def test_vertical_zero():
"""
checks that there is no vertical movement
"""
mv = RandomVerticalMover(mixed_layer_depth=10.0) # mostly defaults
num_elements = 100
sc = sample_sc_release(
num_elements=num_elements,
start_pos=(0.0, 0.0, 0.0),
release_time=model_time,
)
# set z positions:
sc['positions'][:, 2] = np.linspace(0, 50, num_elements)
print sc['positions']
mv.vertical_diffusion_coef_above_ml = 0.0
mv.vertical_diffusion_coef_below_ml = 0.0
delta = mv.get_move(
sc,
time_step,
model_time,
)
print delta
assert not np.alltrue(delta[:, 2] == 0.0)
def test_horizontal_zero():
"""
checks that there is no horizontal movement
"""
mv = RandomVerticalMover() # all defaults
num_elements = 100
sc = sample_sc_release(
num_elements=num_elements,
start_pos=(0.0, 0.0, 0.0),
release_time=model_time,
)
# set z positions:
sc['positions'][:, 2] = np.linspace(0, 50, num_elements)
mv.horizontal_diffusion_coef_above_ml = 0
mv.horizontal_diffusion_coef_below_ml = 0
delta = mv.get_move(
sc,
time_step,
model_time,
)
print delta
assert np.alltrue(delta[:, 0:2] == 0.0)
def test_vertical_zero():
"""
checks that there is no vertical movement
"""
mv = RandomVerticalMover() # all defaults
num_elements = 100
sc = sample_sc_release(num_elements=num_elements,
start_pos=(0.0, 0.0, 0.0),
release_time=model_time,
)
# set z positions:
sc['positions'][:, 2] = np.linspace(0, 50, num_elements)
mv.vertical_diffusion_coef_above_ml = 0;
mv.vertical_diffusion_coef_below_ml = 0;
delta = mv.get_move(sc,
time_step,
model_time,
)
print delta
assert np.alltrue(delta[:, 2] == 0.0)
def test_init():
"""
test the varios ways it can be initilized
"""
mv = RandomVerticalMover()
mv = RandomVerticalMover(vertical_diffusion_coef_above_ml=10.0) # cm^2/s
mv = RandomVerticalMover(vertical_diffusion_coef_below_ml=2.0) # cm^2/s
mv = RandomVerticalMover(mixed_layer_depth=20) # m
assert True
def test_mixed_layer2():
"""
tests the top layer
elements should end up fairly evenly distributed by the end of along run
"""
D_mixed = 10.0 # cm^2/s
mixed_layer_depth = 10.0 # m
time_step = 60
total_time = 6000
num_elements = 1000
num_timesteps = total_time // time_step
mv = RandomVerticalMover(
vertical_diffusion_coef_above_ml=D_mixed,
vertical_diffusion_coef_below_ml=0.0,
mixed_layer_depth=mixed_layer_depth,
) # m
sc = sample_sc_release(
num_elements=num_elements,
start_pos=(0.0, 0.0, 0.0),
release_time=model_time,
)
# re-set z positions:
sc['positions'][:, 2] = 5.0 # middle of the layer
# call get_move a bunch of times
for i in range(num_timesteps):
# print "positions:\n", sc['positions']
delta = mv.get_move(
sc,
time_step,
model_time,
)
# print "delta:\n", delta
sc['positions'] += delta
# print sc['positions']
# expected mean
# exp_mean = 5.0 # middle of layer
# expected variance:
exp_var = mixed_layer_depth**2 / 12.0
var = sc['positions'][:, 2].var()
print "expected_var:", exp_var, "var:", var
assert np.allclose(exp_var, var, rtol=0.18)
def test_mixed_layer():
"""
tests the top layer
elements should vertically spread according to the diffusion coef of the
top layer. This version uses a really thick mixed layer to stay away from
boundary effects
"""
D_mixed = 10.0 # cm^2/s
time_step = 60
total_time = 600
num_elements = 1000
num_timesteps = total_time // time_step
mv = RandomVerticalMover(
vertical_diffusion_coef_above_ml=D_mixed,
mixed_layer_depth=1000, # HUGE to avoid surface effects.
) # m
sc = sample_sc_release(
num_elements=num_elements,
start_pos=(0.0, 0.0, 0.0),
release_time=model_time,
)
# re-set z positions:
sc['positions'][:, 2] = 500.0 # far from the top
# call get_move a bunch of times
for i in range(num_timesteps):
# print "positions:\n", sc['positions']
delta = mv.get_move(
sc,
time_step,
model_time,
)
# print "delta:\n", delta
sc['positions'] += delta
# print sc['positions']
# expected variance:
exp_var = 2 * num_timesteps * time_step * D_mixed # in cm^2?
