def power(config):
"""Reads 'time series' from netcdf time series file, and adds power as a variable. """
if __name__ == "__main__":
logger = loghelper.create_logger(config)
else:
logger = loghelper.get_logger(config['log.name'])
# Number of samples to use should be in here
# Whether to normalise power should be in here
pnorm = config['pnorm']
pdist = config['pdist']
sstd = config['sstd']
dstd = config['dstd']
pquants = config['pquants']
quantiles = np.array(pquants)
logger.debug(pnorm)
if pdist:
n=pdist
grid_id = config['grid_id']
init_time = config['init_time']
pcurve_dir = config['pcurve_dir']
ts_dir = config['tseries_dir']
tseries_file = expand(config['tseries_file'], config)
power_file = expand(config['power_file'], config)
logger.info('Estimating power from time series: %s ' % tseries_file)
logger.info('Writing power time series to: %s ' % power_file)
dataset_in = Dataset(tseries_file, 'a')
# Get dimensions
dims = dataset_in.dimensions
ntime = len(dims['time'])
nloc = len(dims['location'])
nheight = len(dims['height'])
loc_str_len = len(dims['loc_str_length'])
# Get coordinate variables
nctime = dataset_in.variables['time']
datetimes = netcdftime.num2date(nctime, nctime.units)
location = [''.join(l.filled(' ')).strip() for l in dataset_in.variables['location']]
height = dataset_in.variables['height']
# Get attributes
metadata = config['metadata']
if power_file == tseries_file:
dataset_out = dataset_in
else:
dataset_out = Dataset(power_file, 'w')
# Get number of quantiles
nq = len(quantiles)
pdata = np.ma.zeros((ntime,nloc,nheight,nq+1), np.float) # mean will be 1st value
use_locs = []
for l,loc in enumerate(location):
pcurve_file = '%s/%s.csv' %(pcurve_dir, loc)
# mask power data if no power curve found for this park
if not os.path.exists(pcurve_file):
#logger.debug("Power curve: %s not found, skipping" % pcurve_file)
pdata[:,l,:,:] = np.ma.masked
continue
logger.info('Predicting power output for %s' % loc )
#
# Open power curve
#
use_locs.append(l)
pcurve = from_file(pcurve_file)
for h in range(nheight):
speed = dataset_in.variables['SPEED'][:,l,h]
direction = dataset_in.variables['DIRECTION'][:,l,h]
#pwr = pcurve.power(speed,direction)
# pdist will create a distribution for each timetep based on sampling
# n times from a normal distribution.
pdist = pcurve.power_dist(speed, direction, sstd=sstd,dstd=dstd,n=n, normalise=pnorm)
pmean = np.mean(pdist, axis=1)
pquants = scipy.stats.mstats.mquantiles(pdist, prob=quantiles/100.0,axis=1, alphap=0.5, betap=0.5)
pdata[:,l,h,0] = pmean
pdata[:,l,h,1:] = pquants[:,:]
logger.info('finished %s' % loc)
use_inds = np.array(use_locs)
logger.debug(use_inds)
logger.debug(pdata.shape)
logger.debug(pdata[:,use_inds,:,:].shape)
if dataset_out != dataset_in:
dataset_out.createDimension('time', None)
dataset_out.createVariable('time', 'float', ('time',))
dataset_out.variables['time'][:] = nctime[:]
dataset_out.variables['time'].units = nctime.units
dataset_out.variables['time'].calendar = nctime.calendar
dataset_out.createDimension('location', len(use_locs))
dataset_out.createDimension('loc_str_length', loc_str_len)
loc_data =np.array([list(l.ljust(loc_str_len, ' ')) for l in location])
dataset_out.createVariable('location', 'c', ('location', 'loc_str_length'))
dataset_out.variables['location'][:] = loc_data[use_inds,:]
dataset_out.createDimension('height', nheight)
dataset_out.createVariable('height', 'i', ('height',))
dataset_out.variables['height'][:] = height[:]
dataset_out.GRID_ID = dataset_in.GRID_ID
dataset_out.DX = dataset_in.DX
dataset_out.DY = dataset_in.DY
try:
dataset_out.variables['height'].units = height.units
except Exception:
logger.warn("height units missing")
pdata = pdata[:, use_inds, :, :]
for key in metadata.keys():
key = key.upper()
logger.debug(key)
dataset_out.setncattr(key,dataset_in.getncattr(key))
