fix: new function > add_max_intensities_to_events

Author: MarkusPic

idf_analysis/event_series_analysis.py

@@ -90,6 +90,7 @@ def partial_series(rolling_sum_values, measurement_period):
 
 
 def _lin_regress2(x, y):
+    # based on math - custom function
     sample_size = len(x)
     x_mean = np.mean(x)
     y_mean = np.mean(y)
@@ -102,6 +103,7 @@ def _lin_regress2(x, y):
 
 
 def _lin_regress(x, y):
+    # using scipy
     res = linregress(x, y)
     return {PARAM.U: res.intercept, PARAM.W: res.slope}
 
@@ -128,42 +130,27 @@ def _improve_factor(interval):
                      list(improve_factor.values()))
 
 
-def calculate_u_w(file_input, duration_steps, series_kind):
+def iter_event_series(file_input, duration_steps):
     """
     Statistical analysis for each duration step.
 
     acc. to DWA-A 531 chap. 5.1
 
     Save the parameters of the distribution function as interim results.
 
-    acc. to DWA-A 531 chap. 4.4: use the annual series only for measurement periods over 20 years
-
-
     Args:
         file_input (pandas.Series): precipitation data
         duration_steps (list[int] | numpy.ndarray): in minutes
-        series_kind (str): 'annual' or 'partial'
 
-    Returns:
-        dict[int, dict]: with key=durations and values=dict(u, w)
+    Yields:
+        series: max rainfall intensity per event for each duration step
     """
     ts = file_input.copy()
-    # -------------------------------
-    # measuring time in years
-    measurement_start, measurement_end = ts.index[[0, -1]]
-    measurement_period = (measurement_end - measurement_start) / year_delta(years=1)
-    if round(measurement_period, 1) < 10:
-        warnings.warn("The measurement period is too short. The results may be inaccurate! "
-                      "It is recommended to use at least ten years. "
-                      "(-> Currently {}a used)".format(measurement_period))
 
     # -------------------------------
     base_frequency = guess_freq(ts.index)  # DateOffset/Timedelta
 
     # ------------------------------------------------------------------------------------------------------------------
-    interim_results = {}
-
-    # -------------------------------
     # acc. to DWA-A 531 chap. 4.2:
     # The values must be independent of each other for the statistical evaluations.
     # estimated four hours acc. (Schilling, 1984)
@@ -202,11 +189,47 @@ def calculate_u_w(file_input, duration_steps, series_kind):
 
         roll_sum = ts.rolling(duration).sum()
 
+        year_index = events[COL.START].dt.year.values
+
         # events[COL.rolling_sum_valuesAX_OVERLAPPING_SUM] = agg_events(events, roll_sum, 'max') * improve
-        rolling_sum_values = agg_events(events, roll_sum, 'max') * improve
+        yield agg_events(events, roll_sum, 'max') * improve, duration_integer, year_index
+
+
+def calculate_u_w(file_input, duration_steps, series_kind):
+    """
+    Statistical analysis for each duration step.
+
+    acc. to DWA-A 531 chap. 5.1
+
+    Save the parameters of the distribution function as interim results.
+
+    acc. to DWA-A 531 chap. 4.4: use the annual series only for measurement periods over 20 years
+
+
+    Args:
+        file_input (pandas.Series): precipitation data
+        duration_steps (list[int] | numpy.ndarray): in minutes
+        series_kind (str): 'annual' or 'partial'
+
+    Returns:
+        dict[int, dict]: with key=durations and values=dict(u, w)
+    """
+    # -------------------------------
+    # measuring time in years
+    measurement_start, measurement_end = file_input.index[[0, -1]]
+    measurement_period = (measurement_end - measurement_start) / year_delta(years=1)
+    if round(measurement_period, 1) < 10:
+        warnings.warn("The measurement period is too short. The results may be inaccurate! "
+                      "It is recommended to use at least ten years. "
+                      "(-> Currently {}a used)".format(measurement_period))
+
+    # -------------------------------
+    interim_results = {}
+
+    for rolling_sum_values, duration_integer, year_index in iter_event_series(file_input, duration_steps):
 
