diff --git a/extreme_data/meteo_france_data/scm_models_data/abstract_study.py b/extreme_data/meteo_france_data/scm_models_data/abstract_study.py
index 5cd271a392cd43447353698232277b296a296b73..ffd7d5427e0d3003ef9fded6d1b81d264833a87b 100644
--- a/extreme_data/meteo_france_data/scm_models_data/abstract_study.py
+++ b/extreme_data/meteo_france_data/scm_models_data/abstract_study.py
@@ -123,7 +123,7 @@ class AbstractStudy(object):
year_to_wps[year] = self.df_weather_types.loc[days].iloc[:, 0].values
return year_to_wps
- def df_for_top_annual_maxima(self, nb_top=None, massif_names=None):
+ def df_for_top_annual_maxima(self, nb_top=None, massif_names=None, limit_for_the_percentage=None):
# Replace default arguments
if nb_top is None:
nb_top = self.nb_years
@@ -151,7 +151,10 @@ class AbstractStudy(object):
df.drop(columns=drop_columns, inplace=True)
df.rename(columns={'50%': 'median'}, inplace=True)
df = df.astype(int)
- df.index.name = 'Number Top={}'.format(nb_top)
+ df.index.name = 'Top {} maxima ({} -{})'.format(nb_top, self.year_min, self.year_max)
+ if limit_for_the_percentage is not None:
+ ind = df['%'] > limit_for_the_percentage
+ df = df.loc[ind]
return df
@cached_property
@@ -214,33 +217,6 @@ class AbstractStudy(object):
year_to_annual_maxima[year] = time_serie.max(axis=0)
return year_to_annual_maxima
- @cached_property
- def year_to_winter_annual_maxima(self) -> OrderedDict:
- # Map each year to an array of size nb_massif
- year_to_annual_maxima = OrderedDict()
- for year, time_serie in self._year_to_max_daily_time_serie.items():
- year_to_annual_maxima[year] = time_serie[91:-61].max(axis=0)
- return year_to_annual_maxima
-
- @property
- def observations_winter_annual_maxima(self) -> AnnualMaxima:
- return AnnualMaxima(
- df_maxima_gev=pd.DataFrame(self.year_to_winter_annual_maxima, index=self.study_massif_names))
-
- @cached_property
- def year_to_summer_annual_maxima(self) -> OrderedDict:
- # Map each year to an array of size nb_massif
- year_to_annual_maxima = OrderedDict()
- for year, time_serie in self._year_to_max_daily_time_serie.items():
- annual_maxima = np.concatenate((time_serie[:91], time_serie[-61:]), axis=0).max(axis=0)
- year_to_annual_maxima[year] = annual_maxima
- return year_to_annual_maxima
-
- @property
- def observations_summer_annual_maxima(self) -> AnnualMaxima:
- return AnnualMaxima(
- df_maxima_gev=pd.DataFrame(self.year_to_summer_annual_maxima, index=self.study_massif_names))
-
@cached_property
def year_to_annual_maxima_index(self) -> OrderedDict:
# Map each year to an array of size nb_massif
diff --git a/extreme_trend/visualizers/utils.py b/extreme_trend/visualizers/utils.py
index 43687d4594689d5acb935060ad1e85e65652e9d6..9e2f857ab2c58b142c3424803f97dabd0a8c707b 100644
--- a/extreme_trend/visualizers/utils.py
+++ b/extreme_trend/visualizers/utils.py
@@ -1,5 +1,6 @@
from collections import OrderedDict
+from extreme_data.meteo_france_data.scm_models_data.utils import SeasonForTheMaxima
from extreme_fit.model.margin_model.linear_margin_model.abstract_temporal_linear_margin_model import \
TemporalMarginFitMethod
from extreme_trend.visualizers.study_visualizer_for_non_stationary_trends import \
@@ -10,11 +11,13 @@ def load_altitude_to_visualizer(altitudes, massif_names, model_subsets_for_uncer
uncertainty_methods,
study_visualizer_class=StudyVisualizerForNonStationaryTrends,
save_to_file=True,
- multiprocessing=True):
+ multiprocessing=True,
+ season=SeasonForTheMaxima.annual):
+ print("Season={}".format(season))
fit_method = TemporalMarginFitMethod.extremes_fevd_mle
altitude_to_visualizer = OrderedDict()
for altitude in altitudes:
