def find_hash_id_of_nearest_rainfall_station(curw_obs_pool, curw_sim_pool, lat,
lon):
obs_connection = curw_obs_pool.connection()
Sim_Ts = Timeseries(pool=curw_sim_pool)
try:
with obs_connection.cursor() as cursor0:
cursor0.callproc('getNearestWeatherStation', (lat, lon))
obs_station = cursor0.fetchone()
obs_id = obs_station['id']
obs_station_name = obs_station['name']
grid_id = 'rainfall_{}_{}_MDPA'.format(
obs_id, obs_station_name) # rainfall_100057_Naula_MDPA
return Sim_Ts.get_timeseries_id(grid_id=grid_id, method='MME')
except Exception as e:
traceback.print_exc()
def prepare_mike_rf_input(start, end, coefficients):
try:
#### process staton based hybrid timeseries ####
distinct_obs_ids = coefficients.curw_obs_id.unique()
hybrid_ts_df = pd.DataFrame()
hybrid_ts_df['time'] = pd.date_range(start=start, end=end, freq='5min')
pool = get_Pool(host=con_params.CURW_SIM_HOST,
port=con_params.CURW_SIM_PORT,
user=con_params.CURW_SIM_USERNAME,
password=con_params.CURW_SIM_PASSWORD,
db=con_params.CURW_SIM_DATABASE)
TS = Timeseries(pool)
obs_id_hash_id_mapping = get_all_obs_rain_hashids_from_curw_sim(pool)
for obs_id in distinct_obs_ids:
# taking data from curw_sim database (data prepared based on active stations for hechms)
ts = TS.get_timeseries(id_=obs_id_hash_id_mapping.get(str(obs_id)),
start_date=start,
end_date=end)
ts.insert(0, ['time', obs_id])
ts_df = list_of_lists_to_df_first_row_as_columns(ts)
ts_df[obs_id] = ts_df[obs_id].astype('float64')
hybrid_ts_df = pd.merge(hybrid_ts_df, ts_df, how="left", on='time')
hybrid_ts_df.set_index('time', inplace=True)
hybrid_ts_df = hybrid_ts_df.resample('15min',
label='right',
closed='right').sum()
# pd.set_option('display.max_rows', hybrid_ts_df.shape[0]+1)
# pd.set_option('display.max_columns', hybrid_ts_df.shape[1]+1)
# print(hybrid_ts_df)
hybrid_ts_df = replace_negative_numbers_with_nan(hybrid_ts_df)
# print(hybrid_ts_df)
hybrid_ts_df = replace_nan_with_row_average(hybrid_ts_df)
# print(hybrid_ts_df)
#### process mike input ####
distinct_names = coefficients.name.unique()
mike_input = pd.DataFrame()
mike_input_initialized = False
for name in distinct_names:
catchment_coefficients = coefficients[coefficients.name == name]
# print(catchment_coefficients)
catchment = pd.DataFrame()
catchment_initialized = False
for index, row in catchment_coefficients.iterrows():
# print(index, row['curw_obs_id'], row['coefficient'])
if not catchment_initialized:
catchment = (hybrid_ts_df[row['curw_obs_id']] *
row['coefficient']).to_frame(
name=row['curw_obs_id'])
catchment_initialized = True
else:
new = (hybrid_ts_df[row['curw_obs_id']] *
row['coefficient']).to_frame(
name=row['curw_obs_id'])
catchment = pd.merge(catchment, new, how="left", on='time')
if not mike_input_initialized:
mike_input[name] = catchment.sum(axis=1)
mike_input_initialized = True
else:
mike_input = pd.merge(
mike_input, (catchment.sum(axis=1)).to_frame(name=name),
how="left",
on='time')
mike_input.round(1)
return mike_input
except Exception:
traceback.print_exc()
finally:
destroy_Pool(pool)
def prepare_mike_rf_input(start, end):
try:
mike_obs_stations = read_csv(
os.path.join(ROOT_DIRECTORY, 'inputs', 'params',
'mike_rainfall_stations.csv'))
selected_obs_ids = []
for list in mike_obs_stations:
selected_obs_ids.append(str(list[1]))
# [hash_id,station_id,station_name,latitude,longitude]
#### process staton based hybrid timeseries ####
hybrid_ts_df = pd.DataFrame()
hybrid_ts_df['time'] = pd.date_range(start=start, end=end, freq='5min')
pool = get_Pool(host=con_params.CURW_SIM_HOST,
port=con_params.CURW_SIM_PORT,
user=con_params.CURW_SIM_USERNAME,
password=con_params.CURW_SIM_PASSWORD,
db=con_params.CURW_SIM_DATABASE)
obs_pool = get_Pool(host=con_params.CURW_OBS_HOST,
port=con_params.CURW_OBS_PORT,
user=con_params.CURW_OBS_USERNAME,
password=con_params.CURW_OBS_PASSWORD,
db=con_params.CURW_OBS_DATABASE)
TS = Timeseries(pool)
obs_id_hash_id_mapping = get_all_obs_rain_hashids_from_curw_sim(pool)
obs_stations = extract_active_curw_obs_rainfall_stations(
curw_obs_pool=obs_pool, start_time=start, end_time=end)
obs_obs_mapping = find_nearest_obs_stations_for_mike_rainfall_stations(
mike_obs_stations=mike_obs_stations, obs_stations=obs_stations)
for obs_id in selected_obs_ids:
if obs_id in obs_id_hash_id_mapping.keys():
# taking data from curw_sim database (data prepared based on active stations for hechms)
ts = TS.get_timeseries(id_=obs_id_hash_id_mapping.get(obs_id),