         if series_kind == SERIES.ANNUAL:
-            interim_results[duration_integer] = annual_series(rolling_sum_values, events[COL.START].dt.year.values)
+            interim_results[duration_integer] = annual_series(rolling_sum_values, year_index)
         elif series_kind == SERIES.PARTIAL:
             interim_results[duration_integer] = partial_series(rolling_sum_values, measurement_period)
         else:

idf_analysis/idf_backend.py

@@ -12,7 +12,7 @@
 from pathlib import Path
 
 from .definitions import *
-from .event_series_analysis import calculate_u_w
+from .event_series_analysis import calculate_u_w, iter_event_series
 from .in_out import write_yaml, read_yaml
 from .parameter_formulation_class import FORMULATION_REGISTER, _Formulation, register_formulations_to_yaml
 
@@ -39,6 +39,13 @@ def calc_from_series(self, series):
             self.parameters_series[p] = np.array([d[p] for d in u_w.values()])
         self._calc_params()
 
+    # def get_event_max_series(self, series):  # TODO
+    #     interim_results = {}
+    #
+    #     for rolling_sum_values, duration_integer, _ in iter_event_series(series, self.durations):
+    #         interim_results[duration_integer] = rolling_sum_values
+    #     return interim_results
+
     def reverse_engineering(self, idf_table, linear_interpolation=False):
         durations = idf_table.index.values
         u = idf_table[1].values

idf_analysis/idf_class.py

@@ -326,7 +326,7 @@ def result_table(self, durations=None, return_periods=None, add_names=False):
 
     ####################################################################################################################
     def result_figure(self, min_duration=5.0, max_duration=720.0, logx=False, return_periods=None, color=True, ax=None,
-                      linestyle=None, add_interim=False):
+                      linestyle=None, add_interim=False, alpha=1):
         """
         Create a plot with the idf-curves with duration on the x-axis and rainfall depth on the y-axis.
 
@@ -340,6 +340,7 @@ def result_figure(self, min_duration=5.0, max_duration=720.0, logx=False, return
             ax (plt.Axes): Axes to plot on. Default = create new one.
             linestyle (str): Line-style for the plotted lines.
             add_interim (bool): Add interim results from the series analysis as scatter points.
+            alpha (float): Alpha of the idf-lines in the plot.
 
         Returns:
             (plt.Figure, plt.Axes): figure and axes of the plot.
@@ -352,7 +353,7 @@ def result_figure(self, min_duration=5.0, max_duration=720.0, logx=False, return
         table = self.result_table(durations=duration_steps, return_periods=return_periods)
         if color:
             table.columns.name = 'T$\\mathsf{_N}$ in a'
-        ax = table.plot(color=(None if color else 'black'), logx=logx, legend=color, ax=ax, ls=linestyle)
+        ax = table.plot(color=(None if color else 'black'), logx=logx, legend=color, ax=ax, ls=linestyle, alpha=alpha)
 
         for _, return_time in enumerate(return_periods):
             if add_interim:
@@ -633,6 +634,21 @@ def add_max_return_periods_to_events(self, events):
 
             events[COL.MAX_PERIOD_DURATION] = return_periods_frame.loc[datetime_max].idxmax(axis=1, skipna=True).values
 
+    def add_max_intensities_to_events(self, events):
+        """
+        Add the maximum intensities for all duration steps to the events table.
+
+        Args:
+            events (pandas.DataFrame): events table
+
+        Returns:
+            pandas.DataFrame: events table including the columns with the maximum intensities
+        """
+        sum_frame = self.rainfall_sum_frame
+        for duration in self.duration_steps:
+            events[f'max_sum_{duration:0.0f}'] = agg_events(events, sum_frame[duration], 'max').round(2)
+        return events
+
     ####################################################################################################################
     def event_report(self, filename, min_event_rain_sum=25, min_return_period=0.5, durations=None):
         """