From e9e777bc65977fbee11c5daad8dcb8c90a97e56b Mon Sep 17 00:00:00 2001
From: Le Roux Erwan <erwan.le-roux@irstea.fr>
Date: Wed, 25 Mar 2020 19:24:43 +0100
Subject: [PATCH] [contrasting project] add main_distribution_wps.py to study
the (spatial and temporal) distribution of weather patterns that are causing
maxima in the different region of the French Alps
---
.../scm_models_data/abstract_study.py | 33 ++++++++
.../main_distribution_wps.py | 75 +++++++++++++++++++
2 files changed, 108 insertions(+)
create mode 100644 projects/contrasting_trends_in_snow_loads/weather_types_analysis/main_distribution_wps.py
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 ddafc5d3..68873376 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
@@ -77,6 +77,10 @@ class AbstractStudy(object):
""" Time """
+ @property
+ def nb_years(self):
+ return self.year_max - self.year_min + 1
+
@cached_property
def year_to_days(self) -> OrderedDict:
# Map each year to the 'days since year-08-01 06:00:00'
@@ -101,6 +105,35 @@ class AbstractStudy(object):
year_to_wps[year] = self.df_weather_types.loc[days].iloc[:, 0].values
return year_to_wps
+ def wps_for_top_annual_maxima(self, nb_top, massif_ids):
+ d = self.massif_id_to_df_ordered_by_maxima
+ wps = pd.Series(np.concatenate([d[massif_id]['WP'].values[:nb_top]
+ for massif_id in massif_ids]))
+ s_normalized = wps.value_counts(normalize=True) * 100
+ s_normalized = s_normalized.round()
+ s_not_normalized = wps.value_counts()
+ df = pd.concat([s_normalized, s_not_normalized], axis=1)
+ df.columns = ['Percentage', 'Nb massifs concerned']
+ df.index.name = 'Number Top={}'.format(nb_top)
+ return df
+
+ @cached_property
+ def massif_id_to_df_ordered_by_maxima(self):
+ df_annual_maxima = pd.DataFrame(self.year_to_annual_maxima)
+ df_wps = pd.DataFrame(self.year_to_wp_for_annual_maxima)
+ massif_id_to_df_ordered_by_maxima = {}
+ for massif_id, s_annual_maxima in df_annual_maxima.iterrows():
+ s_annual_maxima.sort_values(inplace=True, ascending=False)
+ d = {
+ 'Year': s_annual_maxima.index,
+ 'Maxima': s_annual_maxima.values,
+ 'WP': df_wps.loc[massif_id, s_annual_maxima.index],
+ }
+ df = pd.DataFrame(d)
+ df.set_index('Year', inplace=True)
+ massif_id_to_df_ordered_by_maxima[massif_id] = df
+ return massif_id_to_df_ordered_by_maxima
+
@cached_property
def year_to_wp_for_annual_maxima(self):
year_to_wp_for_annual_maxima = OrderedDict()
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
new file mode 100644
index 00000000..8191e44a
--- /dev/null
+++ b/projects/contrasting_trends_in_snow_loads/weather_types_analysis/main_distribution_wps.py
@@ -0,0 +1,75 @@
+import pandas as pd
+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
+
+
+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)
+ for region_name in AbstractExtendedStudy.region_names:
+ massifs_ids = AbstractExtendedStudy.region_name_to_massif_ids[region_name]
+ print('\n \n', region_name, '\n')
+ for nb_top in [study.nb_years, 5, 1][1:2]:
+ print(study.wps_for_top_annual_maxima(nb_top=nb_top, massif_ids=massifs_ids), '\n')
+
+
+"""
+Some analysis:
+
+At altitude 1800m,
+for the precipitation in 1 day
+between year_min=1954, year_max=2008
+
+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
+
+ Percentage Nb massifs concerned
+ Northern Alps
+Steady Oceanic 77.0 27
+Atlantic Wave 11.0 4
+
+ Central Alps
+Steady Oceanic 57.0 20
+South Circulation 17.0 6
+East Return 14.0 5
+
+ Southern Alps
+South Circulation 43.0 13
+Central Depression 37.0 11
+East Return 17.0 5
+
+ 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):
+ 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)
+ for region_name in AbstractExtendedStudy.region_names:
+ massifs_ids = AbstractExtendedStudy.region_name_to_massif_ids[region_name]
+ print('\n \n', region_name, '\n')
+ for nb_top in [study_before.nb_years, 10, 5, 1][1:2]:
+ print(study_before.wps_for_top_annual_maxima(nb_top=nb_top, massif_ids=massifs_ids), '\n')
+ print(study_after.wps_for_top_annual_maxima(nb_top=nb_top, massif_ids=massifs_ids), '\n')
+
+"""
+There is no real stationarity in the percentage of the kind of storms that are causing extreme.
+
+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)
--
GitLab