- study = study_class(altitude=altitude, multiprocessing=multiprocessing)
+ study = study_class(altitude=altitude, multiprocessing=multiprocessing, season=season)
study_visualizer = study_visualizer_class(study=study, multiprocessing=multiprocessing,
save_to_file=save_to_file, uncertainty_massif_names=massif_names,
uncertainty_methods=uncertainty_methods,
diff --git a/projects/contrasting_trends_in_snow_loads/main_result.py b/projects/contrasting_trends_in_snow_loads/main_result.py
index 02c5567be8d6f0cb747e54eb9055fb66971aec0c..75b13547e249ae384664642561f0c3e5037a96d5 100644
--- a/projects/contrasting_trends_in_snow_loads/main_result.py
+++ b/projects/contrasting_trends_in_snow_loads/main_result.py
@@ -2,6 +2,7 @@ from multiprocessing.pool import Pool
import matplotlib as mpl
+from extreme_data.meteo_france_data.scm_models_data.utils import SeasonForTheMaxima
from extreme_trend.visualizers.utils import load_altitude_to_visualizer
mpl.use('Agg')
@@ -41,7 +42,8 @@ def intermediate_result(altitudes, massif_names=None,
"""
# Load altitude to visualizer
altitude_to_visualizer = load_altitude_to_visualizer(altitudes, massif_names, model_subsets_for_uncertainty,
- study_class, uncertainty_methods, save_to_file=save_to_file)
+ study_class, uncertainty_methods, save_to_file=save_to_file,
+ season=SeasonForTheMaxima.winter_extended)
# Load variable object efficiently
for v in altitude_to_visualizer.values():
_ = v.study.year_to_variable_object
@@ -65,7 +67,7 @@ def major_result():
model_subsets_for_uncertainty = None
# altitudes = paper_altitudes
# altitudes = paper_altitudes
- altitudes = [900, 1200, 1500, 1800, 2100, 2400, 2700, 3000][:8]
+ altitudes = [900, 1200, 1500, 1800, 2100, 2400, 2700, 3000][:]
snow_load_classes = [CrocusSnowLoad1Day, CrocusSnowLoad3Days, CrocusSnowLoad5Days, CrocusSnowLoad7Days][:]
precipitation_classes = [SafranPrecipitation1Day, SafranPrecipitation3Days, SafranPrecipitation5Days,
SafranPrecipitation7Days][:]
@@ -73,7 +75,7 @@ def major_result():
rainfall_classes = [SafranRainfall1Day, SafranRainfall3Days, SafranRainfall5Days, SafranRainfall7Days]
study_classes = precipitation_classes + snow_load_classes
# study_classes = snowfall_classes + rainfall_classes
- for study_class in study_classes[:1]:
+ for study_class in study_classes[:]:
intermediate_result(altitudes, massif_names, model_subsets_for_uncertainty,
uncertainty_methods, study_class, multiprocessing=True)
diff --git a/projects/contrasting_trends_in_snow_loads/main_spatial_relative_change_in_maxima_at_fixed_altitude.py b/projects/contrasting_trends_in_snow_loads/main_spatial_relative_change_in_maxima_at_fixed_altitude.py
index 5b5a6ea32426c8e07d47306e0cd3106751ebd356..91260c0b93074d2fc48be64ba2cd5db611d4e2e0 100644
--- a/projects/contrasting_trends_in_snow_loads/main_spatial_relative_change_in_maxima_at_fixed_altitude.py
+++ b/projects/contrasting_trends_in_snow_loads/main_spatial_relative_change_in_maxima_at_fixed_altitude.py
@@ -6,13 +6,6 @@ from extreme_data.meteo_france_data.scm_models_data.visualization.study_visualiz
import matplotlib.pyplot as plt
-def test():
- study = CrocusSnowLoad3Days(altitude=1200)
- study_visualizer = StudyVisualizer(study)
- study_visualizer.visualize_max_graphs_poster('Queyras', altitude='noope', snow_abbreviation="ok", color='red')
- plt.show()
-
-
def density_wrt_altitude():
save_to_file = True
study_class = [SafranPrecipitation, SafranRainfall, SafranSnowfall, CrocusSnowLoad3Days, CrocusSnowLoadTotal][-2]
diff --git a/projects/contrasting_trends_in_snow_loads/weather_types_analysis/main_distribution_wps.py b/projects/contrasting_trends_in_snow_loads/weather_types_analysis/main_distribution_wps.py
