start)[:6] + ".png"
lm = sns.lmplot(x="sit_30",
y=kw[1],
col="coastal_region_2",
data=data_90[data_90.area_label == k],
size=3,
fit_reg=True)
lm.set(ylim=y_lim_list[j])
plt.savefig(save_name_30_90, dpi=500)
plt.close()
except:
print("Skipped plot: {}".format(csv_kind_list[1] + "/" +
plot_y_list[j] + "/" +
area_item))
plt.close()
try:
save_name_30_30_w_iw = "../result_h/scatter/sit_by_csv/" + csv_kind_list[
2] + "/" + plot_y_list[
j] + "/" + area_item + "/" + "lmplot_coastal_2_" + str(
start)[:6] + ".png"
lm = sns.lmplot(
x="sit_30",
y=kw[2],
col="coastal_region_2",
data=data_30_w_iw[data_30_w_iw.area_label == k],
size=3,
fit_reg=True)
lm.set(ylim=y_lim_list[j])
plt.savefig(save_name_30_30_w_iw, dpi=500)
plt.close()
except:
print("Skipped plot: {}".format(csv_kind_list[2] + "/" +
plot_y_list[j] + "/" +
area_item))
plt.close()
def get_helmert_ocean_90():
dirs = "../data/csv_Helmert_ocean_90/"
mkdir(dirs)
start_list_plus_1month = start_list + [20170901]
for i, start in enumerate(start_list):
print("******************* {}/{} *******************".format(i+1, M))
month_end = start_list_plus_1month[i+1]
month_end = date(month_end//10000, (month_end%10000)//100, (month_end%10000)%100) - timedelta(days=1)
end = start + month_end.day - 1
#csv_Helmert_90の当該月のcsvを読み込む
hermert_file_name = "../data/csv_Helmert_both_90/Helmert_both_90_" + str(start)[:6] + ".csv"
helmert_90_data = pd.read_csv(hermert_file_name)
#wデータの取得・整形
_, _, _, data_w = main_data(
start, end,
span=30,
get_columns=["w"],
region=None,
accumulate=True
)
data_array_w = np.array(data_w)
data_ave_w = np.nanmean(data_array_w, axis=0)
#iwデータの取得・整形
_, _, _, data_iw = main_data(
start, end,
span=30,
get_columns=["iw"],
region=None,
accumulate=True
)
data_array_iw = np.array(data_iw)
print("\n")
w_array = np.vstack((data_array_iw[:,:,1], data_array_iw[:,:,2]))
ocean_array_u = np.array(helmert_90_data["ocean_u_90"])
ocean_array_v = np.array(helmert_90_data["ocean_v_90"])
ocean_array = np.hstack((ocean_array_u.reshape((-1,1)), ocean_array_v.reshape((-1,1))))
ocean_array = np.tile(ocean_array, (data_array_iw.shape[0],1,1))
ocean_array = np.vstack((ocean_array[:,:,0], ocean_array[:,:,1]))
#print(ocean_array.shape)
#print(w_array[:, 0].shape)
Helmert = []
#for j in range(100):
for j in range(145**2):
print("j: {}".format(j))
#欠損データの処理
not_nan_idx = np.sum(np.isnan(data_array_iw[:, j, :]), axis=1)==False
#print("\tnot_nan_idx: {}".format(not_nan_idx))
x = data_array_w[:, j, 1][not_nan_idx].reshape((-1,1))
y = data_array_w[:, j, 2][not_nan_idx].reshape((-1,1))
w = (w_array[:, j]-ocean_array[:, j])[np.tile(not_nan_idx, 2)].reshape((-1,1))
iw_u_ave = np.nanmean(data_array_iw[:, j, 1])
iw_v_ave = np.nanmean(data_array_iw[:, j, 2])
iw_u_ave_ocean = np.nanmean(data_array_iw[:, j, 1] - ocean_array_u[j])
iw_v_ave_ocean = np.nanmean(data_array_iw[:, j, 2] - ocean_array_v[j])
N_c = np.sum(not_nan_idx==True)
if N_c <= 1:
print("\tskipping for loop...")
Helmert.append([np.nan, np.nan, np.nan, np.nan, N_c, np.nan, np.nan])
continue
one_N = np.ones(N_c).reshape((-1,1))
zero_N = np.zeros(N_c).reshape((-1,1))
#D_1 = np.hstack((one_N, zero_N, x, -y))
#D_2 = np.hstack((zero_N, one_N, y, x))
D_1 = np.hstack((x, -y))
D_2 = np.hstack((y, x))
Dt = np.vstack((D_1, D_2))
#print(N_c, Dt, np.dot(Dt.T, Dt))
D_inv = np.linalg.inv(np.dot(Dt.T, Dt))
gamma = np.dot(D_inv, np.dot(Dt.T, w))
print("\t{}".format(gamma.shape))
A = np.sqrt(gamma[0]**2 + gamma[1]**2)
theta = np.arctan2(gamma[1], gamma[0]) * 180/np.pi
R_denominator = np.dot(w.T, w) - N_c*(iw_u_ave_ocean**2 + iw_v_ave_ocean**2)
epsilon = w - np.dot(Dt, gamma)
R_numerator = np.dot(epsilon.T, epsilon)
R2 = 1 - (R_numerator/R_denominator)[0,0]
#print("\t{}".format([A[0], theta[0], gamma[0,0], gamma[1,0], R2, N_c]))
if N_c <= 20:
Helmert.append([np.nan, np.nan, np.nan, np.nan, N_c, iw_u_ave_ocean, iw_v_ave_ocean])
else:
Helmert.append([A, theta, R2, R_numerator, N_c, iw_u_ave_ocean, iw_v_ave_ocean])
result = np.hstack((Helmert, data_ave_w[:,[1,2]]))
#print(result.shape)
data = pd.DataFrame(result)
data.columns = ["A_30_90", "theta_30_90", "R2_30_90", "epsilon2_30_90", "N_c_30_90", "mean_iw_u_with_ocean_30_90", "mean_iw_v_with_ocean_30_90", "mean_w_u_30", "mean_w_v_30"]
data_ex_dir = "../data/csv_Helmert_ex/Helmert_ex_" + str(start)[:-2] + ".csv"
data_ = pd.read_csv(data_ex_dir).loc[:, ["coastal_region_1", "coastal_region_2", "area_label", "ic0_30", "ic0_30_median", "sit_30", "sit_30_median"]]
data = pd.concat([latlon_ex, data_, data], axis=1)
#print(data.head(3))
save_name = dirs + "Helmert_30_90_" + str(start)[:6] + ".csv"
print(save_name)
data.to_csv(save_name, index=False)
def get_helmert_both_90(data_ex_basic):
dirs = "../data/csv_Helmert_both_90/"
mkdir(dirs)
#start_list = [20030101]
start_list_plus_3month = start_list + [20170901, 20171001, 20171101]
for k, start_1 in enumerate(start_list):
if start_1 == 20170801:
print("Breaking the loop...")