start_date=start,
end_date=end)
else:
ts = []
ts.insert(0, ['time', obs_id])
ts_df = list_of_lists_to_df_first_row_as_columns(ts)
ts_df[obs_id] = ts_df[obs_id].astype('float64')
hybrid_ts_df = pd.merge(hybrid_ts_df, ts_df, how="left", on='time')
hybrid_ts_df.set_index('time', inplace=True)
hybrid_ts_df = hybrid_ts_df.resample('15min',
label='right',
closed='right').sum()
mike_input = replace_negative_numbers_with_nan(hybrid_ts_df)
for col in mike_input.columns:
if len(obs_obs_mapping[col]) > 0:
print(col, obs_obs_mapping[col][0])
mike_input[col] = mike_input[col].fillna(
mike_input[obs_obs_mapping[col][0]])
if len(obs_obs_mapping[col]) > 1:
print(col, obs_obs_mapping[col][1])
mike_input[col] = mike_input[col].fillna(
mike_input[obs_obs_mapping[col][1]])
mike_input = replace_nan_with_row_average(mike_input)
mike_input = mike_input.round(1)
station_name_dict = {}
for i in range(len(mike_obs_stations)):
station_name_dict[str(
mike_obs_stations[i][1])] = mike_obs_stations[i][2]
for col in mike_input.columns:
mike_input = mike_input.rename(
columns={col: station_name_dict.get(col)})
# pd.set_option('display.max_rows', mike_input.shape[0]+1)
# pd.set_option('display.max_columns', mike_input.shape[1]+1)
# print(mike_input)
return mike_input
except Exception:
traceback.print_exc()
finally:
destroy_Pool(pool)
destroy_Pool(obs_pool)
def update_rainfall_obs(target_model, method, timestep, start_time, end_time):
"""
Update rainfall observations for flo2d models
:param model: target model
:param method: value interpolation method
:param grid_interpolation: grid interpolation method
:param timestep:
:return:
"""
obs_start = datetime.strptime(start_time, '%Y-%m-%d %H:%M:%S')
try:
# Connect to the database
curw_obs_pool = get_Pool(host=con_params.CURW_OBS_HOST,
user=con_params.CURW_OBS_USERNAME,
password=con_params.CURW_OBS_PASSWORD,
port=con_params.CURW_OBS_PORT,
db=con_params.CURW_OBS_DATABASE)
curw_obs_connection = curw_obs_pool.connection()
curw_sim_pool = get_Pool(host=con_params.CURW_SIM_HOST,
user=con_params.CURW_SIM_USERNAME,
password=con_params.CURW_SIM_PASSWORD,
port=con_params.CURW_SIM_PORT,
db=con_params.CURW_SIM_DATABASE)
TS = Timeseries(pool=curw_sim_pool)
# [hash_id, station_id, station_name, latitude, longitude]
active_obs_stations = extract_active_curw_obs_rainfall_stations(
start_time=start_time, end_time=end_time)[1:]
obs_stations_dict = {
} # keys: obs station id , value: [hash id, name, latitude, longitude]
for obs_index in range(len(active_obs_stations)):
obs_stations_dict[active_obs_stations[obs_index][1]] = [
active_obs_stations[obs_index][0],
active_obs_stations[obs_index][2],
active_obs_stations[obs_index][3],
active_obs_stations[obs_index][4]
]
for obs_id in obs_stations_dict.keys():
meta_data = {
'latitude':
float('%.6f' % float(obs_stations_dict.get(obs_id)[2])),
'longitude':
float('%.6f' % float(obs_stations_dict.get(obs_id)[3])),
'model':
target_model,
'method':
method,
'grid_id':
'rainfall_{}_{}'.format(obs_id,
obs_stations_dict.get(obs_id)[1])
}
tms_id = TS.get_timeseries_id_if_exists(meta_data=meta_data)
if tms_id is None:
tms_id = TS.generate_timeseries_id(meta_data=meta_data)
meta_data['id'] = tms_id
TS.insert_run(meta_data=meta_data)
TS.update_grid_id(id_=tms_id, grid_id=meta_data['grid_id'])
obs_hash_id = obs_stations_dict.get(obs_id)[0]
obs_timeseries = []
if timestep == 5:
ts = extract_obs_rain_5_min_ts(connection=curw_obs_connection,
start_time=obs_start,
end_time=end_time,
id=obs_hash_id)
if ts is not None and len(ts) > 1:
obs_timeseries.extend(
process_5_min_ts(newly_extracted_timeseries=ts,
expected_start=obs_start)[1:])
# obs_start = ts[-1][0]
elif timestep == 15:
ts = extract_obs_rain_15_min_ts(connection=curw_obs_connection,
start_time=obs_start,
end_time=end_time,
id=obs_hash_id)
if ts is not None and len(ts) > 1:
obs_timeseries.extend(
process_15_min_ts(newly_extracted_timeseries=ts,
expected_start=obs_start)[1:])
# obs_start = ts[-1][0]
# for i in range(len(obs_timeseries)):
# if obs_timeseries[i][1] == -99999:
# obs_timeseries[i][1] = 0
if obs_timeseries is not None and len(obs_timeseries) > 0:
TS.replace_data(timeseries=obs_timeseries, tms_id=tms_id)
except Exception as e:
traceback.print_exc()
logger.error(
"Exception occurred while updating obs rainfalls in curw_sim.")