index 9372f13b45771fbe09760ee51d802cb4931a7d5c..52946ba34c073266ab7da920ea13b8851f492420 100644
--- a/projects/contrasting_trends_in_snow_loads/weather_types_analysis/main_distribution_wps.py
+++ b/projects/contrasting_trends_in_snow_loads/weather_types_analysis/main_distribution_wps.py
@@ -3,76 +3,109 @@ import numpy as np
from extreme_data.meteo_france_data.scm_models_data.abstract_extended_study import AbstractExtendedStudy
from extreme_data.meteo_france_data.scm_models_data.crocus.crocus import CrocusSnowLoad3Days, CrocusSnowLoad1Day
-from extreme_data.meteo_france_data.scm_models_data.safran.safran import SafranPrecipitation1Day
+from extreme_data.meteo_france_data.scm_models_data.safran.safran import SafranPrecipitation1Day, \
+ SafranPrecipitation3Days
+from extreme_data.meteo_france_data.scm_models_data.utils import SeasonForTheMaxima
-def main_spatial_distribution_wps(study_class, year_min=1954, year_max=2008):
- study = study_class(altitude=1800, year_min=year_min, year_max=year_max)
+def main_spatial_distribution_wps(study_class, year_min=1954, year_max=2008, limit_for_the_percentage=None):
+ study = study_class(altitude=1800, year_min=year_min, year_max=year_max, season=SeasonForTheMaxima.winter_extended)
for region_name in AbstractExtendedStudy.region_names:
massif_names = AbstractExtendedStudy.region_name_to_massif_names[region_name]
print('\n \n', region_name, '\n')
- for nb_top in [study.nb_years, 5, 1][1:2]:
- print(study.df_for_top_annual_maxima(nb_top=nb_top, massif_names=massif_names), '\n')
+ for nb_top in [study.nb_years, 10, 1][1:2]:
+ print(study.df_for_top_annual_maxima(nb_top=nb_top, massif_names=massif_names, limit_for_the_percentage=limit_for_the_percentage), '\n')
"""
-Some analysis:
+ Northern Alps
+ % count mean std min median max
+Steady Oceanic 92 65 159 35 121 146 282
+
+ Central Alps
+ % count mean std min median max
+Steady Oceanic 72 51 136 31 97 130 235
+South Circulation 15 11 128 18 105 126 161
-At altitude 1800m,
-for the precipitation in 1 day
-between year_min=1954, year_max=2008
+ Southern Alps
+ % count mean min median max
+South Circulation 61 37 147 79 147 235
+Steady Oceanic 13 8 121 93 114 187
+
+ Extreme South Alps
+ % count mean min median max
+South Circulation 80 24 174 101 166 306
-If we look at the 5 biggest maxima for each massif,
-and look to which Weather pattern they correspond
-then we do some percentage for each climatic region
+
+Process finished with exit code 0
- Northern Alps
- % count mean std min median max
-Steady Oceanic 77 27 105 16 80 104 150
-Atlantic Wave 11 4 111 15 95 111 129
- Central Alps
- % count mean min median max
-Steady Oceanic 57 20 90 70 86 119
-South Circulation 17 6 85 64 88 104
-East Return 14 5 88 74 93 100
- Southern Alps
- % count mean min median max
-South Circulation 43 13 99 72 106 122
-Central Depression 36 11 98 68 97 136
-East Return 16 5 90 70 83 121
- Extreme South Alps
-Central Depression 53.0 8
-South Circulation 47.0 7
"""
-def main_temporal_distribution_wps(study_class, year_min=1954, year_max=2008):
+def main_temporal_distribution_wps(study_class, year_min=1954, year_max=2008, limit_for_the_percentage=None):
altitude = 1800
- study_before = study_class(altitude=altitude, year_min=year_min, year_max=1981)
- study_after = study_class(altitude=altitude, year_min=1981, year_max=2008)
+ study_before = study_class(altitude=altitude, year_min=year_min, year_max=1981, season=SeasonForTheMaxima.winter_extended)
+ study_after = study_class(altitude=altitude, year_min=1981, year_max=year_max, season=SeasonForTheMaxima.winter_extended)
+ # todo: same min and max year ?