break
print("******************* {}/{} *******************".format(k+1, M))
if k == 0:
month_end_1 = start_list_plus_3month[k+1]
month_end_1 = date(month_end_1//10000, (month_end_1%10000)//100, (month_end_1%10000)%100) - timedelta(days=1)
end_1 = start_1 + month_end_1.day - 1
start_2 = month_end_1 + timedelta(days=1)
month_end_2 = start_list_plus_3month[k+2]
month_end_2 = date(month_end_2//10000, (month_end_2%10000)//100, (month_end_2%10000)%100) - timedelta(days=1)
end_2 = start_2 + timedelta(days=month_end_2.day-1)
start_2 = int(start_2.strftime('%Y/%m/%d').replace("/", ""))
end_2 = int(end_2.strftime('%Y/%m/%d').replace("/", ""))
start_list_3month = [start_1, start_2, 0]
end_list_3month = [end_1, end_2, 0]
else:
kk = k - 1
start_1_1 = start_list[kk]
month_end_1 = start_list_plus_3month[kk+1]
month_end_1 = date(month_end_1//10000, (month_end_1%10000)//100, (month_end_1%10000)%100) - timedelta(days=1)
end_1 = start_1_1 + month_end_1.day - 1
start_2 = month_end_1 + timedelta(days=1)
month_end_2 = start_list_plus_3month[kk+2]
month_end_2 = date(month_end_2//10000, (month_end_2%10000)//100, (month_end_2%10000)%100) - timedelta(days=1)
end_2 = start_2 + timedelta(days=month_end_2.day-1)
start_3 = month_end_2 + timedelta(days=1)
month_end_3 = start_list_plus_3month[kk+3]
month_end_3 = date(month_end_3//10000, (month_end_3%10000)//100, (month_end_3%10000)%100) - timedelta(days=1)
end_3 = start_3 + timedelta(days=month_end_3.day-1)
start_2 = int(start_2.strftime('%Y/%m/%d').replace("/", ""))
start_3 = int(start_3.strftime('%Y/%m/%d').replace("/", ""))
end_2 = int(end_2.strftime('%Y/%m/%d').replace("/", ""))
end_3 = int(end_3.strftime('%Y/%m/%d').replace("/", ""))
start_list_3month = [start_1_1, start_2, start_3]
end_list_3month = [end_1, end_2, end_3]
data_w_90 = np.zeros((1, 145**2, 3))
data_iw_90 = np.zeros((1, 145**2, 3))
for i in range(3):
start = start_list_3month[i]
end = end_list_3month[i]
if start == 0 and end == 0:
continue
#wデータの取得・整形
_, _, _, data_w = main_data(
start, end,
span=30,
get_columns=["w"],
region=None,
accumulate=True
)
data_array_w = np.array(data_w)
#data_ave_w = np.nanmean(data_array_w, axis=0)
data_w_90 = np.concatenate([data_w_90, data_array_w], axis=0)
#iwデータの取得・整形
_, _, _, data_iw = main_data(
start, end,
span=30,
get_columns=["iw"],
region=None,
accumulate=True
)
data_array_iw = np.array(data_iw)
data_iw_90 = np.concatenate([data_iw_90, data_array_iw], axis=0)
print("\n")
data_w_90 = data_w_90[1:, :, :]
data_iw_90 = data_iw_90[1:, :, :]
data_ave_w = np.nanmean(data_w_90, axis=0)
data_ave_iw = np.nanmean(data_iw_90, axis=0)
w_array = np.vstack((data_iw_90[:,:,1], data_iw_90[:,:,2]))
Helmert = []
for j in range(145**2):
print("j: {}".format(j))
#欠損データの処理
not_nan_idx = np.sum(np.isnan(data_iw_90[:, j, :]), axis=1)==False
#print("\tnot_nan_idx: {}".format(not_nan_idx))
x = data_w_90[:, j, 1][not_nan_idx].reshape((-1,1))
y = data_w_90[:, j, 2][not_nan_idx].reshape((-1,1))
w = w_array[:, j][np.tile(not_nan_idx, 2)].reshape((-1,1))
iw_u_ave = np.nanmean(data_iw_90[:, j, 1])
iw_v_ave = np.nanmean(data_iw_90[:, j, 2])
N_c = np.sum(not_nan_idx==True)
if N_c <= 1:
print("\tskipping for loop...")
Helmert.append([np.nan, np.nan, np.nan, np.nan, np.nan, np.nan, N_c, np.nan, np.nan])
continue
one_N = np.ones(N_c).reshape((-1,1))
zero_N = np.zeros(N_c).reshape((-1,1))
D_1 = np.hstack((one_N, zero_N, x, -y))
D_2 = np.hstack((zero_N, one_N, y, x))
Dt = np.vstack((D_1, D_2))
#print(N_c, Dt, np.dot(Dt.T, Dt))
D_inv = np.linalg.inv(np.dot(Dt.T, Dt))
gamma = np.dot(D_inv, np.dot(Dt.T, w))
A = np.sqrt(gamma[2]**2 + gamma[3]**2)
theta = np.arctan2(gamma[3], gamma[2]) * 180/np.pi
R_denominator = np.dot(w.T, w) - N_c*(iw_u_ave**2 + iw_v_ave**2)
epsilon = w - np.dot(Dt, gamma)
R_numerator = np.dot(epsilon.T, epsilon)
R2 = 1 - (R_numerator/R_denominator)[0,0]
print("\t{}".format([A[0], theta[0], gamma[0,0], gamma[1,0], R2, N_c]))
if N_c < 45:
Helmert.append([np.nan, np.nan, gamma[0], gamma[1], np.nan, np.nan, N_c, iw_u_ave, iw_v_ave])
else:
Helmert.append([A, theta, gamma[0], gamma[1], R2, R_numerator, N_c, iw_u_ave, iw_v_ave])
result = np.hstack((Helmert, data_ave_w[:,[1,2]]))
#print(result.shape)
data = pd.DataFrame(result)
data.columns = ["A_90", "theta_90", "ocean_u_90", "ocean_v_90", "R2_90", "epsilon2_90", "N_c_90", "mean_iw_u_90", "mean_iw_v_90", "mean_w_u_90", "mean_w_v_90"]
data["ocean_speed_90"] = np.sqrt(data["ocean_u_90"]**2 + data["ocean_v_90"]**2)
#print(data.head(3))
data = pd.concat([latlon_ex, data_ex_basic, data], axis=1)
save_name = dirs + "Helmert_both_90_" + str(start_1)[:6] + ".csv"
print(save_name)
data.to_csv(save_name, index=False)
def H_scatter_6win_2area(worker):
dirs_basic = "../result_h/scatter/"
if worker == 0:
dirs_A_ic0 = dirs_basic + "A_ic0/"
mkdir(dirs_A_ic0)
dirs_theta_ic0 = dirs_basic + "theta_ic0/"
mkdir(dirs_theta_ic0)
dirs_e2_ic0 = dirs_basic + "e2_ic0/"
mkdir(dirs_e2_ic0)
elif worker == 1:
dirs_A_sit = dirs_basic + "A_sit/"
mkdir(dirs_A_sit)
dirs_theta_sit = dirs_basic + "theta_sit/"
mkdir(dirs_theta_sit)
dirs_e2_sit = dirs_basic + "e2_sit/"
mkdir(dirs_e2_sit)
sns.set_style("darkgrid")
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(
i + 1, M))
month_end = start_list_plus_1month[i + 1]
month_end = date(month_end // 10000, (month_end % 10000) // 100,
(month_end % 10000) % 100) - timedelta(days=1)
end = start + month_end.day - 1
_, _, _, data_hermert = main_data(start,
start,
span=30,
get_columns=["hermert"],
