def test_model_gr4j_run(dataset_l0123001):
air_gr_parameters = [265.072, 1.040]
air_gr_rmse = 17.804161
# One month aggregation using period as index :
precipitation = dataset_l0123001['precipitation'].resample(
'M', kind='period').sum()
temperature = dataset_l0123001['temperature'].resample(
'M', kind='period').mean()
evapotranspiration = dataset_l0123001['evapotranspiration'].resample(
'm', kind='period').sum()
flow_mm = dataset_l0123001['flow_mm'].resample('M', kind='period').sum()
df = pd.concat([precipitation, temperature, evapotranspiration, flow_mm],
axis=1)
df['date'] = df.index
input_handler = InputDataHandler(ModelGr2m, df)
warm_up_start_date = datetime.datetime(1989, 1, 1, 0, 0)
start_date = datetime.datetime(1990, 1, 1, 0, 0)
end_date = datetime.datetime(1999, 12, 1, 0, 0)
sub_input = input_handler.get_sub_period(warm_up_start_date, end_date)
model = ModelGr2m(sub_input, air_gr_parameters)
outputs = model.run()
filtered_input = model.input_data[model.input_data.index >= start_date]
filtered_output = outputs[outputs.index >= start_date]
rmse = sqrt(
mean(
(filtered_output['Qsim'] - filtered_input['flow_mm'].values)**2.0))
assert round(rmse, 6) == round(air_gr_rmse, 6)
def test_model_gr1a_run(dataset_l0123001):
air_gr_parameters = [0.840]
air_gr_rmse = 64.70577504958929 # Slightly different from airGR rmse as R method for resampling return NA for 1996 observed flow...
# One year aggregation :
precipitation = dataset_l0123001['precipitation'].resample('A').sum()
temperature = dataset_l0123001['temperature'].resample('A').mean()
evapotranspiration = dataset_l0123001['evapotranspiration'].resample('A').sum()
flow_mm = dataset_l0123001['flow_mm'].resample('A').sum()
df = pd.concat([precipitation, temperature, evapotranspiration, flow_mm], axis=1)
df['date'] = df.index
input_handler = InputDataHandler(ModelGr1a, df)
warm_up_start_date = datetime.datetime(1989, 1, 1, 0, 0)
start_date = datetime.datetime(1990, 1, 1, 0, 0)
end_date = datetime.datetime(1999, 12, 31, 0, 0)
sub_input = input_handler.get_sub_period(warm_up_start_date, end_date)
model = ModelGr1a(sub_input, air_gr_parameters)
outputs = model.run()
filtered_input = model.input_data[model.input_data.index >= start_date]
filtered_output = outputs[outputs.index >= start_date]
rmse = sqrt(mean((filtered_output['Qsim'] - filtered_input['flow_mm'].values) ** 2.0))
assert round(rmse, 6) == round(air_gr_rmse, 6)
def test_input_data_fail(dataset_l0123001):
with pytest.raises(ValueError):
_ = InputDataHandler(ModelGr1a, dataset_l0123001)
with pytest.raises(ValueError):
_ = InputDataHandler(ModelGr2m, dataset_l0123001)
with pytest.raises(ValueError):
_ = InputDataHandler(ModelGr4h, dataset_l0123001)
def __init__(self, data, start_date, end_date):
self.start = start_date
self.end = end_date
self.data = data
self.model_inputs = InputDataHandler(ModelGr4j, self.data)
self.bound = [[0.0, 200.0], [0.0, 1.0], [0.0, 100.0],
[0.0, 10.0]] # Physical parameter boundaries
def test_input_yearly_data_period_index(dataset_l0123001):
# One year aggregation with datetime index :
precipitation = dataset_l0123001['precipitation'].resample('A', kind='period').sum()
temperature = dataset_l0123001['temperature'].resample('A', kind='period').mean()
evapotranspiration = dataset_l0123001['evapotranspiration'].resample('A', kind='period').sum()
flow_mm = dataset_l0123001['flow_mm'].resample('A', kind='period').sum()