finally:
curw_obs_connection.close()
destroy_Pool(pool=curw_sim_pool)
destroy_Pool(pool=curw_obs_pool)
def update_rainfall_from_file(curw_sim_pool,
flo2d_grid_polygon_map,
stations_dict,
rainfall_df,
flo2d_model,
method,
grid_interpolation,
timestep,
start_time=None,
end_time=None):
"""
Update rainfall observations for flo2d models
:param flo2d_model: flo2d model
:param method: value interpolation method
:param grid_interpolation: grid interpolation method
:param timestep: output timeseries timestep
:return:
"""
# start = datetime.strptime(start_time, '%Y-%m-%d %H:%M:%S')
try:
TS = Sim_Timeseries(pool=curw_sim_pool)
# # [hash_id, station_id, station_name, latitude, longitude]
# flo2d_grid_polygon_map :: [Grid_ ID, X(longitude), Y(latitude), matching_point]
# stations_dict_for_obs = { } # keys: obs station id , value: hash id
for grid in flo2d_grid_polygon_map:
lat = grid[2]
lon = grid[1]
cell_id = grid[0]
meta_data = {
'latitude':
float('%.6f' % float(lat)),
'longitude':
float('%.6f' % float(lon)),
'model':
flo2d_model,
'method':
method,
'grid_id':
'{}_{}_{}'.format(flo2d_model, grid_interpolation,
(str(cell_id)).zfill(10))
}
if len(grid) > 3:
polygon = grid[3]
poly_lat = stations_dict.get(polygon)[1]
poly_lon = stations_dict.get(polygon)[0]
processed_ts = rainfall_df.loc[
(rainfall_df['latitude'] == poly_lat)
& (rainfall_df['longitude'] == poly_lon)][[
'time', 'value'
]].values.tolist()
else:
processed_ts = rainfall_df.groupby('time').mean().round(
3)['value'].reset_index().values.tolist()
tms_id = TS.get_timeseries_id(grid_id=meta_data.get('grid_id'),
method=meta_data.get('method'))
if tms_id is None:
tms_id = TS.generate_timeseries_id(meta_data=meta_data)
meta_data['id'] = tms_id
TS.insert_run(meta_data=meta_data)
print("grid_id:", meta_data['grid_id'])
print(processed_ts)
# for i in range(len(obs_timeseries)):
# if obs_timeseries[i][1] == -99999:
# obs_timeseries[i][1] = 0
if processed_ts is not None and len(processed_ts) > 0:
TS.insert_data(timeseries=processed_ts,
tms_id=tms_id,
upsert=True)
except Exception as e:
traceback.print_exc()
logger.error(
"Exception occurred while updating obs rainfalls in curw_sim.")
finally:
destroy_Pool(pool=curw_sim_pool)
logger.info("Process finished")
def update_rainfall_obs(flo2d_model, method, grid_interpolation, timestep,
start_time, end_time):
"""
Update rainfall observations for flo2d models
:param flo2d_model: flo2d model
:param method: value interpolation method
:param grid_interpolation: grid interpolation method
:param timestep: output timeseries timestep
:return:
"""
obs_start = datetime.strptime(start_time, '%Y-%m-%d %H:%M:%S')
try:
# Connect to the database
curw_obs_pool = get_Pool(host=con_params.CURW_OBS_HOST,
user=con_params.CURW_OBS_USERNAME,
password=con_params.CURW_OBS_PASSWORD,
port=con_params.CURW_OBS_PORT,
db=con_params.CURW_OBS_DATABASE)
curw_obs_connection = curw_obs_pool.connection()
curw_sim_pool = get_Pool(host=con_params.CURW_SIM_HOST,
user=con_params.CURW_SIM_USERNAME,
password=con_params.CURW_SIM_PASSWORD,
port=con_params.CURW_SIM_PORT,
db=con_params.CURW_SIM_DATABASE)
TS = Sim_Timeseries(pool=curw_sim_pool)
# [hash_id, station_id, station_name, latitude, longitude]
# active_obs_stations = read_csv(os.path.join(ROOT_DIR,'grids/obs_stations/rainfall/curw_active_rainfall_obs_stations.csv'))
active_obs_stations = extract_active_curw_obs_rainfall_stations(
start_time=start_time, end_time=end_time)[1:]
flo2d_grids = read_csv(
os.path.join(ROOT_DIR, 'grids/flo2d/{}m.csv'.format(
flo2d_model))) # [Grid_ ID, X(longitude), Y(latitude)]
stations_dict_for_obs = {} # keys: obs station id , value: hash id
for obs_index in range(len(active_obs_stations)):
stations_dict_for_obs[active_obs_stations[obs_index]
[1]] = active_obs_stations[obs_index][0]
# flo2d_obs_mapping = get_flo2d_cells_to_obs_grid_mappings(pool=curw_sim_pool, grid_interpolation=grid_interpolation, flo2d_model=flo2d_model)
flo2d_obs_mapping = find_nearest_obs_stations_for_flo2d_stations(
flo2d_stations_csv=os.path.join(
ROOT_DIR, 'grids/flo2d/{}m.csv'.format(flo2d_model)),
obs_stations=active_obs_stations,
flo2d_model=flo2d_model)
for flo2d_index in range(len(flo2d_grids)):
lat = flo2d_grids[flo2d_index][2]
lon = flo2d_grids[flo2d_index][1]
cell_id = flo2d_grids[flo2d_index][0]
meta_data = {
'latitude':
float('%.6f' % float(lat)),
'longitude':
float('%.6f' % float(lon)),
'model':
flo2d_model,
'method':
method,
'grid_id':
'{}_{}_{}'.format(flo2d_model, grid_interpolation,
(str(cell_id)).zfill(10))
}
tms_id = TS.get_timeseries_id(grid_id=meta_data.get('grid_id'),
method=meta_data.get('method'))
if tms_id is None:
tms_id = TS.generate_timeseries_id(meta_data=meta_data)
meta_data['id'] = tms_id
TS.insert_run(meta_data=meta_data)
print("grid_id:", cell_id)