region=None,
accumulate=False)
_, _, _, data_ic0_sit = main_data(start,
end,
span=30,
get_columns=["ic0_145", "sit_145"],
region=None,
accumulate=True)
data_array = np.array(data_ic0_sit)
data_ave = np.nanmean(data_array, axis=0)
data_ave = pd.DataFrame(data_ave)
data_ave.columns = ["ic0_30", "sit_30"]
data_cross = data_array[:, :, 0] * data_array[:, :, 1]
data_cross_ave = np.nanmean(data_cross, axis=0) / 100
data_ave["cross_ic0_sit"] = data_cross_ave
data_tmp = data_hermert.loc[:, ["A", "theta", "R2", "epsilon2"]]
data_basic = pd.concat([latlon_ex, data_tmp, data_ave], axis=1)
rank_np = np.zeros(145**2)
rank_np[data_basic[data_basic.Name == "north_polar"].index] = 1
rank_R2 = np.ones(145**2)
rank_R2[data_basic[data_basic.R2 <= (0.5)**2].index] = 0
rank_R2[data_basic[data_basic.R2 > (0.6)**2].index] = 2
data_rank = pd.DataFrame({"rank_np": rank_np, "rank_R2": rank_R2})
data = pd.concat([data_basic, data_rank], axis=1)
if worker == 0:
save_name = dirs_A_ic0 + str(start)[:6] + ".png"
sns.lmplot(x="ic0_30",
y="A",
row="rank_R2",
col="rank_np",
data=data,
size=3,
fit_reg=True)
plt.savefig(save_name, dpi=900)
plt.close()
save_name = dirs_theta_ic0 + str(start)[:6] + ".png"
sns.lmplot(x="ic0_30",
y="theta",
row="rank_R2",
col="rank_np",
data=data,
size=3)
plt.savefig(save_name, dpi=900)
plt.close()
save_name = dirs_e2_ic0 + str(start)[:6] + ".png"
sns.lmplot(x="ic0_30",
y="epsilon2",
row="rank_R2",
col="rank_np",
data=data,
size=3)
plt.savefig(save_name, dpi=900)
plt.close()
elif worker == 1:
save_name = dirs_A_sit + str(start)[:6] + ".png"
sns.lmplot(x="sit_30",
y="A",
row="rank_R2",
col="rank_np",
data=data,
size=3)
plt.savefig(save_name, dpi=900)
plt.close()
save_name = dirs_theta_sit + str(start)[:6] + ".png"
sns.lmplot(x="sit_30",
y="theta",
row="rank_R2",
col="rank_np",
data=data,
size=3)
plt.savefig(save_name, dpi=900)
plt.close()
save_name = dirs_e2_sit + str(start)[:6] + ".png"
sns.lmplot(x="sit_30",
y="epsilon2",
row="rank_R2",
col="rank_np",
data=data,
size=3)
plt.savefig(save_name, dpi=900)
plt.close()
print("\n")
def test_ocean_plot():
dirs = "../result_h/test/test_iw/"
mkdir(dirs)
start_list_plus_1month = start_list + [20170901]
for i, start in enumerate(start_list):
print("******************* {}/{} *******************".format(
i + 1, M))
month_end = start_list_plus_1month[i + 1]
month_end = date(month_end // 10000, (month_end % 10000) // 100,
(month_end % 10000) % 100) - timedelta(days=1)
end = start + month_end.day - 1
try:
#csv_Helmert_90の当該月のcsvを読み込む
#hermert_file_name_90 = "../data/csv_Helmert_both_90/Helmert_both_90_" + str(start)[:6] + ".csv"
#helmert_90_data = pd.read_csv(hermert_file_name_90)
#csv_Helmert_30の当該月のcsvを読み込む
hermert_file_name_30 = "../data/csv_Helmert_30/Helmert_30_" + str(
start)[:6] + ".csv"
helmert_30_data = pd.read_csv(hermert_file_name_30)
#木村のcsvの読み込み(90)
#coeff_file_name_90 = "../data/csv_A_90/ssc_amsr_ads" + str(start)[2:6] + "_90_fin.csv"
#coeff_90_data = calc_data.get_1month_coeff_data(coeff_file_name_90)
#木村のcsvの読み込み(30)
coeff_file_name_30 = "../data/csv_A_30/ssc_amsr_ads" + str(
start)[2:6] + "_30_fin.csv"
coeff_30_data = calc_data.get_1month_coeff_data(coeff_file_name_30)
except:
continue
#ocean_90_h_u = helmert_90_data["ocean_u_90"].values
#ocean_90_h_v = helmert_90_data["ocean_v_90"].values
#ocean_90_h_speed = np.sqrt(ocean_90_h_u**2 + ocean_90_h_v**2)
ocean_30_h_u = helmert_30_data["ocean_u"].values
ocean_30_h_v = helmert_30_data["ocean_v"].values
ocean_30_h_speed = np.sqrt(ocean_30_h_u**2 + ocean_30_h_v**2)
#ocean_90_c_u = coeff_90_data["mean_ocean_u"].values
#ocean_90_c_v = coeff_90_data["mean_ocean_v"].values
#ocean_90_c_speed = np.sqrt(ocean_90_c_u**2 + ocean_90_c_v**2)
ocean_30_c_u = coeff_30_data["mean_ocean_u"].values
ocean_30_c_v = coeff_30_data["mean_ocean_v"].values
ocean_30_c_speed = np.sqrt(ocean_30_c_u**2 + ocean_30_c_v**2)
"""
fig, axes = plt.subplots(1, 2)
ratio_90 = ocean_90_c_speed/ocean_90_h_speed
ratio_30 = ocean_30_c_speed/ocean_30_h_speed
axes[0].hist(ratio_30[ocean_idx][~np.isnan(ratio_30[ocean_idx])], range=(0,3), bins=150)
axes[1].hist(ratio_90[ocean_idx][~np.isnan(ratio_90[ocean_idx])], range=(0,3), bins=150)
"""
ratio_w = ocean_30_c_u / ocean_30_h_u
N_c_h = helmert_30_data["mean_iw_v"].values
N_c_c = coeff_30_data["mean_ice_v"].values
N_c_diff = N_c_h - N_c_c
#plt.hist(N_c_diff[~np.isnan(N_c_diff)], range=(-5,5), bins=11, alpha=0.5)
#tmp = N_c_diff[~np.isnan(N_c_diff)]
#print(len(np.where(tmp>0)[0]))
m = Basemap(lon_0=180,
boundinglat=50,
resolution='l',
projection='npstere')
m.drawcoastlines(color='0.15')
m.fillcontinents(color='#555555')
lon = np.array(latlon_ex.Lon)
lat = np.array(latlon_ex.Lat)
x, y = m(lon, lat)
x1 = np.reshape(x, (145, 145), order='F')
y1 = np.reshape(y, (145, 145), order='F')
dx1 = (x1[1, 0] - x1[0, 0]) / 2
dy1 = (y1[0, 1] - y1[0, 0]) / 2
x2 = np.linspace(x1[0, 0], x1[144, 0], 145)
y2 = np.linspace(y1[0, 0], y1[0, 144], 145)
xx, yy = np.meshgrid(x2, y2)
xx, yy = xx.T, yy.T
xx = np.hstack([xx, xx[:, 0].reshape(145, 1)])
xx_ex = np.vstack([xx, (xx[144, :] + (xx[1, 0] - xx[0, 0]))])
yy = np.vstack([yy, yy[0, :]])
yy_ex = np.hstack([(yy[:, 0].reshape(146, 1) + (yy[0, 0] - yy[0, 1])),
yy])
data = np.ma.masked_invalid(N_c_diff)
data1 = np.reshape(data, (145, 145), order='F')
m.pcolormesh(xx_ex - dx1,
yy_ex + dy1,
data1,
cmap=plt.cm.jet,
vmax=0.5,
vmin=-0.5)
#fig.colorbar(im, ax=ax)
m.colorbar()
#plt.hist(N_c_c[~np.isnan(N_c_c)], range=(20,35), bins=16)
