def _write_body(self):
# logger.debug('generating body')
vi = self.ssp.cur.proc_valid
for idx in range(np.sum(vi)):
self.fod.io.write(
"%8.2f%10.2f%10.2f%10.2f%10.2f\n" %
(self.ssp.cur.proc.depth[vi][idx],
self.ssp.cur.proc.speed[vi][idx],
self.ssp.cur.proc.temp[vi][idx],
self.ssp.cur.proc.sal[vi][idx],
Oc.s2c(s=self.ssp.cur.proc.sal[vi][idx],
p=Oc.d2p(d=self.ssp.cur.proc.depth[vi][idx],
lat=self.ssp.cur.meta.latitude),
t=self.ssp.cur.proc.temp[vi][idx])))
def query(self,
lat: Optional[float],
lon: Optional[float],
datestamp: Union[date, dt, None] = None,
server_mode: bool = False):
"""Query RTOFS for passed location and timestamp"""
if datestamp is None:
datestamp = dt.utcnow()
if isinstance(datestamp, dt):
datestamp = datestamp.date()
if not isinstance(datestamp, date):
raise RuntimeError("invalid date passed: %s" % type(datestamp))
logger.debug("query: %s @ (%.6f, %.6f)" % (datestamp, lon, lat))
# check the inputs
if (lat is None) or (lon is None) or (datestamp is None):
logger.error("invalid query: %s @ (%s, %s)" %
(datestamp.strftime("%Y%m%d"), lon, lat))
return None
try:
lat_idx, lon_idx = self.grid_coords(lat,
lon,
datestamp=datestamp,
server_mode=server_mode)
if lat_idx is None:
logger.info("location outside of GoMOFS coverage")
return None
except TypeError as e:
logger.critical("while converting location to grid coords, %s" % e)
return None
logger.debug("idx > lat: %s, lon: %s" % (lat_idx, lon_idx))
lat_s_idx = lat_idx - self._search_half_window
if lat_s_idx < 0:
lat_s_idx = 0
lat_n_idx = lat_idx + self._search_half_window
if lat_n_idx >= self._lat.shape[0]:
lat_n_idx = self._lat.shape[0] - 1
lon_w_idx = lon_idx - self._search_half_window
if lon_w_idx < 0:
lon_w_idx = 0
lon_e_idx = lon_idx + self._search_half_window
if lon_e_idx >= self._lon.shape[1]:
lon_e_idx = self._lon.shape[1] - 1
# logger.info("indices -> %s %s %s %s" % (lat_s_idx, lat_n_idx, lon_w_idx, lon_e_idx))
lat_search_window = lat_n_idx - lat_s_idx + 1
lon_search_window = lon_e_idx - lon_w_idx + 1
logger.info("updated search window: (%s, %s)" %
(lat_search_window, lon_search_window))
# Need +1 on the north and east indices since it is the "stop" value in these slices
t = self._file.variables['temp'][self._day_idx, :,
lat_s_idx:lat_n_idx + 1,
lon_w_idx:lon_e_idx + 1]
s = self._file.variables['salt'][self._day_idx, :,
lat_s_idx:lat_n_idx + 1,
lon_w_idx:lon_e_idx + 1]
# Set 'unfilled' elements to NANs (BUT when the entire array has valid data, it returns numpy.ndarray)
if isinstance(t, np.ma.core.MaskedArray):
t_mask = t.mask
t._sharedmask = False
t[t_mask] = np.nan
if isinstance(s, np.ma.core.MaskedArray):
s_mask = s.mask
s._sharedmask = False
s[s_mask] = np.nan
# Calculate distances from requested position to each of the grid node locations
distances = np.zeros(
(self._d.size, lon_search_window, lat_search_window))
longitudes = self._lon[lat_s_idx:lat_n_idx + 1,
lon_w_idx:lon_e_idx + 1]
latitudes = self._lat[lat_s_idx:lat_n_idx + 1, lon_w_idx:lon_e_idx + 1]
for i in range(lat_search_window):
for j in range(lon_search_window):
dist = self.g.distance(longitudes[i, j], latitudes[i, j], lon,
lat)
distances[:, i, j] = dist