for region_name in AbstractExtendedStudy.region_names:
massif_names = AbstractExtendedStudy.region_name_to_massif_names[region_name]
- print('\n \n', region_name, '\n')
+ print('\n \n', '{} ({} massifs)'.format(region_name, len(massif_names)), '\n')
for nb_top in [study_before.nb_years, 10][1:]:
- print(study_before.df_for_top_annual_maxima(nb_top=nb_top, massif_names=massif_names), '\n')
- print(study_after.df_for_top_annual_maxima(nb_top=nb_top, massif_names=massif_names), '\n')
+ print(study_before.df_for_top_annual_maxima(nb_top=nb_top, massif_names=massif_names, limit_for_the_percentage=limit_for_the_percentage), '\n')
+ print(study_after.df_for_top_annual_maxima(nb_top=nb_top, massif_names=massif_names, limit_for_the_percentage=limit_for_the_percentage), '\n')
"""
-There is no real stationarity in the percentage of the kind of storms that are causing extreme.
+ Northern Alps (7 massifs)
+ % count mean min median max
+Top 10 maxima (1954 -1981)
+Steady Oceanic 94 66 130 102 127 202
+
+Top 10 maxima (1981 -2008)
+Steady Oceanic 82 58 151 104 134 282
+
+ Central Alps (7 massifs)
+ % count mean min median max
+Top 10 maxima (1954 -1981)
+Steady Oceanic 71 50 110 76 107 190
+
+Top 10 maxima (1981 -2008)
+Steady Oceanic 74 52 125 87 115 235
+South Circulation 14 10 123 100 120 161
+
+ Southern Alps (6 massifs)
+ % count mean min median max
+Top 10 maxima (1954 -1981)
+South Circulation 43 26 113 67 112 197
+Steady Oceanic 16 10 95 82 93 122
+Southwest Circulation 15 9 102 68 95 140
+
+Top 10 maxima (1981 -2008)
+South Circulation 63 38 134 73 127 235
+Steady Oceanic 21 13 110 80 105 187
+
+
+ Extreme South Alps (3 massifs)
+ % count mean min median max
+Top 10 maxima (1954 -1981)
+South Circulation 63 19 136 84 139 194
+Southwest Circulation 13 4 122 95 123 146
+
+Top 10 maxima (1981 -2008)
+South Circulation 76 23 165 78 159 306
+
+
-the reparittion of storm before and after (no matter the nb_top consider 10, or all) are the same.
-even for the local region it is the same.
--> so what is really changing is probably the strength associated to each kind of storm.
"""
if __name__ == '__main__':
- study_class = [CrocusSnowLoad1Day, SafranPrecipitation1Day][-1]
- main_spatial_distribution_wps(study_class)
- # main_temporal_distribution_wps(study_class)
+ limit_percentage = 10
+ study_class = [CrocusSnowLoad1Day, SafranPrecipitation1Day, SafranPrecipitation3Days][-1]
+ # main_spatial_distribution_wps(study_class, limit_for_the_percentage=limit_percentage)
+ main_temporal_distribution_wps(study_class, limit_for_the_percentage=limit_percentage)