print("grid map:", flo2d_obs_mapping.get(cell_id))
obs1_station_id = flo2d_obs_mapping.get(cell_id)[0]
obs2_station_id = flo2d_obs_mapping.get(cell_id)[1]
obs3_station_id = flo2d_obs_mapping.get(cell_id)[2]
obs_timeseries = []
if timestep == 5:
if obs1_station_id != str(-1):
obs1_hash_id = stations_dict_for_obs.get(obs1_station_id)
ts = extract_obs_rain_5_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs1_hash_id,
end_time=end_time)
if ts is not None and len(ts) > 1:
obs_timeseries.extend(
process_5_min_ts(newly_extracted_timeseries=ts,
expected_start=obs_start)[1:])
# obs_start = ts[-1][0]
if obs2_station_id != str(-1):
obs2_hash_id = stations_dict_for_obs.get(
obs2_station_id)
ts2 = extract_obs_rain_5_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs2_hash_id,
end_time=end_time)
if ts2 is not None and len(ts2) > 1:
obs_timeseries = fill_missing_values(
newly_extracted_timeseries=ts2,
OBS_TS=obs_timeseries)
if obs_timeseries is not None and len(
obs_timeseries) > 0:
expected_start = obs_timeseries[-1][0]
else:
expected_start = obs_start
obs_timeseries.extend(
process_5_min_ts(
newly_extracted_timeseries=ts2,
expected_start=expected_start)[1:])
# obs_start = ts2[-1][0]
if obs3_station_id != str(-1):
obs3_hash_id = stations_dict_for_obs.get(
obs3_station_id)
ts3 = extract_obs_rain_5_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs3_hash_id,
end_time=end_time)
if ts3 is not None and len(ts3) > 1 and len(
obs_timeseries) > 0:
obs_timeseries = fill_missing_values(
newly_extracted_timeseries=ts3,
OBS_TS=obs_timeseries)
if obs_timeseries is not None:
expected_start = obs_timeseries[-1][0]
else:
expected_start = obs_start
obs_timeseries.extend(
process_5_min_ts(
newly_extracted_timeseries=ts3,
expected_start=expected_start)[1:])
elif timestep == 15:
if obs1_station_id != str(-1):
obs1_hash_id = stations_dict_for_obs.get(obs1_station_id)
ts = extract_obs_rain_15_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs1_hash_id,
end_time=end_time)
if ts is not None and len(ts) > 1:
obs_timeseries.extend(
process_15_min_ts(newly_extracted_timeseries=ts,
expected_start=obs_start)[1:])
# obs_start = ts[-1][0]
if obs2_station_id != str(-1):
obs2_hash_id = stations_dict_for_obs.get(
obs2_station_id)
ts2 = extract_obs_rain_15_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs2_hash_id,
end_time=end_time)
if ts2 is not None and len(ts2) > 1:
obs_timeseries = fill_missing_values(
newly_extracted_timeseries=ts2,
OBS_TS=obs_timeseries)
if obs_timeseries is not None and len(
obs_timeseries) > 0:
expected_start = obs_timeseries[-1][0]
else:
expected_start = obs_start
obs_timeseries.extend(
process_15_min_ts(
newly_extracted_timeseries=ts2,
expected_start=expected_start)[1:])
# obs_start = ts2[-1][0]
if obs3_station_id != str(-1):
obs3_hash_id = stations_dict_for_obs.get(
obs3_station_id)
ts3 = extract_obs_rain_15_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs3_hash_id,
end_time=end_time)
if ts3 is not None and len(ts3) > 1 and len(
obs_timeseries) > 0:
obs_timeseries = fill_missing_values(
newly_extracted_timeseries=ts3,
OBS_TS=obs_timeseries)
if obs_timeseries is not None:
expected_start = obs_timeseries[-1][0]
else:
expected_start = obs_start
obs_timeseries.extend(
process_15_min_ts(
newly_extracted_timeseries=ts3,
expected_start=expected_start)[1:])
for i in range(len(obs_timeseries)):
if obs_timeseries[i][1] == -99999:
obs_timeseries[i][1] = 0
print("### obs timeseries length ###", len(obs_timeseries))
if obs_timeseries is not None and len(
obs_timeseries) > 0 and obs_timeseries[-1][0] != end_time:
obs_timeseries.append(
[datetime.strptime(end_time, DATE_TIME_FORMAT), 0])
final_ts = process_continuous_ts(original_ts=obs_timeseries,
expected_start=datetime.strptime(
start_time, DATE_TIME_FORMAT),
filling_value=0,
timestep=timestep)
if final_ts is not None and len(final_ts) > 0:
TS.insert_data(timeseries=final_ts, tms_id=tms_id, upsert=True)
TS.update_latest_obs(id_=tms_id, obs_end=(final_ts[-1][1]))
except Exception as e:
traceback.print_exc()
logger.error(
"Exception occurred while updating obs rainfalls in curw_sim.")
finally:
curw_obs_connection.close()
destroy_Pool(pool=curw_sim_pool)
destroy_Pool(pool=curw_obs_pool)
logger.info("Process finished")
try:
pool = get_Pool(host=con_params.CURW_SIM_HOST, port=con_params.CURW_SIM_PORT, user=con_params.CURW_SIM_USERNAME,
password=con_params.CURW_SIM_PASSWORD, db=con_params.CURW_SIM_DATABASE)
method = MethodEnum.getAbbreviation(MethodEnum.MME)
run_table = "run"
data_table = "data"
end = (datetime.now() - timedelta(days=51)).strftime("%Y-%m-%d %H:%M:00")
hash_ids = flush_common.get_curw_sim_hash_ids(pool=pool, run_table=run_table, model=HecHMS, method=method, obs_end_start=None,
obs_end_end=None, grid_id=None)