plt.savefig(dirs + "iw_v_" + str(start)[:6] + ".png", dpi=450)
plt.close()
def H_scatter_ic0_np(mode):
if mode == "mean":
dirs_A_30_and_ic0_np = "../result_h/scatter/scatter_A_30_and_ic0_np/"
mkdir(dirs_A_30_and_ic0_np)
dirs_A_30_and_ic0_h_np = "../result_h/scatter/scatter_A_30_and_ic0_h_np/"
mkdir(dirs_A_30_and_ic0_h_np)
dirs_angle_30_and_ic0_np = "../result_h/scatter/scatter_angle_30_and_ic0_np/"
mkdir(dirs_angle_30_and_ic0_np)
dirs_angle_30_and_ic0_h_np = "../result_h/scatter/scatter_angle_30_and_ic0_h_np/"
mkdir(dirs_angle_30_and_ic0_h_np)
elif mode == "median":
dirs_A_30_and_ic0_median_np = "../result_h/scatter/scatter_A_30_and_ic0_median_np/"
mkdir(dirs_A_30_and_ic0_median_np)
dirs_A_30_and_ic0_median_h_np = "../result_h/scatter/scatter_A_30_and_ic0_median_h_np/"
mkdir(dirs_A_30_and_ic0_median_h_np)
dirs_angle_30_and_ic0_median_np = "../result_h/scatter/scatter_angle_30_and_ic0_median_np/"
mkdir(dirs_angle_30_and_ic0_median_np)
dirs_angle_30_and_ic0_median_h_np = "../result_h/scatter/scatter_angle_30_and_ic0_median_h_np/"
mkdir(dirs_angle_30_and_ic0_median_h_np)
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(
i + 1, M))
month_end = start_list_plus_1month[i + 1]
month_end = date(month_end // 10000, (month_end % 10000) // 100,
(month_end % 10000) % 100) - timedelta(days=1)
end = start + month_end.day - 1
_, _, _, data_A_original = main_data(start,
start,
span=30,
get_columns=["hermert"],
region=None,
accumulate=False)
_, _, _, data_ic0_30 = main_data(start,
end,
span=30,
get_columns=["ic0_145"],
region=None,
accumulate=True)
data_array = np.array(data_ic0_30)
if mode == "mean":
data_ave = np.nanmean(data_array, axis=0)
data_ave = pd.DataFrame(data_ave)
data_ave.columns = ["ic0_30"]
#dirs_A_30_and_ic0_np
data_A = data_A_original["A"]
data = pd.concat([latlon_ex, data_A, data_ave], axis=1)
data = data[data.Name == "north_polar"]
save_name = dirs_A_30_and_ic0_np + str(start)[:6] + ".png"
visualize.visual_non_line(data,
mode=["scatter", ["ic0_30", "A"]],
save_name=save_name,
show=False)
#dirs_A_30_and_ic0_h_np
"""
data_A_1 = data_A_original.loc[:, ["A", "R2"]]
data_A_1.A[data_A_1.R2<threshold_R2] = np.nan
data_1 = pd.concat([latlon_ex, data_A_1, data_ave], axis=1)
data_1 = data_1[data_1.Name=="north_polar"]
save_name = dirs_A_30_and_ic0_h_np + str(start)[:6] + ".png"
visualize.visual_non_line(
data_1,
mode=["scatter", ["ic0_30", "A"]],
save_name=save_name,
show=False
)
"""
#dirs_angle_30_and_ic0_np
data_angle = data_A_original["theta"]
data_2 = pd.concat([latlon_ex, data_angle, data_ave], axis=1)
data_2 = data_2[data_2.Name == "north_polar"]
save_name = dirs_angle_30_and_ic0_np + str(start)[:6] + ".png"
visualize.visual_non_line(data_2,
mode=["scatter", ["ic0_30", "theta"]],
save_name=save_name,
show=False)
#dirs_angle_30_and_ic0_h_np
"""
data_angle_1 = data_A_original.loc[:, ["theta", "R2"]]
data_angle_1.theta[data_angle_1.R2<threshold_R2] = np.nan
data_3 = pd.concat([latlon_ex, data_angle_1, data_ave], axis=1)
data_3 = data_3[data_3.Name=="north_polar"]
save_name = dirs_angle_30_and_ic0_h_np + str(start)[:6] + ".png"
visualize.visual_non_line(
data_3,
mode=["scatter", ["ic0_30", "theta"]],
save_name=save_name,
show=False
)
"""
elif mode == "median":
data_ave = np.nanmedian(data_array, axis=0)
data_ave = pd.DataFrame(data_ave)
data_ave.columns = ["ic0_30"]
#dirs_A_30_and_ic0_np
data_A = data_A_original["A"]
data = pd.concat([latlon_ex, data_A, data_ave], axis=1)
data = data[data.Name == "north_polar"]
save_name = dirs_A_30_and_ic0_median_np + str(start)[:6] + ".png"
visualize.visual_non_line(data,
mode=["scatter", ["ic0_30", "A"]],
save_name=save_name,
show=False)
#dirs_A_30_and_ic0_h_np
"""
data_A_1 = data_A_original.loc[:, ["A", "R2"]]
data_A_1.A[data_A_1.R2<threshold_R2] = np.nan
data_1 = pd.concat([latlon_ex, data_A_1, data_ave], axis=1)
data_1 = data_1[data_1.Name=="north_polar"]
save_name = dirs_A_30_and_ic0_median_h_np + str(start)[:6] + ".png"
visualize.visual_non_line(
data_1,
mode=["scatter", ["ic0_30", "A"]],
save_name=save_name,
show=False
)
"""
#dirs_angle_30_and_ic0_np
data_angle = data_A_original["theta"]
data_2 = pd.concat([latlon_ex, data_angle, data_ave], axis=1)
data_2 = data_2[data_2.Name == "north_polar"]
save_name = dirs_angle_30_and_ic0_median_np + str(
start)[:6] + ".png"
visualize.visual_non_line(data_2,
mode=["scatter", ["ic0_30", "theta"]],
save_name=save_name,
show=False)
#dirs_angle_30_and_ic0_h_np
"""
data_angle_1 = data_A_original.loc[:, ["theta", "R2"]]
data_angle_1.theta[data_angle_1.R2<threshold_R2] = np.nan
data_3 = pd.concat([latlon_ex, data_angle_1, data_ave], axis=1)
data_3 = data_3[data_3.Name=="north_polar"]
save_name = dirs_angle_30_and_ic0_median_h_np + str(start)[:6] + ".png"
visualize.visual_non_line(
data_3,
mode=["scatter", ["ic0_30", "theta"]],
save_name=save_name,
show=False
)
"""
print("\n")
def get_csv_ex(array_label, array_coastal_region_1, array_coastal_region_2):
data_ex_dir = "../data/csv_Helmert_ex/"
mkdir(data_ex_dir)
dirs_pair = "../result_h/pairplot/"
mkdir(dirs_pair)
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(
i + 1, M))
month_end = start_list_plus_1month[i + 1]
month_end = date(month_end // 10000, (month_end % 10000) // 100,
(month_end % 10000) % 100) - timedelta(days=1)
end = start + month_end.day - 1
_, _, _, data_ic0_30 = main_data(start,
end,
span=30,
get_columns=["ic0_145"],
region=None,
accumulate=True)
data_array_ic0 = np.array(data_ic0_30)