pavg = dataset_out.createVariable('POWER','f',('time','location','height'))
pavg.units = 'kW'
pavg.description = 'forecast power output'
pavg[:] = pdata[:,:,:,0]
for q, qval in enumerate(quantiles):
varname = 'POWER.P%02d' % qval
logger.debug("creating variable %s" % varname)
var = dataset_out.createVariable(varname,'f',('time','location','height'))
if pnorm:
var.units = 'ratio'
else:
var.units = 'kW'
var.description = 'forecast power output'
print pdata[:,:,:,q+1]
var[:] = pdata[:,:,:,q+1]
dataset_in.close()
if dataset_out!=dataset_in:
dataset_out.close()
def power(config):
"""Reads 'time series' from netcdf time series file, and adds power as a variable. """
if __name__ == "__main__":
logger = loghelper.create_logger(config)
else:
logger = loghelper.get_logger(config['log.name'])
# listify ensures they are returned as a list, even if it is one file
files = shared._listify(config['<files>'])
# Number of samples to use should be in here
# Whether to normalise power should be in here
start = config.get('start')
delay = config.get('delay')
cycles = shared._listify(config.get('cycles'))
pnorm = config.get('pnorm')
pdist = config.get('pdist')
sstd = config.get('sstd')
dstd = config.get('dstd')
pquants = config.get('pquants')
quantiles = np.array(pquants)
pcurve_dir = config.get('pcurve-dir')
ts_dir = config.get('tseries-dir')
out = config.get('out')
metadata = config.get('metadata')
basetime = start if start else datetime.datetime.today()
prior = shared._prior_time(basetime, delay=delay, hours=cycles)
logger.debug("using %s as a start time" % prior)
if not files:
logger.debug("no files specified, finding using options")
file_pattern = config.get('file-pattern')
if not file_pattern: raise ConfigError('either supply files or specify file-pattern')
expanded = substitute.sub_date(file_pattern, init_time=prior)
files = glob.glob(expanded)
print("hello")
logger.debug(files)
# if we get to this point and there are still no files, then we have a problem
if not files: raise IOError("no files found")
logger.debug("input files: ")
logger.debug(files)
for f in files:
logger.debug("\t%s" % f)
# if pdist
if pdist: n=pdist
#grid_id = config['grid_id']
out_pattern = config.get('out')
for tseries_file in files:
dataset_in = Dataset(tseries_file, 'a')
# Get dimensions
dims = dataset_in.dimensions
nreftime = len(dims['reftime'])
ntime = len(dims['leadtime'])
nloc = len(dims['location'])
nheight = len(dims['height'])
loc_str_len = len(dims['loc_str_length'])
# Get coordinate variables
reftime = dataset_in.variables['reftime']
leadtime = dataset_in.variables['leadtime']
validtime = nctools._valid_time(reftime, leadtime)
refdt = num2date(reftime[:], reftime.units)
power_file = substitute.sub_date(out, init_time=refdt[0])
logger.info('Estimating power from time series: %s ' % tseries_file)
logger.info('Writing power time series to: %s ' % power_file)
location = [''.join(l.filled(' ')).strip() for l in dataset_in.variables['location']]
height = dataset_in.variables['height']
if power_file == tseries_file:
dataset_out = dataset_in
else:
dataset_out = Dataset(power_file, 'w')
# Get number of quantiles
nq = len(quantiles)
pdata = np.ma.zeros((ntime,nloc,nheight,nq+1), np.float) # mean will be 1st value
use_locs = []
# loop through locations and look for power-curve file
for l,loc in enumerate(location):
pcurve_file = '%s/%s.csv' %(pcurve_dir, loc)
# mask power data if no power curve found for this park
if not os.path.exists(pcurve_file):
#logger.debug("Power curve: %s not found, skipping" % pcurve_file)
pdata[:,l,:,:] = np.ma.masked
continue
logger.info('Predicting power output for %s' % loc )
#
# Open power curve
#
use_locs.append(l)
pcurve = from_file(pcurve_file)
for h in range(nheight):
speed = dataset_in.variables['SPEED'][0,:,l,h]
direction = dataset_in.variables['DIRECTION'][0,:,l,h]
#pwr = pcurve.power(speed,direction)