TS = flush_common.Timeseries(pool=pool, run_table=run_table, data_table=data_table)
TS_rain = Timeseries(pool)
###########################################################################
# Delete run entries without any observed data within last 50 days period #
###########################################################################
count = 0
for id in hash_ids:
obs_end = TS_rain.get_obs_end(id)
if (obs_end is None) or (obs_end.strftime("%Y-%m-%d %H:%M:%S") < end):
count += 1
TS.delete_all_by_hash_id(id)
print(count, id)
hash_ids.remove(id)
#####################################################################################################
# delete a specific timeseries defined by a given hash id from data table for specified time period #
def update_rainfall_obs(curw_obs_pool, curw_sim_pool, flo2d_model, method, grid_interpolation, timestep, start_time, end_time):
"""
Update rainfall observations for flo2d models
:param flo2d_model: flo2d model
:param method: value interpolation method
:param grid_interpolation: grid interpolation method
:param timestep: output timeseries timestep
:return:
"""
# obs_start = datetime.strptime(start_time, '%Y-%m-%d %H:%M:%S')
try:
curw_obs_connection = curw_obs_pool.connection()
# [hash_id, station_id, station_name, latitude, longitude]
# active_obs_stations = read_csv(os.path.join(ROOT_DIR,'grids/obs_stations/rainfall/curw_active_rainfall_obs_stations.csv'))
active_obs_stations = extract_active_curw_obs_rainfall_stations(curw_obs_pool=curw_obs_pool, start_time=start_time, end_time=end_time)[1:]
flo2d_grids = read_csv(os.path.join(ROOT_DIR,'grids/flo2d/{}m.csv'.format(flo2d_model))) # [Grid_ ID, X(longitude), Y(latitude)]
stations_dict_for_obs = { } # keys: obs station id , value: hash id
for obs_index in range(len(active_obs_stations)):
stations_dict_for_obs[active_obs_stations[obs_index][1]] = active_obs_stations[obs_index][0]
# flo2d_obs_mapping = get_flo2d_cells_to_obs_grid_mappings(pool=curw_sim_pool, grid_interpolation=grid_interpolation, flo2d_model=flo2d_model)
flo2d_obs_mapping = find_nearest_obs_stations_for_flo2d_stations(
flo2d_stations_csv=os.path.join(ROOT_DIR,'grids/flo2d/{}m.csv'.format(flo2d_model)),
obs_stations=active_obs_stations, flo2d_model=flo2d_model)
# retrieve observed timeseries
obs_df = pd.DataFrame()
obs_df['time'] = pd.date_range(start=start_time, end=end_time, freq='5min')
for obs_id in stations_dict_for_obs.keys():
ts = extract_obs_rain_5_min_ts(connection=curw_obs_connection, start_time=start_time,
id=stations_dict_for_obs.get(obs_id), end_time=end_time)
ts.insert(0, ['time', obs_id])
ts_df = list_of_lists_to_df_first_row_as_columns(ts)
ts_df[obs_id] = ts_df[obs_id].astype('float64')
obs_df = pd.merge(obs_df, ts_df, how="left", on='time')
obs_df.set_index('time', inplace=True)
obs_df['0'] = 0
if timestep == 15:
obs_df = obs_df.resample('15min', label='right', closed='right').sum()
TS = Sim_Timeseries(pool=curw_sim_pool)
for flo2d_index in range(len(flo2d_grids)):
lat = flo2d_grids[flo2d_index][2]
lon = flo2d_grids[flo2d_index][1]
cell_id = flo2d_grids[flo2d_index][0]
meta_data = {
'latitude': float('%.6f' % float(lat)), 'longitude': float('%.6f' % float(lon)),
'model': flo2d_model, 'method': method,
'grid_id': '{}_{}_{}'.format(flo2d_model, grid_interpolation, (str(cell_id)).zfill(10))
}
tms_id = TS.get_timeseries_id(grid_id=meta_data.get('grid_id'), method=meta_data.get('method'))
if tms_id is None:
tms_id = TS.generate_timeseries_id(meta_data=meta_data)
meta_data['id'] = tms_id
TS.insert_run(meta_data=meta_data)
print(datetime.now().strftime(DATE_TIME_FORMAT))
print("grid_id:", cell_id)
obs_station_ids = flo2d_obs_mapping.get(cell_id)
if len(obs_station_ids) == 1:
obs_ts_df = obs_df[obs_station_ids].to_frame(name='final')
elif len(obs_station_ids) == 2:
obs_ts_df = obs_df[obs_station_ids]
obs_ts_df[obs_station_ids[0]] = obs_ts_df[obs_station_ids[0]].fillna(obs_ts_df[obs_station_ids[1]])
obs_ts_df['final'] = obs_ts_df[obs_station_ids[0]]
elif len(obs_station_ids) == 3:
obs_ts_df = obs_df[obs_station_ids]
obs_ts_df[obs_station_ids[1]] = obs_ts_df[obs_station_ids[1]].fillna(obs_ts_df[obs_station_ids[2]])
obs_ts_df[obs_station_ids[0]] = obs_ts_df[obs_station_ids[0]].fillna(obs_ts_df[obs_station_ids[1]])
obs_ts_df['final'] = obs_ts_df[obs_station_ids[0]]
else:
obs_ts_df = obs_df['0'].to_frame(name='final')
final_ts_df = obs_ts_df['final'].reset_index()
final_ts_df['time'] = final_ts_df['time'].dt.strftime(DATE_TIME_FORMAT)
final_ts = final_ts_df.values.tolist()
if final_ts is not None and len(final_ts) > 0:
TS.replace_data(timeseries=final_ts, tms_id=tms_id)
TS.update_latest_obs(id_=tms_id, obs_end=(final_ts[-1][1]))
except Exception as e:
traceback.print_exc()
logger.error("Exception occurred while updating obs rainfalls in curw_sim.")
