Homec4science
Diffusion solar_potential (master) Browse

solarPOT/Potential_hourly/merge_commune_all.py
f9c7e8d98b53master

Last Change
Raw File

merge_commune_all.py
View Options

import numpy as np
import pandas as pd
import xarray as xr
import os
import time
import sys
COMM_FP = '/work/hyenergy/output/solar_potential/Sonnendach_Gemeinden_Dachflaechen.csv'
OUTFILE = sys.argv[1]
INFILES = lambda i: (os.path.splitext( OUTFILE )[0] + '_%d.nc' %i)
N_BATCHES = int(sys.argv[2])
OUT_PV = os.path.splitext( OUTFILE )[0] + '_pv.csv'
OUT_GT = os.path.splitext( OUTFILE )[0] + '_gt.csv'
# Load data
comm_map = pd.read_csv(COMM_FP, encoding = 'cp1250', usecols = ['DF_UID', 'BFS_NUMMER', 'NAME'] )
# For-loop
init = True
gt_weight = 0
BFS_IDs = np.sort(comm_map.BFS_NUMMER.unique())
for BATCH in range(N_BATCHES):
tt = time.time()
new_file = xr.open_dataset( INFILES(BATCH) )
new_data = new_file.sel(BFS_NUMMER = new_file.BFS_NUMMER.isin(BFS_IDs))[['pv_potential', 'gt_weighted_by_APV', 'APV_weight_sum']]
if init:
all_data = new_data
init = False
else:
all_data = (all_data + new_data).combine_first(all_data).combine_first(new_data)
"""
new_df = new_data.sel(BFS_NUMMER = new_data.BFS_NUMMER.isin(BFS_IDs)).to_dataframe().dropna().reset_index()
pv_new_df = new_df.pivot(values='pv_potential', index='time', columns='BFS_NUMMER')
gt_new_df = new_df.pivot(values='gt_weighted_by_APV', index='time', columns='BFS_NUMMER')
if init:
init = False
pv_df = pd.DataFrame(index = pv_new_df.index, columns = BFS_IDs).fillna(0)
gt_df = pd.DataFrame(index = pv_new_df.index, columns = BFS_IDs).fillna(0)
A_df = pd.Series(np.zeros(len(BFS_IDs)), index = BFS_IDs)
pv_df[pv_new_df.columns] += pv_new_df
gt_df[gt_new_df.columns] += gt_new_df
A_df[new_data.BFS_NUMMER.values] += new_data.APV_weight_sum.to_pandas()
"""
print('Executed iteration %i in %.2fs' %(BATCH, time.time()-tt))
# Fill all missing values with 0
all_data = all_data.fillna(0)
all_data['time'] = new_file.time
print(all_data)
# Convert data to pandas tables
all_data_df = all_data.drop('APV_weight_sum').to_dataframe().dropna().reset_index()
pv_df = all_data_df.pivot(values='pv_potential', index='time', columns='BFS_NUMMER')
gt_df = all_data_df.pivot(values='gt_weighted_by_APV', index='time', columns='BFS_NUMMER')
A_df = all_data.APV_weight_sum.to_pandas()
# Compute average tilted irrad by dividing by A_df
gt_df.loc[:,A_df > 0] /= A_df[A_df > 0]
print(pv_df.head())
print(gt_df.head())
# CHECK FOR CONSISTENCE WITH ROOFS:
roof_pv = pd.read_csv(os.path.join(os.path.split( OUTFILE )[0], '2016_annual_roofs.csv'), index_col = 0)
roof_pv['PV_roof'] = roof_pv.suitable * roof_pv.EPV_kWh
PV_roof_aggr = roof_pv.groupby('BFS_NUMMER').PV_roof.sum()
PV_roof_aggr.index = PV_roof_aggr.index.astype('int')
PV_comm_aggr = pv_df.sum().rename('PV_comm')
PV_comm_aggr.index = PV_comm_aggr.index.astype('int')
max_diff = (abs(PV_comm_aggr-PV_roof_aggr ) / PV_roof_aggr).max()
if max_diff > 1e-5:
print('Significant deviation (%f)' %(max_diff))
else:
print('Execution OK')
# Save output
tt = time.time()
pv_df.to_csv( OUT_PV )
gt_df.to_csv( OUT_GT )
print('\nSaved outputs %s and %s in %.2fs' %(OUT_PV, OUT_GT, time.time()-tt))
c4science · Help