data_ave_ic0 = np.nanmean(data_array_ic0, axis=0).ravel()
data_med_ic0 = np.nanmedian(data_array_ic0, axis=0).ravel()
data_ic0_ave_med_30 = pd.DataFrame({
"ic0_30": data_ave_ic0,
"ic0_30_median": data_med_ic0
})
_, _, _, data_sit_30 = main_data(start,
end,
span=30,
get_columns=["sit_145"],
region=None,
accumulate=True)
data_array_sit = np.array(data_sit_30)
data_ave_sit = np.nanmean(data_array_sit, axis=0).ravel()
data_med_sit = np.nanmedian(data_array_sit, axis=0).ravel()
data_sit_ave_med_30 = pd.DataFrame({
"sit_30": data_ave_sit,
"sit_30_median": data_med_sit
})
_, _, _, data_A_original = main_data(start,
start,
span=30,
get_columns=["hermert"],
region=None,
accumulate=False)
data_A_original["mean_ocean_speed"] = np.sqrt(
data_A_original["ocean_u"]**2 + data_A_original["ocean_v"]**2)
data_A_original["mean_iw_speed"] = np.sqrt(
data_A_original["mean_iw_u"]**2 + data_A_original["mean_iw_v"]**2)
data_A_original["mean_w_speed"] = np.sqrt(
data_A_original["mean_w_u"]**2 + data_A_original["mean_w_v"]**2)
data_A_original["w_iw_ratio"] = data_A_original[
"mean_iw_speed"] / data_A_original["mean_w_speed"]
data_A_original["mean_iw_speed_no_ocean"] = np.sqrt(
(data_A_original["mean_iw_u"] - data_A_original["ocean_u"])**2 +
(data_A_original["mean_iw_v"] - data_A_original["ocean_v"])**2)
data = pd.concat([
latlon_ex, data_A_original, data_ic0_ave_med_30,
data_sit_ave_med_30
],
axis=1)
#data["t_ocean"] = t_ocean
#data["t_iw"] = t_iw
#print(len(data))
#print(array_coastal_region_1.shape)
data["coastal_region_1"] = array_coastal_region_1
data["coastal_region_2"] = array_coastal_region_2
data["area_label"] = array_label
#print("\n\n{}\n".format(data.head(3)))
data_name = data_ex_dir + "Helmert_ex_" + str(start)[:-2] + ".csv"
data.to_csv(data_name, index=False)
data_no_nan = data_A_original.dropna()
sns.pairplot(
data_no_nan.loc[:, ["A", "R2", "epsilon2", "mean_ocean_speed"]])
plt.savefig(dirs_pair + "ocean_" + str(start)[:-2] + ".png")
plt.close()
sns.pairplot(data_no_nan.loc[:,
["A", "R2", "epsilon2", "mean_iw_speed"]])
plt.savefig(dirs_pair + "iw_" + str(start)[:-2] + ".png")
plt.close()
sns.pairplot(data_no_nan.loc[:, ["A", "R2", "epsilon2", "w_iw_ratio"]])
plt.savefig(dirs_pair + "ratio_" + str(start)[:-2] + ".png")
plt.close()
sns.pairplot(
data_no_nan.loc[:,
["A", "R2", "epsilon2", "mean_iw_speed_no_ocean"]])
plt.savefig(dirs_pair + "iw_no_ocean_" + str(start)[:-2] + ".png")
plt.close()
sns.pairplot(data_no_nan.loc[:, [
"mean_iw_speed", "mean_w_speed", "mean_ocean_speed",
"mean_iw_speed_no_ocean"
]])
plt.savefig(dirs_pair + "speed_" + str(start)[:-2] + ".png")
plt.close()
print("\n")
def plot_anomaly_points():
dirs = "../result_h/A/anomaly_ic0/"
mkdir(dirs)
start_list.pop()
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(
i + 1, M - 1))
helmert_30_30_fname = "../data/csv_Helmert_both_30/Helmert_both_30_" + str(
start)[:6] + ".csv"
data_30 = pd.read_csv(helmert_30_30_fname)
helmert_30_90_fname = "../data/csv_Helmert_ocean_90/Helmert_30_90_" + str(
start)[:6] + ".csv"
data_90 = pd.read_csv(helmert_30_90_fname)
#anomaly_30 = data_30.loc[((data_30.A>=0.005)&(data_30.ic0_30>=97)), "A"]
#anomaly_90 = data_90.loc[((data_90.A_30_90>=0.005)&(data_90.ic0_30>=97)), "A_30_90"]
anomaly_bool_30 = (data_30.A >= 0.005) & (data_30.ic0_30 >= 97)
data_30["anomaly"] = data_30.epsilon2[anomaly_bool_30]
anomaly_bool_90 = (data_90.A_30_90 >= 0.005) & (data_90.ic0_30 >= 97)
data_90["anomaly"] = data_90.epsilon2_30_90[anomaly_bool_90]
fig, axes = plt.subplots(1, 2)
data_list = [
np.array(data_30["anomaly"]),
np.array(data_90["anomaly"])
]
for j in range(2):
ax = axes[j]
m = Basemap(lon_0=180,
boundinglat=50,
resolution='l',
projection='npstere')
m.ax = ax
m.drawcoastlines(color='0.15')
m.fillcontinents(color='#555555')
lon = np.array(latlon_ex.Lon)
lat = np.array(latlon_ex.Lat)
x, y = m(lon, lat)
x1 = np.reshape(x, (145, 145), order='F')
y1 = np.reshape(y, (145, 145), order='F')
dx1 = (x1[1, 0] - x1[0, 0]) / 2
dy1 = (y1[0, 1] - y1[0, 0]) / 2
x2 = np.linspace(x1[0, 0], x1[144, 0], 145)
y2 = np.linspace(y1[0, 0], y1[0, 144], 145)
xx, yy = np.meshgrid(x2, y2)
xx, yy = xx.T, yy.T
xx = np.hstack([xx, xx[:, 0].reshape(145, 1)])
xx_ex = np.vstack([xx, (xx[144, :] + (xx[1, 0] - xx[0, 0]))])
yy = np.vstack([yy, yy[0, :]])
yy_ex = np.hstack([
(yy[:, 0].reshape(146, 1) + (yy[0, 0] - yy[0, 1])), yy
])
data = np.ma.masked_invalid(data_list[j])
data1 = np.reshape(data, (145, 145), order='F')
im = m.pcolormesh(xx_ex - dx1,
yy_ex + dy1,
data1,
cmap=plt.cm.jet,
vmax=1,
vmin=0)
fig.colorbar(im, ax=ax)
#m.colorbar(location='bottom')
save_name = dirs + str(start)[:6] + ".png"
print(save_name)
plt.savefig(save_name, dpi=500)
plt.close()
def ocean_30_vs_90():
dirs = "../result_h/mean_vector/ocean_currents_30_vs_90/"
mkdir(dirs)
start_list.pop()
for i, start in enumerate(start_list):
print("****************** {}/{} *******************".format(
i + 1, M))
helmert_30_30_fname = "../data/csv_Helmert_30/Helmert_30_" + str(
start)[:6] + ".csv"
data_30 = pd.read_csv(helmert_30_30_fname)
data_30_vec = [
np.array(data_30["ocean_u"]),
np.array(data_30["ocean_v"])
]
helmert_90_90_fname = "../data/csv_Helmert_both_90/Helmert_both_90_" + str(
start)[:6] + ".csv"
data_90 = pd.read_csv(helmert_90_90_fname)
data_90_vec = [
np.array(data_90["ocean_u_90"]),
np.array(data_90["ocean_v_90"])
]
helmert_30_30_w_iw_fname = "../data/csv_Helmert_both_30_w_iw/Helmert_both_30_w_iw_" + str(
start)[:6] + ".csv"
