def load_particles(particle_dataset):
"""
Load a netCDF4 dataset of particles and return it. Close the dataset when
done.
:param dataset: open readable NetCDF dataset
:type dataset: netCDF.Dataset
:return: (
x coordinates of particles (longitude),
y coordinates of particles (latitude),
z coordinates of particles in meters,
ages of particles in seconds,
times converted to datetime objects,
original time units,
original time base date,
)
:rtype: tuple of (
TxN array,
TxN array,
TxN array,
TxN array,
T array of datetime objects,
str,
array,
)
Where T is the number of time steps and N is the number of particles.
"""
ncparticle = ncp.nc_particle_file(particle_dataset)
NumLE = ncparticle.particle_count[:]
times = ncparticle.times[:]
NumTimeSteps = len(times)
data = ncparticle.get_all_timesteps(
['longitude', 'latitude', 'depth', 'mass', 'age'])
#print data.keys()
#Check whether the number of particles remains constant in each time-step
#(instaneous release) or changes over time (continuous release) (Zelenke).
if len(np.unique(NumLE)) == 1:
#Reshape LE time-series vectors into rectangular matricies (NumTimeSteps x NumLE).
lon = data['longitude'].reshape((NumTimeSteps, -1))
lat = data['latitude'].reshape((NumTimeSteps, -1))
depth = data['depth'].reshape((NumTimeSteps, -1))
mass = data['mass'].reshape((NumTimeSteps, -1))
age = data['age'].reshape((NumTimeSteps, -1))
else:
#Put varying number of LEs in each time-step into uniformly sized vectors
#to allow for resizing into rectangular matricies (NumTimeSteps x NumLE).
MaxNumLE = np.max(NumLE)
lons = np.ones(MaxNumLE * NumTimeSteps,
dtype=data['longitude'].dtype) * np.array(
np.NaN, dtype=data['longitude'].dtype
) #Preallocate vector for subsequent loop.
lats = np.ones(MaxNumLE * NumTimeSteps,
dtype=data['latitude'].dtype) * np.array(
np.NaN, dtype=data['latitude'].dtype)
depths = np.ones(MaxNumLE * NumTimeSteps,
dtype=data['depth'].dtype) * np.array(
np.NaN, dtype=data['depth'].dtype)
masses = np.ones(MaxNumLE * NumTimeSteps,
dtype=data['mass'].dtype) * np.array(
np.NaN, dtype=data['mass'].dtype)
ages = np.copy(masses)
CumSumLE = np.append(0, np.cumsum(NumLE))
for i in range(NumTimeSteps):
#Put however many LEs there were in each time-step into that time-step's (uniformly sized) portion of the vector.
#Leave the rest of the chunk populated with NaNs, if there aren't enough LEs in that particular time-step to fill-up the chunk.
lons[i * MaxNumLE:(i * MaxNumLE) +
NumLE[i]] = data['longitude'][CumSumLE[i]:CumSumLE[i + 1]]
lats[i * MaxNumLE:(i * MaxNumLE) +
NumLE[i]] = data['latitude'][CumSumLE[i]:CumSumLE[i + 1]]
depths[i * MaxNumLE:(i * MaxNumLE) +
NumLE[i]] = data['depth'][CumSumLE[i]:CumSumLE[i + 1]]
masses[i * MaxNumLE:(i * MaxNumLE) +
NumLE[i]] = data['mass'][CumSumLE[i]:CumSumLE[i + 1]]
ages[i * MaxNumLE:(i * MaxNumLE) +
NumLE[i]] = data['age'][CumSumLE[i]:CumSumLE[i + 1]]
#Reshape (NumTimeSteps x NumLE).
lon = lons.reshape((NumTimeSteps, -1))
lat = lats.reshape((NumTimeSteps, -1))
depth = depths.reshape((NumTimeSteps, -1))
mass = masses.reshape((NumTimeSteps, -1))
age = ages.reshape((NumTimeSteps, -1))
time_units = ncparticle.time_units
mass_units = ncparticle.mass_units #Included units of mass specified in the NetCDF file as an output variable (Zelenke).
return (
lon,
lat,
depth,
age,
times,
time_units,
mass,
mass_units,
)
def load_particles(particle_dataset):
"""
Load a netCDF4 dataset of particles and return it. Close the dataset when
done.
