fig.set_size_inches(10, 5)
ax = fig.add_subplot(1, 2, 1)
state['air_temperature'].mean(dim='lon').plot.contourf(ax=ax, levels=16)
ax.set_title('Temperature')
ax = fig.add_subplot(1, 2, 2)
state['eastward_wind'].mean(dim='lon').plot.contourf(ax=ax, levels=16)
ax.set_title('Zonal Wind')
plt.suptitle('Time: ' + str(state['time']))
model_time_step = timedelta(seconds=600)
monitor = PlotFunctionMonitor(plot_function)
grid = climt.get_grid(nx=128, ny=62)
held_suarez = climt.HeldSuarez()
dycore = climt.GFSDynamicalCore([held_suarez])
my_state = climt.get_default_state([dycore], grid_state=grid)
my_state['eastward_wind'].values[:] = np.random.randn(
*my_state['eastward_wind'].shape)
for i in range(10000):
diag, output = dycore(my_state, model_time_step)
if (my_state['time'].hour % 2 == 0 and my_state['time'].minute == 0):
print('max. zonal wind: ', np.amax(my_state['eastward_wind'].values))
monitor.store(my_state)
def main():
# ============ Adjustable Variables ============
# Integration Options
dt = timedelta(minutes=15) # timestep
duration = '48_00:00' # run duration ('<days>_<hours>:<mins>')t
linearized = True
ncout_freq = 6 # netcdf write frequency (hours)
plot_freq = 6 # plot Monitor call frequency (hours)
ntrunc = 42 # triangular truncation for spharm (e.g., 21 --> T21)
# Diffusion Options
diff_on = True # Use diffusion?
k = 2.338e16 # Diffusion coefficient for del^4 hyperdiffusion
# Forcing Options
forcing_on = True # Apply vort. tendency forcing?
damp_ts = 14.7 # Damping timescale (in days)
# I/O Options
ncoutfile = os.path.join(os.path.dirname(__file__), 'sardeshmukh88.nc')
append_nc = False # Append to an existing netCDF file?
# ==============================================
start = time()
# Get the initial state
state = super_rotation(linearized=linearized, ntrunc=ntrunc)
# Set up the Timestepper with the desired Prognostics
if linearized:
dynamics_prog = LinearizedDynamics(ntrunc=ntrunc)
diffusion_prog = LinearizedDiffusion(k=k, ntrunc=ntrunc)
damping_prog = LinearizedDamping(tau=damp_ts)
else:
dynamics_prog = NonlinearDynamics(ntrunc=ntrunc)
diffusion_prog = NonlinearDiffusion(k=k, ntrunc=ntrunc)
damping_prog = NonlinearDamping(tau=damp_ts)
prognostics = [TendencyInDiagnosticsWrapper(dynamics_prog, 'dynamics')]
if diff_on:
prognostics.append(TendencyInDiagnosticsWrapper(diffusion_prog, 'diffusion'))
if forcing_on:
# Get our suptropical RWS forcing (from equatorial divergence)
rws, rlat, rlon = rws_from_tropical_divergence(state)
prognostics.append(TendencyInDiagnosticsWrapper(Forcing.from_numpy_array(rws, rlat, rlon, ntrunc=ntrunc,
linearized=linearized), 'forcing'))
prognostics.append(TendencyInDiagnosticsWrapper(damping_prog, 'damping'))
stepper = Leapfrog(prognostics)
# Create Monitors for plotting & storing data
plt_monitor = PlotFunctionMonitor(debug_plots.fourpanel)
if os.path.isfile(ncoutfile) and not append_nc:
os.remove(ncoutfile)
aliases = get_component_aliases(*prognostics)
nc_monitor = NetCDFMonitor(ncoutfile, write_on_store=True, aliases=aliases)
# Figure out the end date of this run
d, h, m = re.split('[_:]', duration)
end_date = state['time'] + timedelta(days=int(d), hours=int(h), minutes=int(m))
# Begin the integration loop
idate = state['time']
while state['time'] <= end_date:
# Get the state at the next timestep using our Timestepper
diagnostics, next_state = stepper(state, dt)
# Add any calculated diagnostics to our current state
state.update(diagnostics)
# Write state to netCDF every <ncout_freq> hours
fhour = (state['time'] - idate).days*24 + (state['time'] - idate).seconds/3600
if fhour % ncout_freq == 0:
print(state['time'])
nc_monitor.store(state)
# Make plot(s) every <plot_freq> hours
if fhour % plot_freq == 0:
plt_monitor.store(state)
# Advance the state to the next timestep
next_state['time'] = state['time'] + dt
state = next_state
print('TOTAL INTEGRATION TIME: {:.02f} min\n'.format((time()-start)/60.))