# pdist will create a distribution for each timetep based on sampling
# n times from a normal distribution.
pdist = pcurve.power_dist(speed, direction, sstd=sstd,dstd=dstd,n=n, normalise=pnorm)
pmean = np.mean(pdist, axis=1)
pquants = scipy.stats.mstats.mquantiles(pdist, prob=quantiles/100.0,axis=1, alphap=0.5, betap=0.5)
pdata[:,l,h,0] = pmean
pdata[:,l,h,1:] = pquants[:,:]
#logger.info('finished %s' % loc)
use_inds = np.array(use_locs)
if dataset_out != dataset_in:
dataset_out.createDimension('reftime', None)
dataset_out.createVariable('reftime', 'float', ('reftime',))
dataset_out.variables['reftime'][:] = reftime[:]
dataset_out.variables['reftime'].units = reftime.units
dataset_out.variables['reftime'].calendar = reftime.calendar
dataset_out.variables['reftime'].long_name = reftime.long_name
dataset_out.variables['reftime'].standard_name = reftime.standard_name
dataset_out.createDimension('leadtime', len(leadtime))
dataset_out.createVariable('leadtime', 'int', ('leadtime',))
dataset_out.variables['leadtime'][:] = leadtime[:]
dataset_out.variables['leadtime'].units = leadtime.units
dataset_out.variables['leadtime'].long_name = leadtime.long_name
dataset_out.variables['leadtime'].standard_name = leadtime.standard_name
dataset_out.createDimension('location', len(use_locs))
dataset_out.createDimension('loc_str_length', loc_str_len)
loc_data =np.array([list(l.ljust(loc_str_len, ' ')) for l in location])
dataset_out.createVariable('location', 'c', ('location', 'loc_str_length'))
dataset_out.variables['location'][:] = loc_data[use_inds,:]
dataset_out.createDimension('height', nheight)
dataset_out.createVariable('height', 'i', ('height',))
dataset_out.variables['height'][:] = height[:]
dataset_out.GRID_ID = dataset_in.GRID_ID
dataset_out.DX = dataset_in.DX
dataset_out.DY = dataset_in.DY
try:
dataset_out.variables['height'].units = height.units
except Exception:
logger.warn("height units missing")
pdata = pdata[:, use_inds, :, :]
for key in metadata.keys():
key = key.upper()
dataset_out.setncattr(key,dataset_in.getncattr(key))
pavg = dataset_out.createVariable('POWER','f',('reftime','leadtime','location','height'))
pavg.units = 'kW'
pavg.description = 'forecast power output'
pavg[0,:,:,:] = pdata[:,:,:,0]
for q, qval in enumerate(quantiles):
varname = 'POWER.P%02d' % qval
var = dataset_out.createVariable(varname,'f',('reftime','leadtime','location','height'))
if pnorm:
var.units = 'ratio'
else:
var.units = 'kW'
var.description = 'forecast power output'
var[0,:,:,:] = pdata[:,:,:,q+1]
#logger.debug(dataset_out)
dataset_in.close()
if dataset_out!=dataset_in:
dataset_out.close()