# logger.info("node (%s %s), pos: %3.2f, %3.2f, dist: %3.1f"
# % (i, j, latitudes[i, j], longitudes[i, j], distances[0, i, j]))
# Get mask of "no data" elements and replace these with NaNs in distance array
t_mask = np.isnan(t)
distances[t_mask] = np.nan
s_mask = np.isnan(s)
distances[s_mask] = np.nan
# logger.info("distance array:\n%s" % distances[0])
# Spin through all the depth levels
temp_pot = np.zeros(self._d.size)
temp_in_situ = np.zeros(self._d.size)
d = np.zeros(self._d.size)
sal = np.zeros(self._d.size)
num_values = 0
for i in range(self._d.size):
t_level = t[i]
s_level = s[i]
d_level = distances[i]
try:
ind = np.nanargmin(d_level)
except ValueError:
# logger.info("%s: all-NaN slices" % i)
continue
if np.isnan(ind):
logger.info("%s: bottom of valid data" % i)
break
ind2 = np.unravel_index(ind, t_level.shape)
t_closest = t_level[ind2]
s_closest = s_level[ind2]
# d_closest = d_level[ind2]
temp_pot[i] = t_closest
sal[i] = s_closest
d[i] = self._d[i]
# Calculate in-situ temperature
p = Oc.d2p(d[i], lat)
temp_in_situ[i] = Oc.in_situ_temp(s=sal[i],
t=t_closest,
p=p,
pr=self._ref_p)
# logger.info("%02d: %6.1f %6.1f > T/S/Dist: %3.1f %3.1f %3.1f [pot.temp. %3.1f]"
# % (i, d[i], p, temp_in_situ[i], s_closest, d_closest, t_closest))
num_values += 1
if num_values == 0:
logger.info("no data from lookup!")
return None
# ind = np.nanargmin(distances[0])
# ind2 = np.unravel_index(ind, distances[0].shape)
# switching to the query location
# lat_out = latitudes[ind2]
# lon_out = longitudes[ind2]
# while lon_out > 180.0:
# lon_out -= 360.0
# Make a new SV object to return our query in
ssp = Profile()
ssp.meta.sensor_type = Dicts.sensor_types['Synthetic']
ssp.meta.probe_type = Dicts.probe_types['GoMOFS']
ssp.meta.latitude = lat
if lon > 180.0: # Go back to negative longitude
lon -= 360.0
ssp.meta.longitude = lon
ssp.meta.utc_time = dt(year=datestamp.year,
month=datestamp.month,
day=datestamp.day)
ssp.meta.original_path = "GoMOFS_%s" % datestamp.strftime("%Y%m%d")
ssp.init_data(num_values)
ssp.data.depth = d[0:num_values]
ssp.data.temp = temp_in_situ[0:num_values]
ssp.data.sal = sal[0:num_values]
ssp.calc_data_speed()
ssp.clone_data_to_proc()
ssp.init_sis()
profiles = ProfileList()
profiles.append_profile(ssp)
return profiles
def query(self, lat: Optional[float], lon: Optional[float], dtstamp: Optional[dt] = None,
server_mode: bool = False):
"""Query RTOFS for passed location and timestamp"""
if dtstamp is None:
dtstamp = dt.utcnow()
if not isinstance(dtstamp, dt):
raise RuntimeError("invalid datetime passed: %s" % type(dtstamp))
logger.debug("query: %s @ (%.6f, %.6f)" % (dtstamp, lon, lat))
# check the inputs
if (lat is None) or (lon is None):
logger.error("invalid query: %s @ (%s, %s)" % (dtstamp.strftime("%Y/%m/%d %H:%M:%S"), lon, lat))
return None
try:
lat_idx, lon_idx = self.grid_coords(lat, lon, dtstamp=dtstamp, server_mode=server_mode)
except TypeError as e:
logger.critical("while converting location to grid coords, %s" % e)
return None
# logger.debug("idx > lat: %s, lon: %s" % (lat_idx, lon_idx))
lat_s_idx = lat_idx - self._search_half_window
lat_n_idx = lat_idx + self._search_half_window
lon_w_idx = lon_idx - self._search_half_window
