From c8734e359aea9d40e13b68494500a28f73bcc96f Mon Sep 17 00:00:00 2001
From: Le Roux Erwan <erwan.le-roux@irstea.fr>
Date: Wed, 25 Mar 2020 19:52:42 +0100
Subject: [PATCH] [contrasting project] modify dictionary related to wp to make
them depend
---
.../scm_models_data/abstract_study.py | 32 ++++++++++++++-----
.../main_distribution_wps.py | 16 +++++-----
.../test_edf_data/test_weather_types.py | 13 +++++---
3 files changed, 41 insertions(+), 20 deletions(-)
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 68873376..17bbf6f7 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
@@ -105,24 +105,39 @@ 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]))
+ def df_for_top_annual_maxima(self, nb_top=None, massif_names=None):
+ # Replace default arguments
+ if nb_top is None:
+ nb_top = self.nb_years
+ if massif_names is None:
+ massif_names = self.study_massif_names
+ # Load percentages of massifs
+ wps = pd.Series(np.concatenate([self.massif_name_to_df_ordered_by_maxima[massif_name]['WP'].values[:nb_top]
+ for massif_name in massif_names]))
s_normalized = wps.value_counts(normalize=True) * 100
s_normalized = s_normalized.round()
s_not_normalized = wps.value_counts()
+ # todo: do that, complete the last columns with the mean maxima
+ # Add a column that indicate the mean maxima associated to each weather pattern
+ # f = {}
+ # for wp in s_normalized.index:
+ # print(wp)
+ # for massif_id
+
+
+ # Concatenate all the results in one dataframe
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):
+ def massif_name_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 = {}
+ massif_name_to_df_ordered_by_maxima = {}
for massif_id, s_annual_maxima in df_annual_maxima.iterrows():
+ massif_name = self.study_massif_names[massif_id]
s_annual_maxima.sort_values(inplace=True, ascending=False)
d = {
'Year': s_annual_maxima.index,
@@ -131,8 +146,9 @@ class AbstractStudy(object):
}
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
+ massif_name_to_df_ordered_by_maxima[massif_name] = df
+ assert set(self.study_massif_names) == set(massif_name_to_df_ordered_by_maxima.keys())
+ return massif_name_to_df_ordered_by_maxima
@cached_property
def year_to_wp_for_annual_maxima(self):
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 8191e44a..5018c1b4 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
@@ -9,10 +9,10 @@ from extreme_data.meteo_france_data.scm_models_data.safran.safran import SafranP
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]
+ 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.wps_for_top_annual_maxima(nb_top=nb_top, massif_ids=massifs_ids), '\n')
+ print(study.df_for_top_annual_maxima(nb_top=nb_top, massif_names=massif_names), '\n')
"""
@@ -53,11 +53,11 @@ def main_temporal_distribution_wps(study_class, year_min=1954, year_max=2008):
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]
+ massif_names = AbstractExtendedStudy.region_name_to_massif_names[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')
+ for nb_top in [study_before.nb_years, 10, 5, 1][-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')
"""
There is no real stationarity in the percentage of the kind of storms that are causing extreme.
@@ -71,5 +71,5 @@ even for the local region it is the same.
if __name__ == '__main__':
study_class = [CrocusSnowLoad1Day, SafranPrecipitation1Day][-1]
- # main_spatial_distribution_wps(study_class)
- main_temporal_distribution_wps(study_class)
+ main_spatial_distribution_wps(study_class)
+ # main_temporal_distribution_wps(study_class)
diff --git a/test/test_extreme_data/test_edf_data/test_weather_types.py b/test/test_extreme_data/test_edf_data/test_weather_types.py
index 78af1c08..740e0d33 100644
--- a/test/test_extreme_data/test_edf_data/test_weather_types.py
+++ b/test/test_extreme_data/test_edf_data/test_weather_types.py
@@ -63,12 +63,17 @@ class TestWeatherTypes(unittest.TestCase):
def test_weather_patterns_maxima(self):
study = SafranPrecipitation1Day(altitude=900, year_min=1954, year_max=2008)
- s = pd.Series(np.concatenate([v for v in study.year_to_wp_for_annual_maxima.values()]))
- storms_ranking = s.value_counts()
+ storms_ranking = study.df_for_top_annual_maxima()
self.assertEqual(storms_ranking.index[0], STEADY_OCEANIC)
self.assertEqual(storms_ranking.index[-1], ANTICYCLONIC)
- self.assertEqual(storms_ranking.values[0], 376)
- self.assertEqual(storms_ranking.values[-1], 9)
+ self.assertEqual(storms_ranking.values[0, 1], 376)
+ self.assertEqual(storms_ranking.values[-1, 1], 9)
+
+ def test_massif_id_to_weather_type_df(self):
+ study = SafranPrecipitation1Day(altitude=900, year_min=2004, year_max=2008)
+ df = study.massif_name_to_df_ordered_by_maxima['Chablais']
+ self.assertAlmostEqual(df.loc[2007].values[0], 57.384655)
+ self.assertEqual(df.loc[2007].values[1], 'Steady Oceanic')
if __name__ == '__main__':
--
GitLab