:param dataset: open readable NetCDF dataset
:type dataset: netCDF.Dataset
:return: (
x coordinates of particles (longitude),
y coordinates of particles (latitude),
z coordinates of particles in meters,
ages of particles in seconds,
times converted to datetime objects,
original time units,
original time base date,
)
:rtype: tuple of (
TxN array,
TxN array,
TxN array,
TxN array,
T array of datetime objects,
str,
array,
)
Where T is the number of time steps and N is the number of particles.
"""
ncparticle = ncp.nc_particle_file(particle_dataset)
NumLE = ncparticle.particle_count[:]
times = ncparticle.times[:]
NumTimeSteps = len(times)
data = ncparticle.get_all_timesteps(['longitude','latitude','depth','mass','age'])
#print data.keys()
#Check whether the number of particles remains constant in each time-step
#(instaneous release) or changes over time (continuous release) (Zelenke).
if len( np.unique(NumLE) ) == 1:
#Reshape LE time-series vectors into rectangular matricies (NumTimeSteps x NumLE).
lon = data['longitude'].reshape( (NumTimeSteps, -1) )
lat = data['latitude'].reshape( (NumTimeSteps, -1) )
depth = data['depth'].reshape( (NumTimeSteps, -1) )
mass = data['mass'].reshape( (NumTimeSteps, -1) )
age = data['age'].reshape( (NumTimeSteps, -1) )
else:
#Put varying number of LEs in each time-step into uniformly sized vectors
#to allow for resizing into rectangular matricies (NumTimeSteps x NumLE).
MaxNumLE = np.max(NumLE)
lons = np.ones( MaxNumLE*NumTimeSteps, dtype = data['longitude'].dtype ) * np.array(np.NaN, dtype = data['longitude'].dtype ) #Preallocate vector for subsequent loop.
lats = np.ones( MaxNumLE*NumTimeSteps, dtype = data['latitude'].dtype ) * np.array(np.NaN, dtype = data['latitude'].dtype )
depths = np.ones( MaxNumLE*NumTimeSteps, dtype = data['depth'].dtype ) * np.array(np.NaN, dtype = data['depth'].dtype )
masses = np.ones( MaxNumLE*NumTimeSteps, dtype = data['mass'].dtype ) * np.array(np.NaN, dtype = data['mass'].dtype )
ages = np.copy(masses)
CumSumLE = np.append(0,np.cumsum(NumLE))
for i in range(NumTimeSteps):
#Put however many LEs there were in each time-step into that time-step's (uniformly sized) portion of the vector.
#Leave the rest of the chunk populated with NaNs, if there aren't enough LEs in that particular time-step to fill-up the chunk.
lons[ i*MaxNumLE : (i*MaxNumLE)+NumLE[i] ] = data['longitude'][ CumSumLE[i] : CumSumLE[i+1] ]
lats[ i*MaxNumLE : (i*MaxNumLE)+NumLE[i] ] = data['latitude'][ CumSumLE[i] : CumSumLE[i+1] ]
depths[ i*MaxNumLE : (i*MaxNumLE)+NumLE[i] ] = data['depth'][ CumSumLE[i] : CumSumLE[i+1] ]
masses[ i*MaxNumLE : (i*MaxNumLE)+NumLE[i] ] = data['mass'][ CumSumLE[i] : CumSumLE[i+1] ]
ages[ i*MaxNumLE : (i*MaxNumLE)+NumLE[i] ] = data['age'][ CumSumLE[i] : CumSumLE[i+1] ]
#Reshape (NumTimeSteps x NumLE).