exp_var /= 10**4 # convert to m
var = sc['positions'][:, 2].var()
print "expected_var:", exp_var, "var:", var
assert np.allclose(exp_var, var, rtol=0.1)
def test_mixed_layer2():
"""
tests the top layer
elements should end up fairly evenly distributed by the end of along run
"""
D_mixed = 10.0 # cm^2/s
mixed_layer_depth = 10.0 # m
time_step = 60
total_time = 6000
num_elements = 1000
num_timesteps = total_time // time_step
mv = RandomVerticalMover(vertical_diffusion_coef_above_ml=D_mixed,
vertical_diffusion_coef_below_ml=0.0,
mixed_layer_depth=mixed_layer_depth,
) # m
sc = sample_sc_release(num_elements=num_elements,
start_pos=(0.0, 0.0, 0.0),
release_time=model_time,
)
# re-set z positions:
sc['positions'][:, 2] = 5.0 # middle of the layer
# call get_move a bunch of times
for i in range(num_timesteps):
# print "positions:\n", sc['positions']
delta = mv.get_move(sc,
time_step,
model_time,
)
# print "delta:\n", delta
sc['positions'] += delta
# print sc['positions']
# expected mean
exp_mean = 5.0 # middle of layer
# expected variance:
exp_var = mixed_layer_depth**2 / 12.0
var = sc['positions'][:,2].var()
print "expected_var:", exp_var, "var:", var
assert np.allclose(exp_var, var, rtol=0.18)
def test_mixed_layer():
"""
tests the top layer
elements should vertically spread according to the diffusion coef of the top layer
this version uses a really thick mixed layer to stay away from boundary effects
"""
D_mixed = 10.0 # cm^2/s
time_step = 60
total_time = 600
num_elements = 1000
num_timesteps = total_time // time_step
mv = RandomVerticalMover(vertical_diffusion_coef_above_ml=D_mixed,
mixed_layer_depth=1000, # HUGE to avoid surface effects.
) # m
sc = sample_sc_release(num_elements=num_elements,
start_pos=(0.0, 0.0, 0.0),
release_time=model_time,
)
# re-set z positions:
sc['positions'][:, 2] = 500.0 # far from the top
# call get_move a bunch of times
for i in range(num_timesteps):
# print "positions:\n", sc['positions']
delta = mv.get_move(sc,
time_step,
model_time,
)
# print "delta:\n", delta
sc['positions'] += delta
# print sc['positions']
# expected variance:
exp_var = 2 * num_timesteps*time_step * D_mixed # in cm^2?
exp_var /= 10**4 # convert to m
var = sc['positions'][:,2].var()
print "expected_var:", exp_var, "var:", var
assert np.allclose(exp_var, var, rtol = 0.1)
def test_bottom_layer():
"""
tests the bottom layer
elements should vertically spread according to the diffusion coef of the bottom layer
"""
D_lower = .11 #m^2/s (or cm^2/s?)
time_step = 60
total_time = 60000
num_elements = 100
num_timesteps = total_time // time_step
mv = RandomVerticalMover(vertical_diffusion_coef_below_ml=D_lower) # m
sc = sample_sc_release(
num_elements=num_elements,
start_pos=(0.0, 0.0, 0.0),
release_time=model_time,
)
# re-set z positions:
sc['positions'][:, 2] = 1000.0 # far from the top
# sc['positions'][ :num_elements/2 , 2] = 5.0 # near top
# sc['positions'][ num_elements/2:, 2] = 50.0 # down deep
# call get_move a bunch of times
for i in range(num_timesteps):
# print "positions:\n", sc['positions']
delta = mv.get_move(
sc,
time_step,
model_time,
)
# print "delta:\n", delta
sc['positions'] += delta
# print sc['positions']
# expected variance:
exp_var = 2 * num_timesteps * time_step * D_lower # in cm^2?
exp_var /= 10**4 # convert to m
var = sc['positions'][:, 2].var()
print "expected_var:", exp_var, "var:", var
assert np.allclose(exp_var, var, rtol=0.25)
def test_bottom_layer():
"""
tests the bottom layer
elements should vertically spread according to the diffusion coef of the bottom layer
"""
D_lower = .11 #m^2/s (or cm^2/s?)
time_step = 60
total_time = 60000
num_elements = 100
num_timesteps = total_time // time_step
mv = RandomVerticalMover(vertical_diffusion_coef_below_ml=D_lower) # m
sc = sample_sc_release(num_elements=num_elements,
start_pos=(0.0, 0.0, 0.0),
release_time=model_time,
)
# re-set z positions:
sc['positions'][:, 2] = 1000.0 # far from the top
# sc['positions'][ :num_elements/2 , 2] = 5.0 # near top
# sc['positions'][ num_elements/2:, 2] = 50.0 # down deep
# call get_move a bunch of times
for i in range(num_timesteps):
# print "positions:\n", sc['positions']
delta = mv.get_move(sc,
time_step,
model_time,
)
# print "delta:\n", delta
sc['positions'] += delta
# print sc['positions']
# expected variance:
exp_var = 2 * num_timesteps*time_step * D_lower # in cm^2?
exp_var /= 10**4 # convert to m
var = sc['positions'][:,2].var()
print "expected_var:", exp_var, "var:", var
assert np.allclose(exp_var, var, rtol=0.1)