landsea_mask.units = "" landsea_mask.scale_factor = 1.0 landsea_mask.add_offset = 0.0 landsea_mask.missing_value = -9999.0 landsea_mask.vmin = 0.0 landsea_mask.vmax = 0.0 landsea_mask.num_bins = 0 # What about extents and DX? # ... landsea.description = 'Land-Sea mask with binary 0/1 values' landsea.history = 'Created ' + time.ctime(time.time()) landsea.MAP_PROJECTION = "EQUIDISTANT CYLINDRICAL" landsea.SOUTH_WEST_CORNER_LAT = -90.0 landsea.SOUTH_WEST_CORNER_LON = -180.0 landsea.DX = 0.5 landsea.DY = 0.5 landsea_lons.units = 'east_west' landsea_lats.units = 'north_south' # Now put some data into out variables # lets do the lats and lons first # Raw data values for lat and lon lats = np.arange(-90, 90, 0.5) lons = np.arange(-180, 180, 0.5) # Now assign the raw data to netcdf coordinate variable landsea_lons[:] = lons landsea_lats[:] = lats
def power(config):
"""Reads 'time series' from netcdf time series file, and adds power as a variable. """
if __name__ == "__main__":
logger = loghelper.create_logger(config)
else:
logger = loghelper.get_logger(config['log.name'])
# Number of samples to use should be in here
# Whether to normalise power should be in here
pnorm = config['pnorm']
pdist = config['pdist']
sstd = config['sstd']
dstd = config['dstd']
pquants = config['pquants']
quantiles = np.array(pquants)
logger.debug(pnorm)
if pdist:
n = pdist
grid_id = config['grid_id']
init_time = config['init_time']
pcurve_dir = config['pcurve_dir']
ts_dir = config['tseries_dir']
tseries_file = expand(config['tseries_file'], config)
power_file = expand(config['power_file'], config)
logger.info('Estimating power from time series: %s ' % tseries_file)
logger.info('Writing power time series to: %s ' % power_file)
dataset_in = Dataset(tseries_file, 'a')
# Get dimensions
dims = dataset_in.dimensions
ntime = len(dims['time'])
nloc = len(dims['location'])
nheight = len(dims['height'])
loc_str_len = len(dims['loc_str_length'])
# Get coordinate variables
nctime = dataset_in.variables['time']
datetimes = netcdftime.num2date(nctime, nctime.units)
location = [
''.join(l.filled(' ')).strip()
for l in dataset_in.variables['location']
]
height = dataset_in.variables['height']
# Get attributes
metadata = config['metadata']
if power_file == tseries_file:
dataset_out = dataset_in
else:
dataset_out = Dataset(power_file, 'w')
# Get number of quantiles
nq = len(quantiles)
pdata = np.ma.zeros((ntime, nloc, nheight, nq + 1),
np.float) # mean will be 1st value
use_locs = []
for l, loc in enumerate(location):
pcurve_file = '%s/%s.csv' % (pcurve_dir, loc)
# mask power data if no power curve found for this park
if not os.path.exists(pcurve_file):
#logger.debug("Power curve: %s not found, skipping" % pcurve_file)
pdata[:, l, :, :] = np.ma.masked
continue
logger.info('Predicting power output for %s' % loc)
#
# Open power curve
#
use_locs.append(l)
pcurve = from_file(pcurve_file)
for h in range(nheight):
speed = dataset_in.variables['SPEED'][:, l, h]
direction = dataset_in.variables['DIRECTION'][:, l, h]
#pwr = pcurve.power(speed,direction)