data_30_w_iw = pd.read_csv(helmert_30_30_w_iw_fname)
data_30_w_iw_vec = [
np.array(data_30_w_iw["ocean_u_w_iw"]),
np.array(data_30_w_iw["ocean_v_w_iw"])
]
fig, axes = plt.subplots(1, 3)
fig.figsize = (6, 9)
title_list = ["30", "90", "30_w_iw"]
vector_list = [data_30_vec, data_90_vec, data_30_w_iw_vec]
for j, title in enumerate(title_list):
ax = axes[j]
ax.set_title(title)
m = Basemap(lon_0=180,
boundinglat=50,
resolution='l',
projection='npstere')
m.ax = ax
m.drawcoastlines(color='0.15')
m.fillcontinents(color='#555555')
lon = np.array(latlon_ex.Lon)
lat = np.array(latlon_ex.Lat)
x, y = m(lon, lat)
x1 = np.reshape(x, (145, 145), order='F')
y1 = np.reshape(y, (145, 145), order='F')
dx1 = (x1[1, 0] - x1[0, 0]) / 2
dy1 = (y1[0, 1] - y1[0, 0]) / 2
x2 = np.linspace(x1[0, 0], x1[144, 0], 145)
y2 = np.linspace(y1[0, 0], y1[0, 144], 145)
xx, yy = np.meshgrid(x2, y2)
xx, yy = xx.T, yy.T
vector_u = np.ma.masked_invalid(vector_list[j][0])
vector_v = np.ma.masked_invalid(vector_list[j][1])
vector_speed = np.sqrt(vector_u * vector_u + vector_v * vector_v)
data_non_wind = vector_speed
data_non_wind = np.ma.masked_invalid(data_non_wind)
data1 = np.reshape(data_non_wind, (145, 145), order='F')
xx = np.hstack([xx, xx[:, 0].reshape(145, 1)])
xx_ex = np.vstack([xx, (xx[144, :] + (xx[1, 0] - xx[0, 0]))])
yy = np.vstack([yy, yy[0, :]])
yy_ex = np.hstack([
(yy[:, 0].reshape(146, 1) + (yy[0, 0] - yy[0, 1])), yy
])
m.pcolormesh(xx_ex - dx1,
yy_ex + dy1,
data1,
cmap=plt.cm.jet,
vmax=0.2,
vmin=0)
#m.colorbar(location='bottom')
m.quiver(x, y, vector_u, vector_v, color="k")
save_name = dirs + str(start)[:6] + ".png"
print(save_name)
plt.savefig(save_name, dpi=750)
plt.close()
print("\n")
def H_ts_A_month(m_plot=True, y_plot=True):
dirs_1g = "../result_h/H_ts_month_1g/"
mkdir(dirs_1g)
dirs_2g = "../result_h/H_ts_month_2g/"
mkdir(dirs_2g)
dirs_3g = "../result_h/H_ts_month_3g/"
mkdir(dirs_3g)
dirs_year = "../result_h/H_ts_year/"
mkdir(dirs_year)
def plot_param_1g(data_m_1g, save_name):
fig, axes = plt.subplots(3, 1)
dates = pd.date_range("2001", periods=12, freq='MS')
plot_data_1g = data_m_1g["A"].loc[:, ["mean", "std", "50%"]]
plot_data_1g["2sigma_pos"] = plot_data_1g['mean'] + 1 * np.sqrt(
plot_data_1g['std'])
plot_data_1g["2sigma_neg"] = plot_data_1g['mean'] - 1 * np.sqrt(
plot_data_1g['std'])
#plot_data_1g["Month"] = dates
axes[0].plot(dates, plot_data_1g['mean'], '-', color="k")
#d = plot_data_1g['Month'].values
axes[0].fill_between(dates,
plot_data_1g['2sigma_pos'],
plot_data_1g['2sigma_neg'],
facecolor='green',
alpha=0.3,
interpolate=True)
#axes[0].set_ylim([0, 0.025])
axes[0].set_ylabel('A')
#axes[0].set_title('A, ' + str_year)
axes[0].xaxis.set_major_formatter(mdates.DateFormatter('%m'))
plot_data_1g_theta = data_m_1g["theta"].loc[:, ["mean", "std", "50%"]]
plot_data_1g_theta["2sigma_pos"] = plot_data_1g_theta[
'mean'] + 2 * np.sqrt(plot_data_1g_theta['std'])
plot_data_1g_theta["2sigma_neg"] = plot_data_1g_theta[
'mean'] - 2 * np.sqrt(plot_data_1g_theta['std'])
axes[1].plot(dates, plot_data_1g_theta['mean'], '-', color="k")
axes[1].fill_between(dates,
plot_data_1g_theta['2sigma_pos'],
plot_data_1g_theta['2sigma_neg'],
facecolor='lightskyblue',
alpha=0.3,
interpolate=True)
#axes[1].set_ylim([-180, 180])
axes[1].set_ylim([-60, 60])
#axes[1].set_yticks([-180, -120, -60, 0, 60, 120, 180])
axes[1].set_ylabel('theta')
#axes[1].set_title('theta, ' + str_year)
axes[1].xaxis.set_major_formatter(mdates.DateFormatter('%m'))
plot_data_1g_e2 = data_m_1g["epsilon2"].loc[:, ["mean", "std", "50%"]]
plot_data_1g_e2["2sigma_pos"] = plot_data_1g_e2['mean'] + 2 * np.sqrt(
plot_data_1g_e2['std'])
plot_data_1g_e2["2sigma_neg"] = plot_data_1g_e2['mean'] - 2 * np.sqrt(
plot_data_1g_e2['std'])
axes[2].plot(dates, plot_data_1g_e2['mean'], '-', color="k")
axes[2].fill_between(dates,
plot_data_1g_e2['2sigma_pos'],
plot_data_1g_e2['2sigma_neg'],
facecolor='silver',
alpha=0.3,
interpolate=True)
#axes[2].set_ylim([-180, 180])
#axes[2].set_yticks([-180, -120, -60, 0, 60, 120, 180])
axes[2].set_ylabel('e2')
#axes[2].set_title('e2, ' + str_year)
axes[2].xaxis.set_major_formatter(mdates.DateFormatter('%m'))
#plt.setp(axes[0].get_xticklabels(), visible=False)
#plt.setp(axes[1].get_xticklabels(), visible=False)
plt.tight_layout()
plt.savefig(save_name, dpi=900)
plt.close()
def plot_param_2g(data_m_2g, save_name):
plot_param_list = ["A", "theta", "epsilon2"]
fig, axes = plt.subplots(3, 2)
dates = pd.date_range("2001", periods=12, freq='MS')
axes[0, 0].set_title("Polar Region")
axes[0, 1].set_title("Coastal Region")
for i, item in enumerate(plot_param_list):
for j, is_np in enumerate([1, 0]):
plot_data_2g_np_pos = data_m_2g.loc[(is_np),
(item,
["mean", "std", "50%"])]
plot_data_2g_np_pos["2sigma_pos"] = plot_data_2g_np_pos[
(item,
'mean')] + 2 * np.sqrt(plot_data_2g_np_pos[(item, 'std')])
plot_data_2g_np_pos["2sigma_neg"] = plot_data_2g_np_pos[
(item,
'mean')] - 2 * np.sqrt(plot_data_2g_np_pos[(item, 'std')])
#plot_data_2g_np_pos["Month"] = np.arange(1, 13, 1)
ax = axes[i, j]
#plt.subplot(321+j+i*2)
ax.plot_date(dates,
plot_data_2g_np_pos[(item, 'mean')],
'-',
color="k")
#d = plot_data_2g_np_pos['Month'].values
ax.fill_between(dates,
plot_data_2g_np_pos['2sigma_pos'],
plot_data_2g_np_pos['2sigma_neg'],
facecolor='lightskyblue',
alpha=0.3,
interpolate=True)
if i == 0:
ax.set_ylim([0, 0.025])
#ax.set_yticks([0, 0.025])
elif i == 1:
ax.set_ylim([-180, 180])
ax.set_yticks([-180, -120, -60, 0, 60, 120, 180])