lon_e_idx = lon_idx + self._search_half_window
# logger.info("indices -> %s %s %s %s" % (lat_s_idx, lat_n_idx, lon_w_idx, lon_e_idx))
if lon < self._lon_0: # Make all longitudes safe
lon += 360.0
longitudes = np.zeros((self._search_window, self._search_window))
if (lon_e_idx < self._lon.size) and (lon_w_idx >= 0):
# logger.info("safe case")
# Need +1 on the north and east indices since it is the "stop" value in these slices
t = self._file_temp.variables['temperature'][self._day_idx, :, lat_s_idx:lat_n_idx + 1,
lon_w_idx:lon_e_idx + 1]
s = self._file_sal.variables['salinity'][self._day_idx, :, lat_s_idx:lat_n_idx + 1, lon_w_idx:lon_e_idx + 1]
# Set 'unfilled' elements to NANs (BUT when the entire array has valid data, it returns numpy.ndarray)
if isinstance(t, np.ma.core.MaskedArray):
t_mask = t.mask
t._sharedmask = False
t[t_mask] = np.nan
if isinstance(s, np.ma.core.MaskedArray):
s_mask = s.mask
s._sharedmask = False
s[s_mask] = np.nan
lons = self._lon[lon_w_idx:lon_e_idx + 1]
for i in range(self._search_window):
longitudes[i, :] = lons
else:
logger.info("split case")
# --- Do the left portion of the array first, this will run into the wrap longitude
lon_e_idx = self._lon.size - 1
# lon_west_index can be negative if lon_index is on the westernmost end of the array
if lon_w_idx < 0:
lon_w_idx = lon_w_idx + self._lon.size
# logger.info("using lon west/east indices -> %s %s" % (lon_w_idx, lon_e_idx))
# Need +1 on the north and east indices since it is the "stop" value in these slices
t_left = self._file_temp.variables['temperature'][self._day_idx, :, lat_s_idx:lat_n_idx + 1,
lon_w_idx:lon_e_idx + 1]
s_left = self._file_sal.variables['salinity'][self._day_idx, :, lat_s_idx:lat_n_idx + 1,
lon_w_idx:lon_e_idx + 1]
# Set 'unfilled' elements to NANs (BUT when the entire array has valid data, it returns numpy.ndarray)
if isinstance(t_left, np.ma.core.MaskedArray):
t_mask = t_left.mask
t_left[t_mask] = np.nan
if isinstance(s_left, np.ma.core.MaskedArray):
s_mask = s_left.mask
s_left[s_mask] = np.nan
lons_left = self._lon[lon_w_idx:lon_e_idx + 1]
for i in range(self._search_window):
longitudes[i, 0:lons_left.size] = lons_left
# logger.info("longitudes are now: %s" % longitudes)
# --- Do the right portion of the array first, this will run into the wrap
# longitude so limit it accordingly
lon_w_idx = 0
lon_e_idx = self._search_window - lons_left.size - 1
# Need +1 on the north and east indices since it is the "stop" value in these slices
t_right = self._file_temp.variables['temperature'][self._day_idx, :, lat_s_idx:lat_n_idx + 1,
lon_w_idx:lon_e_idx + 1]
s_right = self._file_sal.variables['salinity'][self._day_idx, :, lat_s_idx:lat_n_idx + 1,
lon_w_idx:lon_e_idx + 1]
# Set 'unfilled' elements to NANs (BUT when the entire array has valid data, it returns numpy.ndarray)
if isinstance(t_right, np.ma.core.MaskedArray):
t_mask = t_right.mask
t_right[t_mask] = np.nan
if isinstance(s_right, np.ma.core.MaskedArray):
s_mask = s_right.mask
s_right[s_mask] = np.nan
lons_right = self._lon[lon_w_idx:lon_e_idx + 1]
for i in range(self._search_window):
longitudes[i, lons_left.size:self._search_window] = lons_right
