def test_iteritems():
d = {'a': 0, 'b': 1, 'c': 2}
for k, v in iteritems(d):
assert type(k) == str
assert type(v) == int
def rasm_soil(data, soil_file):
"""Write VIC formatted soil parameter file"""
message = """
*** ---------------------------------------------------------------------- ***
Notes about RASM soil parameter file generations:
- To fill in missing grid cells 'mask' variable must be the same as the
domain file mask.
- Inactive grid cells will have a dummy line printed for all variables except
the lons/lats.
- Any grid cells with nans will be copied from the previous line without nans
or FILL_VALUEs.
*** --------------------------------------------------------------------- ***\n
"""
print(message)
# ---------------------------------------------------------------- #
c = grid_params.cols(nlayers=3)
f = grid_params.format(nlayers=3)
c.soil_param['Nveg'] = np.array([0])
f.soil_param['Nveg'] = '%1i'
numcells = data['mask'].size
arrayshape = (numcells, 1 +
np.max([np.max(cols)
for v, cols in iteritems(c.soil_param)]))
soil_params = np.empty(arrayshape)
dtypes = ['%1i'] * arrayshape[1]
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# find nearest real grid cell for all grid cells where frozen soil mask is
# active
# This is needed because the RASM mask is often different than existing
# datasets
print('Finding/filling nearest neighbors for all variables based on '
'fs_active mask')
# real mask (from input dataset)
ry, rx = np.nonzero(data['fs_active'])
# fill mask (we will fill all of these grid cells with their nearest real
# value)
my_mask = np.zeros(data['fs_active'].shape, dtype=int)
my_mask[ry, rx] = 1
fy, fx = np.nonzero(my_mask == 0)
# Find nearest real grid cell
combined = np.dstack(([data['yc'][ry, rx], data['xc'][ry, rx]]))[0]
points = list(
np.vstack((data['yc'][fy, fx], data['xc'][fy, fx])).transpose())
mytree = cKDTree(combined)
dist, inds = mytree.query(points, k=1)
# loop over all variables and fill in values
for var in c.soil_param:
if var not in [
'run_cell', 'grid_cell', 'lats', 'lons', 'fs_active', 'Nveg'
]:
# unmask the variable
data[var] = np.array(data[var])
# fill with nearest data
if data[var].ndim == 2:
data[var][fy, fx] = data[var][ry[inds], rx[inds]]
elif data[var].ndim == 3:
data[var][:, fy, fx] = data[var][:, ry[inds], rx[inds]]
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# This is the domain mask
yinds, xinds = np.nonzero(data['mask'] == 1)
ymask, xmask = np.nonzero(data['mask'] == 0)
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Fix problematic avg_T values
print('Finding/filling nearest neighbors for avg_T\n')
# real mask (from input dataset)
ry, rx = np.nonzero((data['avg_T'] > -50) & (data['avg_T'] < 99))
# fill mask (we will fill all of these grid cells with their nearest
# real value)
my_mask = np.zeros(data['avg_T'].shape)
my_mask[ry, rx] = 1
fy, fx = np.nonzero(my_mask == 0)
# Find nearest real grid cell
if len(fy) > 0:
combined = np.dstack(([data['yc'][ry, rx], data['xc'][ry, rx]]))[0]
points = list(
np.vstack((data['yc'][fy, fx], data['xc'][fy, fx])).transpose())
mytree = cKDTree(combined)
dist, inds = mytree.query(points, k=1)
data['avg_T'] = np.array(data['avg_T'])
data['avg_T'][fy, fx] = data['avg_T'][ry[inds], rx[inds]]
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# For rasm, all cols are shifted one to right to make room for nveg in
# col 0
i = -1
for var, cols in iteritems(c.soil_param):
for col in cols:
dtypes[col] = f.soil_param[var]
for (y, x), maskval in np.ndenumerate(data['mask']):
# advance the row count
i += 1
# put real data
for var in c.soil_param:
cols = c.soil_param[var]
if data[var].ndim == 2:
soil_params[i, cols] = data[var][y, x]
elif data[var].ndim == 3:
for j, col in enumerate(cols):
soil_params[i, col] = data[var][j, y, x]
# write the grid cell number
soil_params[i, 2] = i + 1
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Write phi_s
soil_params[:, c.soil_param['phi_s']] = -999
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Write nveg
soil_params[:, 0] = data['Nveg'][:, :].flatten()
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Set all grid cells to run
soil_params[:, 1] = 1 # run
soil_params[:, -1] = 1 # run with frozen soils on
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# rewrite the lons/lats columns
soil_params[:, 3] = data['lats'].flatten()
soil_params[:, 4] = data['lons'].flatten()
assert soil_params[-1, 3] == data['lats'][y, x]
assert soil_params[-1, 4] == data['lons'][y, x]