finally:
curw_obs_connection.close()
destroy_Pool(pool=curw_sim_pool)
destroy_Pool(pool=curw_obs_pool)
logger.info("Process finished")
def prepare_rain(curw_sim_pool, rain_file_path, curw_sim_hash_id, start_time,
end_time, target_model):
# retrieve observed timeseries
df = pd.DataFrame()
df['time'] = pd.date_range(start=start_time, end=end_time, freq='5min')
TS = Timeseries(curw_sim_pool)
ts = TS.get_timeseries(id_=curw_sim_hash_id,
start_date=start_time,
end_date=end_time)
ts.insert(0, ['time', 'value'])
ts_df = list_of_lists_to_df_first_row_as_columns(ts)
ts_df['value'] = ts_df['value'].astype('float64')
df = pd.merge(df, ts_df, how="left", on='time')
df.set_index('time', inplace=True)
df = df.dropna()
if target_model == "flo2d_250":
timestep = 5
elif target_model in ("flo2d_150", "flo2d_150_v2"):
timestep = 15
else:
timestep = 5
if timestep == 15:
df = df.resample('15min', label='right', closed='right').sum()
df = replace_negative_numbers_with_nan(df)
timeseries = df['value'].reset_index().values.tolist()
start_time = datetime.strptime(start_time, DATE_TIME_FORMAT)
rain_dat = []
total_rain = 0
cumulative_timeseries = []
for i in range(len(timeseries)):
total_rain += float(timeseries[i][1])
cumulative_timeseries.append(total_rain)
for i in range(len(timeseries)):
time_col = '%.3f' % ((
(timeseries[i][0] - start_time).total_seconds()) / 3600)
if total_rain > 0:
rain_col = '%.3f' % (cumulative_timeseries[i] / total_rain)
else:
rain_col = '%.3f' % (0)
rain_dat.append("R " + time_col.ljust(14) + rain_col +
" ")
rain_dat.insert(
0, " {} 5 0 0 ".format('%.3f' %
total_rain))
rain_dat.insert(0, " 0 0 ")
write_to_file(rain_file_path, rain_dat)
def update_rainfall_fcsts(flo2d_model, method, grid_interpolation, model_list,
timestep):
"""
Update rainfall forecasts for flo2d models
:param flo2d_model: flo2d model
:param method: value interpolation method
:param grid_interpolation: grid interpolation method
:param model_list: list of forecast model and their versions used to calculate the rainfall
e.g.: [["WRF_E", "v4"],["WRF_SE", "v4"]]
:param timestep: output timeseries timestep
:return:
"""
try:
# Connect to the database
curw_sim_pool = get_Pool(host=CURW_SIM_HOST,
user=CURW_SIM_USERNAME,
password=CURW_SIM_PASSWORD,
port=CURW_SIM_PORT,
db=CURW_SIM_DATABASE)
curw_fcst_pool = get_Pool(host=CURW_FCST_HOST,
user=CURW_FCST_USERNAME,
password=CURW_FCST_PASSWORD,
port=CURW_FCST_PORT,
db=CURW_FCST_DATABASE)
Sim_TS = Sim_Timeseries(pool=curw_sim_pool)
Fcst_TS = Fcst_Timeseries(pool=curw_fcst_pool)
flo2d_grids = read_csv('grids/flo2d/{}m.csv'.format(
flo2d_model)) # [Grid_ ID, X(longitude), Y(latitude)]
flo2d_wrf_mapping = get_flo2d_cells_to_wrf_grid_mappings(
pool=curw_sim_pool,
grid_interpolation=grid_interpolation,
flo2d_model=flo2d_model)
for flo2d_index in range(len(flo2d_grids)): # len(flo2d_grids)
lat = flo2d_grids[flo2d_index][2]
lon = flo2d_grids[flo2d_index][1]
cell_id = flo2d_grids[flo2d_index][0]
meta_data = {
'latitude':
float('%.6f' % float(lat)),
'longitude':
float('%.6f' % float(lon)),
'model':
flo2d_model,
'method':
method,
'grid_id':
'{}_{}_{}'.format(flo2d_model, grid_interpolation,
(str(cell_id)).zfill(10))
}
tms_id = Sim_TS.get_timeseries_id(grid_id=meta_data.get('grid_id'),
method=meta_data.get('method'))
if tms_id is None:
tms_id = Sim_TS.generate_timeseries_id(meta_data=meta_data)
meta_data['id'] = tms_id
Sim_TS.insert_run(meta_data=meta_data)
obs_end = Sim_TS.get_obs_end(id_=tms_id)
fcst_timeseries = []
for i in range(len(model_list)):
source_id = get_source_id(pool=curw_fcst_pool,
model=model_list[i][0],
version=model_list[i][1])
sim_tag = model_list[i][2]
coefficient = model_list[i][3]
temp_timeseries = []
if timestep == 5:
if obs_end is not None:
temp_timeseries = convert_15_min_ts_to_5_mins_ts(
newly_extracted_timeseries=Fcst_TS.
get_latest_timeseries(sim_tag=sim_tag,
station_id=flo2d_wrf_mapping.
get(meta_data['grid_id']),
start=obs_end,
source_id=source_id,
variable_id=1,
unit_id=1))
else:
temp_timeseries = convert_15_min_ts_to_5_mins_ts(
newly_extracted_timeseries=Fcst_TS.
get_latest_timeseries(sim_tag=sim_tag,
station_id=flo2d_wrf_mapping.
get(meta_data['grid_id']),
source_id=source_id,
variable_id=1,
unit_id=1))
elif timestep == 15:
if obs_end is not None:
temp_timeseries = Fcst_TS.get_latest_timeseries(
sim_tag=sim_tag,
station_id=flo2d_wrf_mapping.get(
meta_data['grid_id']),
start=obs_end,
source_id=source_id,
variable_id=1,
unit_id=1)
else:
temp_timeseries = Fcst_TS.get_latest_timeseries(
sim_tag=sim_tag,
station_id=flo2d_wrf_mapping.get(
meta_data['grid_id']),
source_id=source_id,
variable_id=1,
unit_id=1)
if coefficient != 1:
for j in range(len(temp_timeseries)):
temp_timeseries[j][1] = float(
temp_timeseries[j][1]) * coefficient
if i == 0:
fcst_timeseries = temp_timeseries
else:
fcst_timeseries = append_value_for_timestamp(
existing_ts=fcst_timeseries, new_ts=temp_timeseries)
sum_timeseries = summed_timeseries(fcst_timeseries)
for i in range(len(sum_timeseries)):
if float(sum_timeseries[i][1]) < 0:
sum_timeseries[i][1] = 0
if sum_timeseries is not None and len(sum_timeseries) > 0:
Sim_TS.insert_data(timeseries=sum_timeseries,
tms_id=tms_id,
upsert=True)
except Exception as e:
traceback.print_exc()
logger.error(
"Exception occurred while updating fcst rainfalls in curw_sim.")
