ueb_df.columns = [
'ueb_' + name if 'ueb' not in name else name for name in ueb_df.columns
]
if os.path.isdir(snow17_dir):
ts_file_list = [
os.path.join(snow17_dir, name) for name in os.listdir(snow17_dir)
if name[-2:] == 'ts'
]
snow17_df = get_sac_ts_dataframe(ts_file_list, sim_skip=snow17_skip)
snow17_df.columns = [
'snow17_' + name if 'snow17' not in name else name
for name in snow17_df.columns
]
obs_df = get_obs_dataframe(obs_discharge_path)
# step2 get plots for each water year
def get_discharge_stats(obs, sim):
nse = 1 - sum(np.power(
(obs - sim), 2)) / sum(np.power((obs - obs.mean()), 2))
bias = (sim - obs).mean()
rmse = np.sqrt(np.mean(np.power((obs - sim), 2)))
return {'nse': nse, 'bias': bias, 'rmse': rmse}
fig, ax_list = plt.subplots(len(water_year_list),
2,
figsize=(20, 5 * len(water_year_list)))
snow17_skip = 136
ueb_skip = 121
start_time = '1989-10-1'
end_time = '2005-9-30'
dt = 6
result_dir = os.path.join(
os.getcwd(), 'discharge_table_analysis_{}_{}'.format(
watershed, 'all' if end_time == '' else start_time[:4] + end_time[:4]))
if not os.path.isdir(result_dir):
os.makedirs(result_dir)
# step 1 get discharge data frame ##############################################
obs_df = get_obs_dataframe(obs_file, start_time, end_time)
ueb_df = get_sim_dataframe(ueb_file,
start_time=start_time,
end_time=end_time,
sim_skip=ueb_skip)
snow17_df = get_sim_dataframe(snow17_file,
start_time=start_time,
end_time=end_time,
sim_skip=snow17_skip)
DF = pd.concat(
[obs_df, ueb_df, snow17_df], axis=1
) # obs should be first used by the loop to get the discharge threshold
DF.columns = [
'obs', 'ueb', 'snow17'
] # obs should be first used by the loop to get the discharge threshold
folder_path = os.path.join(os.getcwd(), output_folder, folder)
if os.path.isdir(folder_path):
ts_file_name = [
x for x in os.listdir(folder_path) if x[-3:] == '.ts'
].pop()
ts_file_path = os.path.join(folder_path, ts_file_name)
sim = get_sim_dataframe(ts_file_path,
start_time=start_time,
end_time=end_time)
sim_list.append(sim)
column_names.append(folder)
DF = pd.concat(sim_list, axis=1, join='inner')
DF.columns = column_names
obs = get_obs_dataframe(obs_file_path)
# get the statistics for all the options
stat_result = pd.DataFrame(columns=[
'rmse',
'nse',
'mae',
'r',
'bias',
# 'monthly_bias', 'annual_bias', 'vol_err'
])
for column in column_names:
df = pd.concat([DF[column], obs], axis=1,
join_axes=[DF[column].index]).reset_index()
df.columns = ['time', 'sim', 'obs']
rmse, nse, mae, r, bias = get_basic_stats(df)
# compare discharge with obs.
concat_df['discharge_diff'] = concat_df['ueb_discharge'] - concat_df[
'snow17_discharge']
fig, ax = plt.subplots(2, 1, figsize=(10, 8))
concat_df.plot(
y=['ueb_discharge', 'snow17_discharge'],
ax=ax[0],
style=['-', ':'],
title='Compare discharge between UEB and Snow17',
)
ax[0].set_ylabel('discharge(cms)')
if obs_discharge_path:
obs_discharge = get_obs_dataframe(obs_discharge_path,
start_time=start_time,
end_time=end_time)
if len(obs_discharge) != 0:
obs_discharge.plot.area(ax=ax[0], alpha=0.3)
concat_df.plot(
y=['discharge_diff'],
ax=ax[1],
title='Difference of discharge between UEB and Snow17',
legend=False,
)
ax[1].legend(['ueb - snow17'])
ax[1].set_ylabel('discharge(cms)')
plt.tight_layout()
fig.savefig(os.path.join(result_dir, 'compare_discharge.png'))