# pdist will create a distribution for each timetep based on sampling
# n times from a normal distribution.
pdist = pcurve.power_dist(speed,
direction,
sstd=sstd,
dstd=dstd,
n=n,
normalise=pnorm)
pmean = np.mean(pdist, axis=1)
pquants = scipy.stats.mstats.mquantiles(pdist,
prob=quantiles / 100.0,
axis=1,
alphap=0.5,
betap=0.5)
pdata[:, l, h, 0] = pmean
pdata[:, l, h, 1:] = pquants[:, :]
logger.info('finished %s' % loc)
use_inds = np.array(use_locs)
logger.debug(use_inds)
logger.debug(pdata.shape)
logger.debug(pdata[:, use_inds, :, :].shape)
if dataset_out != dataset_in:
dataset_out.createDimension('time', None)
dataset_out.createVariable('time', 'float', ('time', ))
dataset_out.variables['time'][:] = nctime[:]
dataset_out.variables['time'].units = nctime.units
dataset_out.variables['time'].calendar = nctime.calendar
dataset_out.createDimension('location', len(use_locs))
dataset_out.createDimension('loc_str_length', loc_str_len)
loc_data = np.array(
[list(l.ljust(loc_str_len, ' ')) for l in location])
dataset_out.createVariable('location', 'c',
('location', 'loc_str_length'))
dataset_out.variables['location'][:] = loc_data[use_inds, :]
dataset_out.createDimension('height', nheight)
dataset_out.createVariable('height', 'i', ('height', ))
dataset_out.variables['height'][:] = height[:]
dataset_out.GRID_ID = dataset_in.GRID_ID
dataset_out.DX = dataset_in.DX
dataset_out.DY = dataset_in.DY
try:
dataset_out.variables['height'].units = height.units
except Exception:
logger.warn("height units missing")
pdata = pdata[:, use_inds, :, :]
for key in metadata.keys():
key = key.upper()
logger.debug(key)
dataset_out.setncattr(key, dataset_in.getncattr(key))
pavg = dataset_out.createVariable('POWER', 'f',
('time', 'location', 'height'))
pavg.units = 'kW'
pavg.description = 'forecast power output'
pavg[:] = pdata[:, :, :, 0]
for q, qval in enumerate(quantiles):
varname = 'POWER.P%02d' % qval
logger.debug("creating variable %s" % varname)
var = dataset_out.createVariable(varname, 'f',
('time', 'location', 'height'))
if pnorm:
var.units = 'ratio'
else:
var.units = 'kW'
var.description = 'forecast power output'
print pdata[:, :, :, q + 1]
var[:] = pdata[:, :, :, q + 1]
dataset_in.close()
if dataset_out != dataset_in:
dataset_out.close()
def power(config):
"""Reads 'time series' from netcdf time series file, and adds power as a variable. """
if __name__ == "__main__":
logger = loghelper.create_logger(config)
else:
logger = loghelper.get_logger(config['log.name'])
# listify ensures they are returned as a list, even if it is one file
files = shared._listify(config['<files>'])
# Number of samples to use should be in here
# Whether to normalise power should be in here
start = config.get('start')
delay = config.get('delay')
cycles = shared._listify(config.get('cycles'))
pnorm = config.get('pnorm')
pdist = config.get('pdist')
sstd = config.get('sstd')
dstd = config.get('dstd')
pquants = config.get('pquants')
quantiles = np.array(pquants)
pcurve_dir = config.get('pcurve-dir')
ts_dir = config.get('tseries-dir')
out = config.get('out')
metadata = config.get('metadata')
basetime = start if start else datetime.datetime.today()
prior = shared._prior_time(basetime, delay=delay, hours=cycles)
logger.debug("using %s as a start time" % prior)
if not files:
logger.debug("no files specified, finding using options")
file_pattern = config.get('file-pattern')
if not file_pattern:
raise ConfigError('either supply files or specify file-pattern')
expanded = substitute.sub_date(file_pattern, init_time=prior)
files = glob.glob(expanded)
print("hello")
logger.debug(files)
# if we get to this point and there are still no files, then we have a problem
if not files: raise IOError("no files found")
logger.debug("input files: ")
logger.debug(files)
for f in files:
logger.debug("\t%s" % f)
# if pdist
if pdist: n = pdist
#grid_id = config['grid_id']
out_pattern = config.get('out')
for tseries_file in files:
dataset_in = Dataset(tseries_file, 'a')
# Get dimensions
dims = dataset_in.dimensions
nreftime = len(dims['reftime'])
ntime = len(dims['leadtime'])
nloc = len(dims['location'])
nheight = len(dims['height'])
loc_str_len = len(dims['loc_str_length'])
# Get coordinate variables
reftime = dataset_in.variables['reftime']
leadtime = dataset_in.variables['leadtime']
validtime = nctools._valid_time(reftime, leadtime)
refdt = num2date(reftime[:], reftime.units)
power_file = substitute.sub_date(out, init_time=refdt[0])
logger.info('Estimating power from time series: %s ' % tseries_file)
logger.info('Writing power time series to: %s ' % power_file)
location = [
''.join(l.filled(' ')).strip()
for l in dataset_in.variables['location']
]
height = dataset_in.variables['height']
if power_file == tseries_file:
dataset_out = dataset_in
else:
dataset_out = Dataset(power_file, 'w')
# Get number of quantiles
nq = len(quantiles)
pdata = np.ma.zeros((ntime, nloc, nheight, nq + 1),
np.float) # mean will be 1st value
use_locs = []
# loop through locations and look for power-curve file
for l, loc in enumerate(location):
pcurve_file = '%s/%s.csv' % (pcurve_dir, loc)
# mask power data if no power curve found for this park
if not os.path.exists(pcurve_file):
#logger.debug("Power curve: %s not found, skipping" % pcurve_file)
pdata[:, l, :, :] = np.ma.masked
continue
logger.info('Predicting power output for %s' % loc)
#
# Open power curve
#
use_locs.append(l)
pcurve = from_file(pcurve_file)
for h in range(nheight):
speed = dataset_in.variables['SPEED'][0, :, l, h]
direction = dataset_in.variables['DIRECTION'][0, :, l, h]
#pwr = pcurve.power(speed,direction)