finally:
destroy_Pool(curw_sim_pool)
destroy_Pool(curw_fcst_pool)
def update_rainfall_obs(flo2d_model, method, grid_interpolation, timestep):
"""
Update rainfall observations for flo2d models
:param flo2d_model: flo2d model
:param method: value interpolation method
:param grid_interpolation: grid interpolation method
:param timestep: output timeseries timestep
:return:
"""
now = datetime.now()
OBS_START_STRING = (now - timedelta(days=10)).strftime('%Y-%m-%d %H:00:00')
OBS_START = datetime.strptime(OBS_START_STRING, '%Y-%m-%d %H:%M:%S')
try:
# Connect to the database
curw_obs_pool = get_Pool(host=CURW_OBS_HOST,
user=CURW_OBS_USERNAME,
password=CURW_OBS_PASSWORD,
port=CURW_OBS_PORT,
db=CURW_OBS_DATABASE)
curw_obs_connection = curw_obs_pool.connection()
curw_sim_pool = get_Pool(host=CURW_SIM_HOST,
user=CURW_SIM_USERNAME,
password=CURW_SIM_PASSWORD,
port=CURW_SIM_PORT,
db=CURW_SIM_DATABASE)
# test ######
# pool = get_Pool(host=HOST, user=USERNAME, password=PASSWORD, port=PORT, db=DATABASE)
TS = Sim_Timeseries(pool=curw_sim_pool)
# [hash_id, station_id, station_name, latitude, longitude]
active_obs_stations = read_csv(
'grids/obs_stations/rainfall/curw_active_rainfall_obs_stations.csv'
)
flo2d_grids = read_csv('grids/flo2d/{}m.csv'.format(
flo2d_model)) # [Grid_ ID, X(longitude), Y(latitude)]
stations_dict_for_obs = {} # keys: obs station id , value: hash id
for obs_index in range(len(active_obs_stations)):
stations_dict_for_obs[active_obs_stations[obs_index]
[1]] = active_obs_stations[obs_index][0]
flo2d_obs_mapping = get_flo2d_cells_to_obs_grid_mappings(
pool=curw_sim_pool,
grid_interpolation=grid_interpolation,
flo2d_model=flo2d_model)
for flo2d_index in range(len(flo2d_grids)):
obs_start = OBS_START
lat = flo2d_grids[flo2d_index][2]
lon = flo2d_grids[flo2d_index][1]
cell_id = flo2d_grids[flo2d_index][0]
meta_data = {
'latitude':
float('%.6f' % float(lat)),
'longitude':
float('%.6f' % float(lon)),
'model':
flo2d_model,
'method':
method,
'grid_id':
'{}_{}_{}'.format(flo2d_model, grid_interpolation,
(str(cell_id)).zfill(10))
}
tms_id = TS.get_timeseries_id(grid_id=meta_data.get('grid_id'),
method=meta_data.get('method'))
if tms_id is None:
tms_id = TS.generate_timeseries_id(meta_data=meta_data)
meta_data['id'] = tms_id
TS.insert_run(meta_data=meta_data)
obs_end = TS.get_obs_end(id_=tms_id)
if obs_end is not None:
obs_start = obs_end - timedelta(hours=1)
obs1_station_id = str(
flo2d_obs_mapping.get(meta_data['grid_id'])[0])
obs2_station_id = str(
flo2d_obs_mapping.get(meta_data['grid_id'])[1])
obs3_station_id = str(
flo2d_obs_mapping.get(meta_data['grid_id'])[2])
obs_timeseries = []
if timestep == 5:
if obs1_station_id != str(-1):
obs1_hash_id = stations_dict_for_obs.get(obs1_station_id)
ts = extract_obs_rain_5_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs1_hash_id)
if ts is not None and len(ts) > 1:
obs_timeseries.extend(
process_5_min_ts(newly_extracted_timeseries=ts,
expected_start=obs_start)[1:])
# obs_start = ts[-1][0]
if obs2_station_id != str(-1):
obs2_hash_id = stations_dict_for_obs.get(
obs2_station_id)
ts2 = extract_obs_rain_5_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs2_hash_id)
if ts2 is not None and len(ts2) > 1:
obs_timeseries = fill_missing_values(
newly_extracted_timeseries=ts2,
OBS_TS=obs_timeseries)
if obs_timeseries is not None and len(
obs_timeseries) > 0:
expected_start = obs_timeseries[-1][0]
else:
expected_start = obs_start
obs_timeseries.extend(
process_5_min_ts(
newly_extracted_timeseries=ts2,
expected_start=expected_start)[1:])
# obs_start = ts2[-1][0]
if obs3_station_id != str(-1):
obs3_hash_id = stations_dict_for_obs.get(
obs3_station_id)
ts3 = extract_obs_rain_5_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs3_hash_id)
if ts3 is not None and len(ts3) > 1 and len(
obs_timeseries) > 0:
obs_timeseries = fill_missing_values(
newly_extracted_timeseries=ts3,
OBS_TS=obs_timeseries)
if obs_timeseries is not None:
expected_start = obs_timeseries[-1][0]
else:
expected_start = obs_start
obs_timeseries.extend(
process_5_min_ts(
newly_extracted_timeseries=ts3,
expected_start=expected_start)[1:])
elif timestep == 15:
if obs1_station_id != str(-1):
obs1_hash_id = stations_dict_for_obs.get(obs1_station_id)
ts = extract_obs_rain_15_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs1_hash_id)
if ts is not None and len(ts) > 1:
obs_timeseries.extend(
process_15_min_ts(newly_extracted_timeseries=ts,
expected_start=obs_start)[1:])
# obs_start = ts[-1][0]
if obs2_station_id != str(-1):
obs2_hash_id = stations_dict_for_obs.get(
obs2_station_id)
ts2 = extract_obs_rain_15_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs2_hash_id)
if ts2 is not None and len(ts2) > 1:
obs_timeseries = fill_missing_values(
newly_extracted_timeseries=ts2,
OBS_TS=obs_timeseries)
if obs_timeseries is not None and len(
obs_timeseries) > 0:
expected_start = obs_timeseries[-1][0]
else:
expected_start = obs_start
obs_timeseries.extend(
process_15_min_ts(
newly_extracted_timeseries=ts2,
expected_start=expected_start)[1:])
# obs_start = ts2[-1][0]
if obs3_station_id != str(-1):
obs3_hash_id = stations_dict_for_obs.get(
obs3_station_id)
ts3 = extract_obs_rain_15_min_ts(
connection=curw_obs_connection,
start_time=obs_start,
id=obs3_hash_id)
if ts3 is not None and len(ts3) > 1 and len(
obs_timeseries) > 0:
obs_timeseries = fill_missing_values(
newly_extracted_timeseries=ts3,
OBS_TS=obs_timeseries)
if obs_timeseries is not None:
expected_start = obs_timeseries[-1][0]
else:
expected_start = obs_start
obs_timeseries.extend(
process_15_min_ts(
newly_extracted_timeseries=ts3,
expected_start=expected_start)[1:])
for i in range(len(obs_timeseries)):
if obs_timeseries[i][1] == -99999:
obs_timeseries[i][1] = 0
if obs_timeseries is not None and len(obs_timeseries) > 0:
TS.insert_data(timeseries=obs_timeseries,
tms_id=tms_id,
upsert=True)
TS.update_latest_obs(id_=tms_id,
obs_end=(obs_timeseries[-1][1]))
except Exception as e:
traceback.print_exc()
logger.error(
"Exception occurred while updating obs rainfalls in curw_sim.")