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# check for nans
# sometimes RASM has land grid cells that the input file does not.
# In this case, we will use the previous lnd grid cell
if np.isnan(soil_params.sum()):
bad_cells = []
replacements = []
for i, cell in enumerate(soil_params):
if np.isnan(np.sum(cell)):
bad_cells.append(i)
j = i
while True:
j -= 1
if (FILL_VALUE not in soil_params[j, :]) and \
not np.isnan(np.sum(soil_params[j, :])):
soil_params[i, 5:] = soil_params[j, 5:]
replacements.append(j)
break
print('Fixed {0} bad cells'.format(len(bad_cells)))
print('Example: {0}-->{1}'.format(bad_cells[0], replacements[0]))
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Print summary of variables
for var, cols in iteritems(c.soil_param):
print('{0: <12}--> min: {1:<09.3f}, max: {2:<09.3f}, mean:'
' {3:<09.3f}'.format(var, soil_params[:, cols].min(),
soil_params[:, cols].max(),
soil_params[:, cols].mean()))
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Finish up
assert soil_params[-1, 3] == data['lats'][y, x]
assert soil_params[-1, 4] == data['lons'][y, x]
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Write the file
print('writing soil parameter file: {0}'.format(soil_file))
np.savetxt(soil_file, soil_params, fmt=dtypes)
print('done RASM with soil')
# ---------------------------------------------------------------- #
return
def rasm_soil(data, soil_file):
"""Write VIC formatted soil parameter file"""
message = """
*** ---------------------------------------------------------------------- ***
Notes about RASM soil parameter file generations:
- To fill in missing grid cells 'mask' variable must be the same as the
domain file mask.
- Inactive grid cells will have a dummy line printed for all variables except
the lons/lats.
- Any grid cells with nans will be copied from the previous line without nans
or FILL_VALUEs.
*** --------------------------------------------------------------------- ***\n
"""
print(message)
# ---------------------------------------------------------------- #
c = grid_params.cols(nlayers=3)
f = grid_params.format(nlayers=3)
c.soil_param['Nveg'] = np.array([0])
f.soil_param['Nveg'] = '%1i'
numcells = data['mask'].size
arrayshape = (numcells, 1 + np.max([np.max(cols) for v, cols in
iteritems(c.soil_param)]))
soil_params = np.empty(arrayshape)
dtypes = ['%1i'] * arrayshape[1]
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# find nearest real grid cell for all grid cells where frozen soil mask is
# active
# This is needed because the RASM mask is often different than existing
# datasets
print('Finding/filling nearest neighbors for all variables based on '
'fs_active mask')
# real mask (from input dataset)
ry, rx = np.nonzero(data['fs_active'])
# fill mask (we will fill all of these grid cells with their nearest real
# value)
my_mask = np.zeros(data['fs_active'].shape, dtype=int)
my_mask[ry, rx] = 1
fy, fx = np.nonzero(my_mask == 0)
# Find nearest real grid cell
combined = np.dstack(([data['yc'][ry, rx], data['xc'][ry, rx]]))[0]
points = list(np.vstack((data['yc'][fy, fx],
data['xc'][fy, fx])).transpose())
mytree = cKDTree(combined)
dist, inds = mytree.query(points, k=1)
# loop over all variables and fill in values
for var in c.soil_param:
if var not in ['run_cell', 'grid_cell', 'lats', 'lons', 'fs_active',
'Nveg']:
# unmask the variable
data[var] = np.array(data[var])
# fill with nearest data
if data[var].ndim == 2:
data[var][fy, fx] = data[var][ry[inds], rx[inds]]
elif data[var].ndim == 3:
data[var][:, fy, fx] = data[var][:, ry[inds], rx[inds]]
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# This is the domain mask