df = pd.concat([precipitation, temperature, evapotranspiration, flow_mm], axis=1)
df['date'] = df.index
_ = InputDataHandler(ModelGr1a, df)
def test_model_gr6j_run(dataset_l0123001):
air_gr_parameters = [242.257, 0.637, 53.517, 2.218, 0.424, 4.759]
air_gr_rmse = 0.785089
input_handler = InputDataHandler(ModelGr6j, dataset_l0123001)
warm_up_start_date = datetime.datetime(1989, 1, 1, 0, 0)
start_date = datetime.datetime(1990, 1, 1, 0, 0)
end_date = datetime.datetime(1999, 12, 31, 0, 0)
sub_input = input_handler.get_sub_period(warm_up_start_date, end_date)
model = ModelGr6j(sub_input, air_gr_parameters)
outputs = model.run()
filtered_input = model.input_data[model.input_data.index >= start_date]
filtered_output = outputs[outputs.index >= start_date]
rmse = sqrt(
mean(
(filtered_output['Qsim'] - filtered_input['flow_mm'].values)**2.0))
assert pytest.approx(rmse) == air_gr_rmse
def test_model_gr5j_run(dataset_l0123001):
air_gr_parameters = [245.918, 1.027, 90.017, 2.198, 0.434]
air_gr_rmse = 0.8072707
input_handler = InputDataHandler(ModelGr5j, dataset_l0123001)
warm_up_start_date = datetime.datetime(1989, 1, 1, 0, 0)
start_date = datetime.datetime(1990, 1, 1, 0, 0)
end_date = datetime.datetime(1999, 12, 31, 0, 0)
sub_input = input_handler.get_sub_period(warm_up_start_date, end_date)
model = ModelGr5j(sub_input, air_gr_parameters)
outputs = model.run()
filtered_input = model.input_data[model.input_data.index >= start_date]
filtered_output = outputs[outputs.index >= start_date]
rmse = sqrt(
mean(
(filtered_output['Qsim'] - filtered_input['flow_mm'].values)**2.0))
assert pytest.approx(rmse) == air_gr_rmse
def test_input_data_get_sub_period(dataset_l0123001):
input_data_handler = InputDataHandler(ModelGr4j, dataset_l0123001)
main_start_date = datetime.datetime(1984, 1, 1, 0, 0, 0)
main_end_date = datetime.datetime(2012, 12, 31, 0, 0, 0)
assert input_data_handler.start_date == main_start_date
assert input_data_handler.end_date == main_end_date
start_date = datetime.datetime(1989, 1, 1, 0, 0)
end_date = datetime.datetime(1999, 12, 31, 0, 0)
sub_input = input_data_handler.get_sub_period(start_date, end_date)
assert input_data_handler.start_date == main_start_date
assert input_data_handler.end_date == main_end_date
assert sub_input.start_date == start_date
assert sub_input.end_date == end_date
start_date = datetime.datetime(1970, 1, 1, 0, 0)
end_date = datetime.datetime(2222, 12, 31, 0, 0)
sub_input = input_data_handler.get_sub_period(start_date, end_date)
assert input_data_handler.start_date == main_start_date
assert input_data_handler.end_date == main_end_date
assert sub_input.start_date == main_start_date
assert sub_input.end_date == main_end_date
def test_model_gr4h_run(dataset_l0123003):
air_gr_parameters = [521.113, -2.918, 218.009, 4.124]
air_gr_rmse = 0.07847187
input_handler = InputDataHandler(ModelGr4h, dataset_l0123003)
warm_up_start_date = datetime.datetime(2004, 1, 1, 0, 0)
start_date = datetime.datetime(2004, 3, 1, 0, 0)
end_date = datetime.datetime(2008, 12, 31, 0, 0)
sub_input = input_handler.get_sub_period(warm_up_start_date, end_date)
model = ModelGr4h(sub_input, air_gr_parameters)
outputs = model.run()
filtered_input = model.input_data[model.input_data.index >= start_date]
filtered_output = outputs[outputs.index >= start_date]
rmse = sqrt(
mean(
(filtered_output['Qsim'] - filtered_input['flow_mm'].values)**2.0))
assert round(rmse,
4) == round(air_gr_rmse,
4) # slightly different after the 4th decimal...