lon=lons.reshape( (NumTimeSteps, -1) )
lat = lats.reshape( (NumTimeSteps, -1) )
depth = depths.reshape( (NumTimeSteps, -1) )
mass = masses.reshape( (NumTimeSteps, -1) )
age = ages.reshape( (NumTimeSteps, -1) )
time_units = ncparticle.time_units
mass_units = ncparticle.mass_units #Included units of mass specified in the NetCDF file as an output variable (Zelenke).
return (
lon,
lat,
depth,
age,
times,
time_units,
mass,
mass_units,
)
def CompThicknessCube(FileList, OutputTimes, Grid, Weather=None):
"""
CompThicknessCube computes the average thickness of the oil over
each receptor site. It only works for grid receptors.
Filelist is a list of netcdf file names: one for each trajectory
OutputTimes is a sequence of output times, in hours, from the beginning of the run.
Grid is a Grid object, specifying the grid parameters
If Weather is not None it must be a tap_comp_volume.weather_curve object
If Weather is None, then there is no change in volume.
"""
## read the header of the first trajectory file: It is assumed that
## the others will match..no check is made to assure this, but it
## will crash if anything is very wrong.
#(junk, junk, HeaderData, junk) = ReadTrajectory(FileList[0],2)
# read the trajectory data from the first netcdf file
#print "**************"
#print "getting header info from file:", FileList[0]
print "nc_particles module:", nc_particles.__file__
traj_file = nc_particles.nc_particle_file(FileList[0])
if traj_file.get_units('age') != 'seconds':
raise ValueError(
"particle age units in netcdf file must be in seconds")
NumTimesteps = traj_file.num_times
MaxNumLEs = traj_file.particle_count[:].max()
TimeStep = traj_file.times[1] - traj_file.times[
0] # assume constant timestep!
traj_file.close()
TimeStepHours = TimeStep.total_seconds() / 3600.00
## OutputTimes should already be in hours
OutputSteps = (np.array([0] + OutputTimes) / TimeStepHours).astype(
np.int32) # in integer units of time step
# Allocate the Cube
NumSpills, NumSites, NumTimes = len(FileList), Grid.num_cells, len(
OutputTimes)
## fixme: need to make this a float cube!
Cube = np.zeros((NumTimes, NumSites, NumSpills), np.float32)
start = time.time() # just for timing how long it takes to run
## Loop through each individual trajectory
for SpillNum in range(NumSpills):
#print "computing spill number %i"%(SpillNum,)
# read new trajectory file:
#print "working with file:", FileList[SpillNum]
traj_file = nc_particles.nc_particle_file(FileList[SpillNum])
VolTable = np.zeros(
(NumSites),
np.float32) # this will store the Maximum volume in each grid box.
## Step through the Cube output time steps
for step in xrange(len(OutputSteps) - 1):
## step through the Trajectory time steps between each Cube Timestep
for t in xrange(OutputSteps[step], OutputSteps[step + 1]):
LE_lat = traj_file.get_timestep_single_var(t, 'latitude')
LE_long = traj_file.get_timestep_single_var(t, 'longitude')
LE_positions = np.column_stack((LE_long, LE_lat))
NumLEs = LE_positions.shape[0]
LE_age = traj_file.get_timestep_single_var(t, 'age').astype(
np.float32) / 3600.00 # age needs to be in hours
#print "age:", LE_age
#NOTE: for TAP -- we assume that are the particles have unit mass at the start
# so we don't read it from the file
LE_mass = np.ones((NumLEs, ), dtype=np.float32)
#print "before"
#print LE_mass
if Weather:
#print "weathering the LEs"
LE_mass = Weather.weather(LE_mass, LE_age)
flags = traj_file.get_timestep_single_var(t, 'flag').astype(
np.uint8)
Vol = comp_volume(LE_positions, LE_mass, flags, Grid)
# keep the largest volume computed between output timesteps
VolTable = np.maximum(Vol.flat, VolTable)
## put the max volume in the Cube at this Cube time step
#Cube[step,:,SpillNum] = transform(VolTable, MaxNumLEs)
Cube[step, :, SpillNum] = VolTable
traj_file.close()
#print "cube took %s seconds to generate"%(time.time() - start)
return Cube
def CompThicknessCube(FileList, OutputTimes, Grid, Weather=None):
"""
CompThicknessCube computes the average thickness of the oil over
each receptor site. It only works for grid receptors.