elif i == 2:
ax.set_ylim([0, 1.5])
#ax.set_yticks([0, 0.025])
ax.xaxis.set_major_formatter(mdates.DateFormatter('%m'))
axes[0, 0].set_ylabel('A')
axes[1, 0].set_ylabel('theta')
axes[2, 0].set_ylabel('e2')
axes[2, 0].set_xlabel("Month")
axes[2, 1].set_xlabel("Month")
"""
plt.setp(axes[0, 0].get_xticklabels(), visible=False)
plt.setp(axes[1, 0].get_xticklabels(), visible=False)
plt.setp(axes[0, 1].get_xticklabels(), visible=False)
plt.setp(axes[1, 1].get_xticklabels(), visible=False)
plt.setp(axes[0, 1].get_yticklabels(), visible=False)
plt.setp(axes[1, 1].get_yticklabels(), visible=False)
plt.setp(axes[2, 1].get_yticklabels(), visible=False)
"""
plt.tight_layout()
plt.savefig(save_name, dpi=900)
plt.close()
def plot_param_3g(data_m_3g, plot_param_item, save_name):
fig, axes = plt.subplots(3, 2)
dates = pd.date_range("2001", periods=12, freq='MS')
axes[0, 0].set_title("Polar Region")
axes[0, 1].set_title("Coastal Region")
for i, is_np in enumerate([1, 0]):
plot_data_3g_np_pos = data_m_3g.loc[(is_np),
(plot_param_item,
["mean", "std", "50%"])]
plot_data_3g_np_pos["2sigma_pos"] = plot_data_3g_np_pos[
(plot_param_item, "mean")] + 2 * np.sqrt(plot_data_3g_np_pos[
(plot_param_item, "std")])
plot_data_3g_np_pos["2sigma_neg"] = plot_data_3g_np_pos[
(plot_param_item, "mean")] - 2 * np.sqrt(plot_data_3g_np_pos[
(plot_param_item, "std")])
for j, subplot_idx in enumerate([2, 1, 0]):
ax = axes[j, i]
ax.plot(dates,
plot_data_3g_np_pos.loc[(subplot_idx),
(plot_param_item, "mean")],
'-',
color="k")
ax.fill_between(dates,
plot_data_3g_np_pos.loc[(subplot_idx),
("2sigma_pos")],
plot_data_3g_np_pos.loc[(subplot_idx),
("2sigma_neg")],
facecolor='lightskyblue',
alpha=0.3,
interpolate=True)
if plot_param_item == "A":
for ax in axes:
ax.set_ylim([0, 0.025])
elif plot_param_item == "theta":
for ax in axes:
ax.set_ylim([-180, 180])
ax.set_yticks([-180, -120, -60, 0, 60, 120, 180])
elif plot_param_item == "e2":
for ax in axes:
ax.set_ylim([0, 1.5])
axes[0, 0].set_ylabel('R2 High')
axes[1, 0].set_ylabel('R2 Middle')
axes[2, 0].set_ylabel('R2 Low')
axes[2, 0].set_xlabel("Month")
axes[2, 1].set_xlabel("Month")
"""
plt.setp(axes[0, 0].get_xticklabels(), visible=False)
plt.setp(axes[1, 0].get_xticklabels(), visible=False)
plt.setp(axes[0, 1].get_xticklabels(), visible=False)
plt.setp(axes[1, 1].get_xticklabels(), visible=False)
plt.setp(axes[0, 1].get_yticklabels(), visible=False)
plt.setp(axes[1, 1].get_yticklabels(), visible=False)
plt.setp(axes[2, 1].get_yticklabels(), visible=False)
"""
plt.tight_layout()
plt.savefig(save_name, dpi=900)
plt.close()
def plot_param_1g_through_years(data_1g_dic, save_name):
fig, axes = plt.subplots(3, 1)
dates1 = pd.date_range("2003", "2010", freq='MS').append(
pd.date_range("2013", "2017", freq='MS'))
dates = dates1[:-1]
data_A = pd.DataFrame([])
data_theta = pd.DataFrame([])
data_e2 = pd.DataFrame([])
for y in y_list:
data_A = pd.concat([data_A, data_1g_dic[y]["1g_A"]])
data_theta = pd.concat([data_theta, data_1g_dic[y]["1g_theta"]])
data_e2 = pd.concat([data_e2, data_1g_dic[y]["1g_e2"]])
axes[0].plot(dates, data_A['mean'], '-', color="k")
axes[0].fill_between(dates,
data_A['mean'] + 2 * np.sqrt(data_A['std']),
data_A['mean'] - 2 * np.sqrt(data_A['std']),
facecolor='green',
alpha=0.3,
interpolate=True)
axes[0].set_ylim([0, 0.025])
axes[0].set_ylabel('A')
axes[0].xaxis.set_major_formatter(mdates.DateFormatter('%y%m'))
axes[1].plot(dates, data_theta['mean'], '-', color="k")
axes[1].fill_between(
dates,
data_theta['mean'] + 2 * np.sqrt(data_theta['std']),
data_theta['mean'] - 2 * np.sqrt(data_theta['std']),
facecolor='green',
alpha=0.3,
interpolate=True)
axes[1].set_ylim([-180, 180])
axes[1].set_ylabel('theta')
axes[1].xaxis.set_major_formatter(mdates.DateFormatter('%y%m'))
axes[2].plot(dates, data_e2['mean'], '-', color="k")
axes[2].fill_between(dates,
data_e2['mean'] + 2 * np.sqrt(data_e2['std']),
data_e2['mean'] - 2 * np.sqrt(data_e2['std']),
facecolor='green',
alpha=0.3,
interpolate=True)
#axes[2].set_ylim([0, 0.025])
axes[2].set_ylabel('e2')
axes[2].xaxis.set_major_formatter(mdates.DateFormatter('%y%m'))
plt.tight_layout()
plt.savefig(save_name, dpi=900)
plt.close()
def plot_param_2g_through_years(data_2g_dic, save_name):
fig, axes = plt.subplots(3, 2)
dates1 = pd.date_range("2003", "2010", freq='MS').append(
pd.date_range("2013", "2017", freq='MS'))
dates = dates1[:-1]
data_A_np = pd.DataFrame([])
data_A_coastal = pd.DataFrame([])
data_theta_np = pd.DataFrame([])
data_theta_coastal = pd.DataFrame([])
data_e2_np = pd.DataFrame([])
data_e2_coastal = pd.DataFrame([])
for y in y_list:
data_A_np = pd.concat([data_A_np, data_2g_dic[y]["2g_A_polar"]])
data_A_coastal = pd.concat(
[data_A_coastal, data_2g_dic[y]["2g_A_coastal"]])
data_theta_np = pd.concat(
[data_theta_np, data_2g_dic[y]["2g_theta_polar"]])
data_theta_coastal = pd.concat(
[data_theta_coastal, data_2g_dic[y]["2g_theta_coastal"]])
data_e2_np = pd.concat([data_e2_np, data_2g_dic[y]["2g_e2_polar"]])
data_e2_coastal = pd.concat(
[data_e2_coastal, data_2g_dic[y]["2g_e2_coastal"]])
axes[0, 0].plot(dates, data_A_np['mean'], '-', color="k")
axes[0,
0].fill_between(dates,
data_A_np['mean'] + 2 * np.sqrt(data_A_np['std']),
data_A_np['mean'] - 2 * np.sqrt(data_A_np['std']),
facecolor='lightskyblue',
alpha=0.3,
interpolate=True)
axes[0, 0].set_ylim([0, 0.025])
axes[0, 0].set_ylabel('A')
axes[0, 0].xaxis.set_major_formatter(mdates.DateFormatter('%y%m'))
axes[0, 1].plot(dates, data_A_coastal['mean'], '-', color="k")
