markuspic / intensity_duration_frequency_analysis Goto Github PK

Which function did the author use to make the IDF curve, the GEV distribution?

Dear Markus,

a relatively short time series behaved as expected in testing purposes. However, I have noticed that w takes on negative sign and hinders further computations depending on log10 when elongating the time series. Do you have any hints on how this might (perhaps mathematically) occur or what this could imply?

I find your package great and unique, would really love to experiment with it without having to implement it myself at the moment, and thus hope for your thoughts.

Best
Philipp

Dear all,

thanks for this lovely opensource. During usage I stumbled over something which I would call a minor issue. One may reconfirm as below.

import io
from pandas import read_csv
from ehyd_tools.synthetic_rainseries import RainModeller

rain = """duration,1,2,5,10,20
5,12.9,14.7,17.2,19.0,20.9
10,18.9,21.9,25.6,28.8,31.9
15,24.6,28.2,34.4,37.9,43.0
30,34.4,40.9,49.5,55.6,62.3
60,45.4,55.0,69.8,79.2,88.8
90,47.4,56.8,68.9,78.6,88.6
120,50.0,59.3,71.5,81.5,91.3
240,55.9,66.6,81.1,91.5,102.2
720,61.9,72.9,87.9,100.5,111.9
1440,66.1,78.1,92.3,104.8,115.9
"""

kostra = read_csv(io.StringIO(rain), index_col=0)
model_rain = RainModeller()
model_rain.idf_table = kostra
model_rain.idf_table.columns = model_rain.idf_table.columns.astype(int) 
ts = model_rain.euler.get_time_series(return_period=20, duration=1440, interval=5, kind=2, start_time='2021-01-01 00:00')
print(ts.tail(10))

Small negative values are returned for the final timesteps. The cumulative amount of the output (including negative values) balances to 115.9, which is to be expected from the input.

2021-01-01 23:30:00    0.027778
2021-01-01 23:35:00   -0.033333
2021-01-01 23:40:00   -0.033333
2021-01-01 23:45:00   -0.033333
2021-01-01 23:50:00   -0.033333
2021-01-01 23:55:00   -0.033333
2021-01-02 00:00:00   -0.033333
dtype: float64

Repeating for different duration I observe that negative values start from 90min and above.

Kind regards

Sebastian