start = timevar.units.date2num(jd_start)
istart = timevar.nearest_neighbour_index(start)
stop = timevar.units.date2num(jd_stop)
istop = timevar.nearest_neighbour_index(stop)
# Only proceed if we have data in the range requested.
if istart != istop:
nsta = len(stations)
if len(r) == 4:
# HYCOM and ROMS
# Dimensions are time, elevation, lat, lon
d = u[0, 0:1, :, :].data
# Find the closest non-land point from a structured grid model.
if len(lon.shape) == 1:
lon, lat = np.meshgrid(lon, lat)
j, i, dd = find_ij(lon, lat, d, obs_lon, obs_lat)
for n in range(nsta):
# Only use if model cell is within max_dist of station
if dd[n] <= max_dist:
u_arr = u[istart:istop, 0:1, j[n], i[n]].data
v_arr = v[istart:istop, 0:1, j[n], i[n]].data
arr = np.sqrt(u_arr ** 2 + v_arr ** 2)
if u_arr.std() >= min_var:
c = mod_df(arr, timevar, istart, istop, mod_name, ts)
name = obs_df[n].name
model_df[n] = pd.concat([model_df[n], c], axis=1)
model_df[n].name = name
else:
print "min_var error"
else:
jd = timevar.units.num2date(timevar.points)
start = timevar.units.date2num(jd_start)
istart = timevar.nearest_neighbour_index(start)
stop = timevar.units.date2num(jd_stop)
istop = timevar.nearest_neighbour_index(stop)
# Only proceed if we have data in the range requested.
if istart != istop:
nsta = len(station_list)
if len(r) == 3:
print('[Structured grid model]:', url)
d = a[0, :, :].data
# Find the closest non-land point from a structured grid model.
if len(lon.shape) == 1:
lon, lat = np.meshgrid(lon, lat)
j, i, dd = find_ij(lon, lat, d, obs_lon, obs_lat)
for n in range(nsta):
# Only use if model cell is within 0.04 degree of requested
# location.
if dd[n] <= max_dist:
arr = a[istart:istop, j[n], i[n]].data
if arr.std() >= min_var:
c = mod_df(arr, timevar, istart, istop, mod_name,
ts)
name = station_list[n]['long_name']
model_df[n] = pd.concat([model_df[n], c], axis=1)
model_df[n].name = name
elif len(r) == 2:
print('[Unstructured grid model]:', url)
# Find the closest point from an unstructured grid model.
# Only proceed if we have data in the range requested.
if istart != istop:
nsta = len(stations)
if len(r) == 3:
print('[Structured grid model]:', url)
d = u[0, :, :].data
if len(lon.shape) == 1:
new_lon, new_lat = np.meshgrid(lon, lat)
else:
new_lon, new_lat = lon, lat
# Find the closest non-land point from a structured grid model.
j, i, dd = find_ij(new_lon, new_lat, d, obs_lon, obs_lat)
# Keep the lat lon of the grid point
model_lat = lat[j].tolist()
model_lon = lon[i].tolist()
for n in range(nsta):
# Only use if model cell is within max_dist of station
if dd[n] <= max_dist:
u_arr = u[istart:istop, j[n], i[n]].data
v_arr = v[istart:istop, j[n], i[n]].data
# Model data is in m/s so convert to cm/s
arr = np.sqrt((u_arr)**2 + (v_arr)**2)
if u_arr.std() >= min_var:
c = mod_df(arr, timevar, istart, istop, mod_name,
ts)
start = timevar.units.date2num(jd_start)
istart = timevar.nearest_neighbour_index(start)
stop = timevar.units.date2num(jd_stop)
istop = timevar.nearest_neighbour_index(stop)
# Only proceed if we have data in the range requested.
if istart != istop:
# Wave Watch III uses a 4D grid (time, z, lat, lon)
d = cube[0, 0, :, :].data
if len(lon.shape) == 1:
new_lon, new_lat = np.meshgrid(lon, lat)
else:
new_lon, new_lat = lon, lat
# Find the closest non-land point from a structured grid model.
j, i, dd = find_ij(new_lon, new_lat, d, station_lon, station_lat)
# Keep the lat lon of the grid point
model_lat = lat[j].tolist()
model_lon = lon[i].tolist()
# Only use if model cell is within max_dist of station
if dd <= max_dist:
arr = cube[istart:istop, 0, j, i].data
if arr.std() >= min_var:
c = mod_df(arr, timevar, istart, istop, mod_name, ts)
model_df = pd.concat([model_df, c], axis=1)
model_df.name = get_station_longName(str(station_id))
else:
print 'Min variance error'
else:
# Only proceed if we have data in the range requested.
if istart != istop:
nsta = len(stations)
if len(r) == 3:
print('[Structured grid model]:', url)
d = u[0, :, :].data
if len(lon.shape) == 1:
new_lon, new_lat = np.meshgrid(lon, lat)
else:
new_lon, new_lat = lon, lat
# Find the closest non-land point from a structured grid model.
j, i, dd = find_ij(new_lon, new_lat, d, obs_lon, obs_lat)
# Keep the lat lon of the grid point
model_lat = lat[j].tolist()
model_lon = lon[i].tolist()
for n in range(nsta):
# Only use if model cell is within max_dist of station
if dd[n] <= max_dist:
u_arr = u[istart:istop, j[n], i[n]].data
v_arr = v[istart:istop, j[n], i[n]].data
# Model data is in m/s so convert to cm/s
arr = np.sqrt((u_arr)**2 + (v_arr)**2)
if u_arr.std() >= min_var:
c = mod_df(arr, timevar, istart, istop,
mod_name, ts)
start = timevar.units.date2num(jd_start)
istart = timevar.nearest_neighbour_index(start)
stop = timevar.units.date2num(jd_stop)
istop = timevar.nearest_neighbour_index(stop)
# Only proceed if we have data in the range requested.
if istart != istop:
# Wave Watch III uses a 4D grid (time, z, lat, lon)
d = cube[0, 0, :, :].data
if len(lon.shape) == 1:
new_lon, new_lat = np.meshgrid(lon, lat)
else:
new_lon, new_lat = lon, lat
# Find the closest non-land point from a structured grid model.
j, i, dd = find_ij(new_lon, new_lat, d, station_lon, station_lat)
# Keep the lat lon of the grid point
model_lat = lat[j].tolist()
model_lon = lon[i].tolist()
# Only use if model cell is within max_dist of station
if dd <= max_dist:
arr = cube[istart:istop, 0, j, i].data
if arr.std() >= min_var:
c = mod_df(arr, timevar, istart, istop,
mod_name, ts)
model_df = pd.concat([model_df, c], axis=1)
model_df.name = get_station_longName(str(station_id))
else:
print 'Min variance error'