axes[0, 1].fill_between(
dates,
data_A_coastal['mean'] + 2 * np.sqrt(data_A_coastal['std']),
data_A_coastal['mean'] - 2 * np.sqrt(data_A_coastal['std']),
facecolor='lightskyblue',
alpha=0.3,
interpolate=True)
axes[0, 1].set_ylim([0, 0.025])
axes[0, 1].xaxis.set_major_formatter(mdates.DateFormatter('%y%m'))
axes[1, 0].plot(dates, data_theta_np['mean'], '-', color="k")
axes[1, 0].fill_between(
dates,
data_theta_np['mean'] + 2 * np.sqrt(data_theta_np['std']),
data_theta_np['mean'] - 2 * np.sqrt(data_theta_np['std']),
facecolor='lightskyblue',
alpha=0.3,
interpolate=True)
axes[1, 0].set_ylim([-180, 180])
axes[1, 0].set_ylabel('theta')
axes[1, 0].xaxis.set_major_formatter(mdates.DateFormatter('%y%m'))
axes[1, 1].plot(dates, data_theta_coastal['mean'], '-', color="k")
axes[1, 1].fill_between(dates,
data_theta_coastal['mean'] +
2 * np.sqrt(data_theta_coastal['std']),
data_theta_coastal['mean'] -
2 * np.sqrt(data_theta_coastal['std']),
facecolor='lightskyblue',
alpha=0.3,
interpolate=True)
axes[1, 1].set_ylim([-180, 180])
axes[1, 1].xaxis.set_major_formatter(mdates.DateFormatter('%y%m'))
axes[2, 0].plot(dates, data_e2_np['mean'], '-', color="k")
axes[2, 0].fill_between(
dates,
data_e2_np['mean'] + 2 * np.sqrt(data_e2_np['std']),
data_e2_np['mean'] - 2 * np.sqrt(data_e2_np['std']),
facecolor='lightskyblue',
alpha=0.3,
interpolate=True)
#axes[2, 0].set_ylim([0, 0.025])
axes[2, 0].set_ylabel('e2')
axes[2, 0].xaxis.set_major_formatter(mdates.DateFormatter('%y%m'))
axes[2, 1].plot(dates, data_e2_coastal['mean'], '-', color="k")
axes[2, 1].fill_between(
dates,
data_e2_coastal['mean'] + 2 * np.sqrt(data_e2_coastal['std']),
data_e2_coastal['mean'] - 2 * np.sqrt(data_e2_coastal['std']),
facecolor='lightskyblue',
alpha=0.3,
interpolate=True)
#axes[2, 1].set_ylim([0, 0.025])
axes[2, 1].xaxis.set_major_formatter(mdates.DateFormatter('%y%m'))
#axes[2, 0].set_xlabel("Year")
#axes[2, 1].set_xlabel("Year")
plt.tight_layout()
plt.savefig(save_name, dpi=900)
plt.close()
y_list = [
"03", "04", "05", "06", "07", "08", "09", "10", "13", "14", "15", "16"
]
month_list = [
"01", "02", "03", "04", "05", "06", "07", "08", "09", "10", "11", "12"
]
data_1g_dic = {}
data_2g_dic = {}
data_3g_dic = {}
for y in y_list:
data_m = pd.DataFrame([])
for m in month_list:
yymm = "20" + y + m
hermert_file_name = "../data/csv_Helmert_30/Helmert_30_" + yymm + ".csv"
data = calc_data.get_1month_hermert_data(hermert_file_name)
data = pd.concat([latlon_ex["Name"], data], axis=1)
#data = data.drop(["Lat", "Lon", "Label", "idx1", "idx2"], axis=1)
rank_np = np.zeros(145**2)
rank_np[data[data.Name == "north_polar"].index] = 1
data["rank_np"] = rank_np
rank_R2 = np.ones(145**2)
rank_R2[data[data.R2 <= (1 / 3)**2].index] = 0
rank_R2[data[data.R2 > (2 / 3)**2].index] = 2
data["rank_R2"] = rank_R2
data = data.dropna()
#https://code.i-harness.com/ja/q/1c29878
print(data.isnull().sum().sum())
data["yymm"] = [pd.to_datetime(yymm, format="%Y%m")] * len(data)
data_m = pd.concat([data_m, data])
#月ごとに全てのエリアの平均などを取得
data_m_1g = data_m.groupby("yymm")[[
"A", "theta", "epsilon2"
]].describe().sort_index(level='yymm')
#月ごとにrank_npで分類したものを取得
data_m_2g = data_m.groupby(["rank_np", "yymm"
])[["A", "theta", "epsilon2"
]].describe().sort_index(level='yymm')
#月ごとにrank_npとrank_R2で分類したものを取得
data_m_3g = data_m.groupby(["rank_np", "rank_R2", "yymm"
])[["A", "theta", "epsilon2"
]].describe().sort_index(level='yymm')
if m_plot == True:
#1gのプロット
save_name_1g = dirs_1g + "1g_" + y + ".png"
plot_param_1g(data_m_1g, save_name_1g)
#2gのプロット
save_name_2g = dirs_2g + "2g_" + y + ".png"
plot_param_2g(data_m_2g, save_name_2g)
#3gのプロット
plot_param_list = ["A", "theta", "epsilon2"]
for item in plot_param_list:
save_name_3g = dirs_3g + item + "/" + "3g_" + y + ".png"
plot_param_3g(data_m_3g, item, save_name_3g)
"""
sns.tsplot(data=data_m_A, ci="sd")
plt.plot(np.nanmean(data_m_A, axis=0))
data_m_theta = data_m[:,:,1].T
sns.tsplot(data=data_m_theta, ci="sd")
data_m_e2 = data_m[:,:,2].T
sns.tsplot(data=data_m_e2, ci="sd")
"""
tmp_1g = {
"1g_A": data_m_1g["A"].loc[:, ["mean", "std", "50%"]],
"1g_theta": data_m_1g["theta"].loc[:, ["mean", "std", "50%"]],
"1g_e2": data_m_1g["epsilon2"].loc[:, ["mean", "std", "50%"]]
}
data_1g_dic[y] = tmp_1g
tmp_2g = {
"2g_A_polar":
data_m_2g.loc[(1), ("A", ["mean", "std", "50%"])],
"2g_A_coastal":
data_m_2g.loc[(0), ("A", ["mean", "std", "50%"])],
"2g_theta_polar":
data_m_2g.loc[(1), ("theta", ["mean", "std", "50%"])],
"2g_theta_coastal":
data_m_2g.loc[(0), ("theta", ["mean", "std", "50%"])],
"2g_e2_polar":
data_m_2g.loc[(1), ("epsilon2", ["mean", "std", "50%"])],
"2g_e2_coastal":
data_m_2g.loc[(0), ("epsilon2", ["mean", "std", "50%"])]
}
data_2g_dic[y] = tmp_2g
tmp_3g = {
"3g_A_polar":
data_m_3g.loc[(1), ("A", ["mean", "std", "50%"])],
"3g_A_coastal":
data_m_3g.loc[(0), ("A", ["mean", "std", "50%"])],
"3g_theta_polar":
data_m_3g.loc[(1), ("theta", ["mean", "std", "50%"])],
"3g_theta_coastal":
data_m_3g.loc[(0), ("theta", ["mean", "std", "50%"])],
"3g_e2_polar":
data_m_3g.loc[(1), ("epsilon2", ["mean", "std", "50%"])],
"3g_e2_coastal":
data_m_3g.loc[(0), ("epsilon2", ["mean", "std", "50%"])]
}
data_3g_dic[y] = tmp_3g
if y_plot == True:
#1g
save_name_1g_year = dirs_year + "1g.png"
plot_param_1g_through_years(data_1g_dic, save_name_1g_year)
#2g
save_name_2g_year = dirs_year + "2g.png"
plot_param_2g_through_years(data_2g_dic, save_name_2g_year)
return data_1g_dic, data_2g_dic, data_3g_dic