finally:
curw_obs_connection.close()
destroy_Pool(pool=curw_sim_pool)
destroy_Pool(pool=curw_obs_pool)
logger.info("Process finished")
def update_rainfall_fcsts(target_model, method, grid_interpolation, model_list,
timestep):
"""
Update rainfall forecasts for flo2d models
:param target_model: target model for which input ins prepared
:param method: value interpolation method
:param grid_interpolation: grid interpolation method
:param model_list: list of forecast model and their versions used to calculate the rainfall
e.g.: [["WRF_E", "4.0", "evening_18hrs"],["WRF_SE", "v4", ,"evening_18hrs"],["WRF_Ensemble", "4.0", ,"MME"]]
:param timestep: output timeseries timestep
:return:
"""
try:
# Connect to the database
curw_sim_pool = get_Pool(host=CURW_SIM_HOST,
user=CURW_SIM_USERNAME,
password=CURW_SIM_PASSWORD,
port=CURW_SIM_PORT,
db=CURW_SIM_DATABASE)
curw_fcst_pool = get_Pool(host=CURW_FCST_HOST,
user=CURW_FCST_USERNAME,
password=CURW_FCST_PASSWORD,
port=CURW_FCST_PORT,
db=CURW_FCST_DATABASE)
Sim_TS = Sim_Timeseries(pool=curw_sim_pool)
Fcst_TS = Fcst_Timeseries(pool=curw_fcst_pool)
# [hash_id, station_id, station_name, latitude, longitude]
active_obs_stations = read_csv(
'grids/obs_stations/rainfall/curw_active_rainfall_obs_stations.csv'
)
obs_stations_dict = {
} # keys: obs station id , value: [name, latitude, longitude]
for obs_index in range(len(active_obs_stations)):
obs_stations_dict[active_obs_stations[obs_index][1]] = [
active_obs_stations[obs_index][2],
active_obs_stations[obs_index][3],
active_obs_stations[obs_index][4]
]
obs_d03_mapping = get_obs_to_d03_grid_mappings_for_rainfall(
pool=curw_sim_pool, grid_interpolation=grid_interpolation)
for obs_id in obs_stations_dict.keys():
meta_data = {
'latitude':
float('%.6f' % float(obs_stations_dict.get(obs_id)[1])),
'longitude':
float('%.6f' % float(obs_stations_dict.get(obs_id)[2])),
'model':
target_model,
'method':
method,
'grid_id':
'rainfall_{}_{}_{}'.format(obs_id,
obs_stations_dict.get(obs_id)[0],
grid_interpolation)
}
tms_id = Sim_TS.get_timeseries_id_if_exists(meta_data=meta_data)
if tms_id is None:
tms_id = Sim_TS.generate_timeseries_id(meta_data=meta_data)
meta_data['id'] = tms_id
Sim_TS.insert_run(meta_data=meta_data)
obs_end = Sim_TS.get_obs_end(id_=tms_id)
fcst_timeseries = []
for i in range(len(model_list)):
source_id = get_source_id(pool=curw_fcst_pool,
model=model_list[i][0],
version=model_list[i][1])
sim_tag = model_list[i][2]
coefficient = model_list[i][3]
temp_timeseries = []
if timestep == 5:
if obs_end is not None:
temp_timeseries = convert_15_min_ts_to_5_mins_ts(
newly_extracted_timeseries=Fcst_TS.
get_latest_timeseries(sim_tag=sim_tag,
station_id=obs_d03_mapping.
get(meta_data['grid_id'])[0],
start=obs_end,
source_id=source_id,
variable_id=1,
unit_id=1))
else:
temp_timeseries = convert_15_min_ts_to_5_mins_ts(
newly_extracted_timeseries=Fcst_TS.
get_latest_timeseries(sim_tag=sim_tag,
station_id=obs_d03_mapping.
get(meta_data['grid_id'])[0],
source_id=source_id,
variable_id=1,
unit_id=1))
elif timestep == 15:
if obs_end is not None:
temp_timeseries = Fcst_TS.get_latest_timeseries(
sim_tag=sim_tag,
station_id=obs_d03_mapping.get(
meta_data['grid_id'])[0],
start=obs_end,
source_id=source_id,
variable_id=1,
unit_id=1)
else:
temp_timeseries = Fcst_TS.get_latest_timeseries(
sim_tag=sim_tag,
station_id=obs_d03_mapping.get(
meta_data['grid_id'])[0],
source_id=source_id,
variable_id=1,
unit_id=1)
if coefficient != 1:
for j in range(len(temp_timeseries)):
temp_timeseries[j][1] = float(
temp_timeseries[j][1]) * coefficient
if i == 0:
fcst_timeseries = temp_timeseries
else:
fcst_timeseries = append_value_for_timestamp(
existing_ts=fcst_timeseries, new_ts=temp_timeseries)
sum_timeseries = summed_timeseries(fcst_timeseries)
for i in range(len(sum_timeseries)):
if float(sum_timeseries[i][1]) < 0:
sum_timeseries[i][1] = 0
if sum_timeseries is not None and len(sum_timeseries) > 0:
Sim_TS.insert_data(timeseries=sum_timeseries,
tms_id=tms_id,
upsert=True)
except Exception as e:
traceback.print_exc()
logger.error(
"Exception occurred while updating fcst rainfalls in curw_sim.")
finally:
destroy_Pool(curw_sim_pool)
destroy_Pool(curw_fcst_pool)