from pathlib import Path
import pandas as pd
import datetime
from hydrogr import InputDataHandler, ModelGr6j
from hydrogr.plot_tools import plot_hydrograph
if __name__ == '__main__':
data_path = Path.cwd().parent / 'data'
df = pd.read_pickle(data_path / 'L0123001.pkl')
df.columns = [
'date', 'precipitation', 'temperature', 'evapotranspiration', 'flow',
'flow_mm'
]
df.index = df['date']
air_gr_parameters = [242.257, 0.637, 53.517, 2.218, 0.424, 4.759]
input_handler = InputDataHandler(ModelGr6j, df)
start_date = datetime.datetime(1989, 1, 1, 0, 0)
end_date = datetime.datetime(1999, 12, 31, 0, 0)
sub_input = input_handler.get_sub_period(start_date, end_date)
model = ModelGr6j(sub_input, air_gr_parameters)
model.set_initial_conditions(production_store_filling=0.3,
routing_store_filling=0.5)
outputs = model.run()
plot_hydrograph(outputs, sub_input.data['flow_mm'])
'date', 'precipitation', 'temperature', 'evapotranspiration', 'flow',
'flow_mm'
]
df.index = df['date']
# One month aggregation using period as index :
precipitation = df['precipitation'].resample('M', kind='period').sum()
temperature = df['temperature'].resample('M', kind='period').mean()
evapotranspiration = df['evapotranspiration'].resample(
'm', kind='period').sum()
flow_mm = df['flow_mm'].resample('M', kind='period').sum()
df = pd.concat([precipitation, temperature, evapotranspiration, flow_mm],
axis=1)
df['date'] = df.index
input_handler = InputDataHandler(ModelGr2m, df)
warm_up_start_date = datetime.datetime(1989, 1, 1, 0, 0)
start_date = datetime.datetime(1990, 1, 1, 0, 0)
end_date = datetime.datetime(1999, 1, 1, 0, 0)
sub_input = input_handler.get_sub_period(warm_up_start_date, end_date)
air_gr_parameters = [265.072, 1.040]
model = ModelGr2m(sub_input, air_gr_parameters)
outputs = model.run()
filtered_input = model.input_data[model.input_data.index >= start_date]
filtered_output = outputs[outputs.index >= start_date]
plot_hydrograph(filtered_output, sub_input.data['flow_mm'])
def test_input_hourly_data_datetime_index(dataset_l0123003):
_ = InputDataHandler(ModelGr4h, dataset_l0123003)
def test_input_daily_data_datetime_index(dataset_l0123001):
_ = InputDataHandler(ModelGr4j, dataset_l0123001)
import pandas as pd
import datetime
from hydrogr import InputDataHandler, ModelGr4h
from hydrogr.plot_tools import plot_hydrograph
if __name__ == '__main__':
data_path = Path.cwd().parent / 'data'
# df = pd.read_csv(data_path / 'L0123003.csv')
# df['date'] = pd.to_datetime(df['date'], format='%d/%m/%Y %H:%M')
# df.to_pickle(data_path / 'L0123003.pkl')
df = pd.read_pickle(data_path / 'L0123003.pkl')
df.columns = [
'date', 'precipitation', 'temperature', 'evapotranspiration', 'flow',
'flow_mm'
]
df.index = df['date']
my_inputs_handler = InputDataHandler(ModelGr4h, df)
filtered_input = my_inputs_handler.get_sub_period(
datetime.datetime(2004, 1, 1, 0, 0),
datetime.datetime(2008, 1, 1, 0, 0))
model = ModelGr4h(filtered_input, [521.113, -2.918, 218.009, 4.124])
model.set_initial_conditions(production_store_filling=0.9,
routing_store_filling=0.9)
outputs = model.run()
print(outputs.head())
plot_hydrograph(outputs, my_inputs_handler.data['flow_mm'])