yinds, xinds = np.nonzero(data['mask'] == 1)
ymask, xmask = np.nonzero(data['mask'] == 0)
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Fix problematic avg_T values
print('Finding/filling nearest neighbors for avg_T\n')
# real mask (from input dataset)
ry, rx = np.nonzero((data['avg_T'] > -50) & (data['avg_T'] < 99))
# fill mask (we will fill all of these grid cells with their nearest
# real value)
my_mask = np.zeros(data['avg_T'].shape)
my_mask[ry, rx] = 1
fy, fx = np.nonzero(my_mask == 0)
# Find nearest real grid cell
if len(fy) > 0:
combined = np.dstack(([data['yc'][ry, rx], data['xc'][ry, rx]]))[0]
points = list(np.vstack((data['yc'][fy, fx],
data['xc'][fy, fx])).transpose())
mytree = cKDTree(combined)
dist, inds = mytree.query(points, k=1)
data['avg_T'] = np.array(data['avg_T'])
data['avg_T'][fy, fx] = data['avg_T'][ry[inds], rx[inds]]
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# For rasm, all cols are shifted one to right to make room for nveg in
# col 0
i = -1
for var, cols in iteritems(c.soil_param):
for col in cols:
dtypes[col] = f.soil_param[var]
for (y, x), maskval in np.ndenumerate(data['mask']):
# advance the row count
i += 1
# put real data
for var in c.soil_param:
cols = c.soil_param[var]
if data[var].ndim == 2:
soil_params[i, cols] = data[var][y, x]
elif data[var].ndim == 3:
for j, col in enumerate(cols):
soil_params[i, col] = data[var][j, y, x]
# write the grid cell number
soil_params[i, 2] = i + 1
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Write phi_s
soil_params[:, c.soil_param['phi_s']] = -999
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Write nveg
soil_params[:, 0] = data['Nveg'][:, :].flatten()
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Set all grid cells to run
soil_params[:, 1] = 1 # run
soil_params[:, -1] = 1 # run with frozen soils on
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# rewrite the lons/lats columns
soil_params[:, 3] = data['lats'].flatten()
soil_params[:, 4] = data['lons'].flatten()
assert soil_params[-1, 3] == data['lats'][y, x]
assert soil_params[-1, 4] == data['lons'][y, x]
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# check for nans
# sometimes RASM has land grid cells that the input file does not.
# In this case, we will use the previous lnd grid cell
if np.isnan(soil_params.sum()):
bad_cells = []
replacements = []
for i, cell in enumerate(soil_params):
if np.isnan(np.sum(cell)):
bad_cells.append(i)
j = i
while True:
j -= 1
if (FILL_VALUE not in soil_params[j, :]) and \
not np.isnan(np.sum(soil_params[j, :])):
soil_params[i, 5:] = soil_params[j, 5:]
replacements.append(j)
break
print('Fixed {0} bad cells'.format(len(bad_cells)))
print('Example: {0}-->{1}'.format(bad_cells[0], replacements[0]))
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Print summary of variables
for var, cols in iteritems(c.soil_param):
print('{0: <12}--> min: {1:<09.3f}, max: {2:<09.3f}, mean:'
' {3:<09.3f}'.format(var,
soil_params[:, cols].min(),
soil_params[:, cols].max(),
soil_params[:, cols].mean()))
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Finish up
assert soil_params[-1, 3] == data['lats'][y, x]
assert soil_params[-1, 4] == data['lons'][y, x]
# ---------------------------------------------------------------- #
# ---------------------------------------------------------------- #
# Write the file
print('writing soil parameter file: {0}'.format(soil_file))
np.savetxt(soil_file, soil_params, fmt=dtypes)
print('done RASM with soil')
# ---------------------------------------------------------------- #
return