Filelist is a list of netcdf file names: one for each trajectory
OutputTimes is a sequence of output times, in hours, from the beginning of the run.
Grid is a Grid object, specifying the grid parameters
If Weather is not None it must be a tap_comp_volume.weather_curve object
If Weather is None, then there is no change in volume.
"""
## read the header of the first trajectory file: It is assumed that
## the others will match..no check is made to assure this, but it
## will crash if anything is very wrong.
#(junk, junk, HeaderData, junk) = ReadTrajectory(FileList[0],2)
# read the trajectory data from the first netcdf file
#print "**************"
#print "getting header info from file:", FileList[0]
print "nc_particles module:", nc_particles.__file__
traj_file = nc_particles.nc_particle_file(FileList[0])
if traj_file.get_units('age') != 'seconds':
raise ValueError("particle age units in netcdf file must be in seconds")
NumTimesteps = traj_file.num_times
MaxNumLEs = traj_file.particle_count[:].max()
TimeStep = traj_file.times[1] - traj_file.times[0] # assume constant timestep!
traj_file.close()
TimeStepHours = TimeStep.total_seconds() / 3600.00
## OutputTimes should already be in hours
OutputSteps = (np.array([0] + OutputTimes) / TimeStepHours).astype(np.int32) # in integer units of time step
# Allocate the Cube
NumSpills, NumSites, NumTimes = len(FileList), Grid.num_cells, len(OutputTimes)
## fixme: need to make this a float cube!
Cube = np.zeros((NumTimes,NumSites,NumSpills), np.float32)
start = time.time() # just for timing how long it takes to run
## Loop through each individual trajectory
for SpillNum in range(NumSpills):
#print "computing spill number %i"%(SpillNum,)
# read new trajectory file:
#print "working with file:", FileList[SpillNum]
traj_file = nc_particles.nc_particle_file(FileList[SpillNum])
VolTable = np.zeros((NumSites), np.float32) # this will store the Maximum volume in each grid box.
## Step through the Cube output time steps
for step in xrange( len(OutputSteps) - 1 ):
## step through the Trajectory time steps between each Cube Timestep
for t in xrange(OutputSteps[step], OutputSteps[step+1]):
LE_lat = traj_file.get_timestep_single_var(t, 'latitude')
LE_long = traj_file.get_timestep_single_var(t, 'longitude')
LE_positions = np.column_stack((LE_long, LE_lat))
NumLEs = LE_positions.shape[0]
LE_age = traj_file.get_timestep_single_var(t, 'age').astype(np.float32) / 3600.00 # age needs to be in hours
#print "age:", LE_age
#NOTE: for TAP -- we assume that are the particles have unit mass at the start
# so we don't read it from the file
LE_mass = np.ones((NumLEs,), dtype = np.float32)
#print "before"
#print LE_mass
if Weather:
#print "weathering the LEs"
LE_mass = Weather.weather(LE_mass, LE_age)
flags = traj_file.get_timestep_single_var(t, 'flag').astype(np.uint8)
Vol = comp_volume(LE_positions, LE_mass, flags, Grid)
# keep the largest volume computed between output timesteps
VolTable = np.maximum(Vol.flat, VolTable)
## put the max volume in the Cube at this Cube time step
#Cube[step,:,SpillNum] = transform(VolTable, MaxNumLEs)
Cube[step,:,SpillNum] = VolTable
traj_file.close()
#print "cube took %s seconds to generate"%(time.time() - start)
return Cube