# pdist will create a distribution for each timetep based on sampling
# n times from a normal distribution.
pdist = pcurve.power_dist(speed,
direction,
sstd=sstd,
dstd=dstd,
n=n,
normalise=pnorm)
pmean = np.mean(pdist, axis=1)
pquants = scipy.stats.mstats.mquantiles(pdist,
prob=quantiles / 100.0,
axis=1,
alphap=0.5,
betap=0.5)
pdata[:, l, h, 0] = pmean
pdata[:, l, h, 1:] = pquants[:, :]
#logger.info('finished %s' % loc)
use_inds = np.array(use_locs)
if dataset_out != dataset_in:
dataset_out.createDimension('reftime', None)
dataset_out.createVariable('reftime', 'float', ('reftime', ))
dataset_out.variables['reftime'][:] = reftime[:]
dataset_out.variables['reftime'].units = reftime.units
dataset_out.variables['reftime'].calendar = reftime.calendar
dataset_out.variables['reftime'].long_name = reftime.long_name
dataset_out.variables[
'reftime'].standard_name = reftime.standard_name
dataset_out.createDimension('leadtime', len(leadtime))
dataset_out.createVariable('leadtime', 'int', ('leadtime', ))
dataset_out.variables['leadtime'][:] = leadtime[:]
dataset_out.variables['leadtime'].units = leadtime.units
dataset_out.variables['leadtime'].long_name = leadtime.long_name
dataset_out.variables[
'leadtime'].standard_name = leadtime.standard_name
dataset_out.createDimension('location', len(use_locs))
dataset_out.createDimension('loc_str_length', loc_str_len)
loc_data = np.array(
[list(l.ljust(loc_str_len, ' ')) for l in location])
dataset_out.createVariable('location', 'c',
('location', 'loc_str_length'))
dataset_out.variables['location'][:] = loc_data[use_inds, :]
dataset_out.createDimension('height', nheight)
dataset_out.createVariable('height', 'i', ('height', ))
dataset_out.variables['height'][:] = height[:]
dataset_out.GRID_ID = dataset_in.GRID_ID
dataset_out.DX = dataset_in.DX
dataset_out.DY = dataset_in.DY
try:
dataset_out.variables['height'].units = height.units
except Exception:
logger.warn("height units missing")
pdata = pdata[:, use_inds, :, :]
for key in metadata.keys():
key = key.upper()
dataset_out.setncattr(key, dataset_in.getncattr(key))
pavg = dataset_out.createVariable(
'POWER', 'f', ('reftime', 'leadtime', 'location', 'height'))
pavg.units = 'kW'
pavg.description = 'forecast power output'
pavg[0, :, :, :] = pdata[:, :, :, 0]
for q, qval in enumerate(quantiles):
varname = 'POWER.P%02d' % qval
var = dataset_out.createVariable(
varname, 'f', ('reftime', 'leadtime', 'location', 'height'))
if pnorm:
var.units = 'ratio'
else:
var.units = 'kW'
var.description = 'forecast power output'
var[0, :, :, :] = pdata[:, :, :, q + 1]
#logger.debug(dataset_out)
dataset_in.close()
if dataset_out != dataset_in:
dataset_out.close()