from datetime import datetime, timedelta
def my_plot_function(fig, state):
ax = fig.add_subplot(1, 1, 1)
ax.set_xlabel('longitude')
ax.set_ylabel('latitude')
ax.set_title('Lowest model level air temperature (K)')
im = ax.pcolormesh(
state['air_temperature'].to_units('degK').values[0, :, :],
vmin=260.,
vmax=310.)
cbar = fig.colorbar(im)
plot_monitor = PlotFunctionMonitor(my_plot_function)
state = get_initial_state(nx=256, ny=128, nz=64)
state['time'] = datetime(2000, 1, 1)
physics_stepper = AdamsBashforth(
UpdateFrequencyWrapper(Radiation(), timedelta(hours=2)),
BoundaryLayer(),
DeepConvection(),
)
implicit_dynamics = ImplicitDynamics()
timestep = timedelta(minutes=30)
while state['time'] < datetime(2010, 1, 1):
physics_diagnostics, state_after_physics = physics_stepper(state)
dynamics_diagnostics, next_state = implicit_dynamics(state_after_physics)
# Add diffusion
if diff_on:
prognostics.append(
TendencyInDiagnosticsWrapper(diffusion_prog, 'diffusion'))
# Add our forcing
if forcing_on:
# Get our suptropical RWS forcing (from equatorial divergence)
prognostics.append(TendencyInDiagnosticsWrapper(forcing_prog, 'forcing'))
prognostics.append(TendencyInDiagnosticsWrapper(damping_prog, 'damping'))
# Create Timestepper
stepper = Leapfrog(prognostics)
# Create Monitors for plotting & storing data
plt_monitor = PlotFunctionMonitor(debug_plots.fourpanel)
if os.path.isfile(ncoutfile) and not append_nc:
os.remove(ncoutfile)
aliases = get_component_aliases(*prognostics)
nc_monitor = NetCDFMonitor(ncoutfile, write_on_store=True, aliases=aliases)
# Figure out the end date of this run
d, h, m = re.split('[_:]', duration)
end_date = state['time'] + timedelta(days=int(d), hours=int(h), minutes=int(m))
# Begin the integration loop
idate = state['time']
while state['time'] <= end_date:
# Get the state at the next timestep using our Timestepper
diagnostics, next_state = stepper(state, dt)
Created on Tue Nov 14 23:35:51 2017
@author: zifan
"""
import climt
from sympl import PlotFunctionMonitor
def plot_function(fig, state):
ax = fig.add_subplot(1, 1, 1)
state['surface_air_pressure'].transpose().plot.contourf(ax=ax, levels=16)
monitor = PlotFunctionMonitor(plot_function)
dycore = climt.GfsDynamicalCore(number_of_longitudes=198,
number_of_latitudes=94,
dry_pressure=1e5)
dcmip = climt.DcmipInitialConditions()
my_state = climt.get_default_state(
[dycore],
x=dycore.grid_definition['x'],
y=dycore.grid_definition['y'],
mid_levels=dycore.grid_definition['mid_levels'],
interface_levels=dycore.grid_definition['interface_levels'])
my_state['surface_air_pressure'].values[:] = 1e5
dycore(my_state)