# merge data
t = np.zeros((self._file_temp.variables['lev'].size, self._search_window, self._search_window))
t[:, :, 0:lons_left.size] = t_left
t[:, :, lons_left.size:self._search_window] = t_right
s = np.zeros((self._file_temp.variables['lev'].size, self._search_window, self._search_window))
s[:, :, 0:lons_left.size] = s_left
s[:, :, lons_left.size:self._search_window] = s_right
# Calculate distances from requested position to each of the grid node locations
distances = np.zeros((self._d.size, self._search_window, self._search_window))
latitudes = np.zeros((self._search_window, self._search_window))
lats = self._lat[lat_s_idx:lat_n_idx + 1]
for i in range(self._search_window):
latitudes[:, i] = lats
for i in range(self._search_window):
for j in range(self._search_window):
dist = self.g.distance(longitudes[i, j], latitudes[i, j], lon, lat)
distances[:, i, j] = dist
# logger.info("node %s, pos: %3.1f, %3.1f, dist: %3.1f"
# % (i, latitudes[i, j], longitudes[i, j], distances[0, i, j]))
# logger.info("distance array:\n%s" % distances[0])
# Get mask of "no data" elements and replace these with NaNs in distance array
t_mask = np.isnan(t)
distances[t_mask] = np.nan
s_mask = np.isnan(s)
distances[s_mask] = np.nan
# Spin through all the depth levels
temp_pot = np.zeros(self._d.size)
temp_in_situ = np.zeros(self._d.size)
d = np.zeros(self._d.size)
sal = np.zeros(self._d.size)
num_values = 0
for i in range(self._d.size):
t_level = t[i]
s_level = s[i]
d_level = distances[i]
try:
ind = np.nanargmin(d_level)
except ValueError:
# logger.info("%s: all-NaN slices" % i)
continue
if np.isnan(ind):
logger.info("%s: bottom of valid data" % i)
break
ind2 = np.unravel_index(ind, t_level.shape)
t_closest = t_level[ind2]
s_closest = s_level[ind2]
# d_closest = d_level[ind2]
temp_pot[i] = t_closest
sal[i] = s_closest
d[i] = self._d[i]
# Calculate in-situ temperature
p = Oc.d2p(d[i], lat)
temp_in_situ[i] = Oc.in_situ_temp(s=sal[i], t=t_closest, p=p, pr=self._ref_p)
# logger.info("%02d: %6.1f %6.1f > T/S/Dist: %3.1f %3.1f %3.1f [pot.temp. %3.1f]"
# % (i, d[i], p, temp_in_situ[i], s_closest, d_closest, t_closest))
num_values += 1
if num_values == 0:
logger.info("no data from lookup!")
return None
# ind = np.nanargmin(distances[0])
# ind2 = np.unravel_index(ind, distances[0].shape)
# switching to the query location
# lat_out = latitudes[ind2]
# lon_out = longitudes[ind2]
# while lon_out > 180.0:
# lon_out -= 360.0
# Make a new SV object to return our query in
ssp = Profile()
ssp.meta.sensor_type = Dicts.sensor_types['Synthetic']
ssp.meta.probe_type = Dicts.probe_types['RTOFS']
ssp.meta.latitude = lat
if lon > 180.0: # Go back to negative longitude
lon -= 360.0
ssp.meta.longitude = lon
ssp.meta.utc_time = dt(year=dtstamp.year, month=dtstamp.month, day=dtstamp.day,
hour=dtstamp.hour, minute=dtstamp.minute, second=dtstamp.second)
ssp.meta.original_path = "RTOFS_%s" % dtstamp.strftime("%Y%m%d_%H%M%S")
ssp.init_data(num_values)
ssp.data.depth = d[0:num_values]
ssp.data.temp = temp_in_situ[0:num_values]
ssp.data.sal = sal[0:num_values]
ssp.calc_data_speed()
ssp.clone_data_to_proc()
ssp.init_sis()
profiles = ProfileList()
profiles.append_profile(ssp)
return profiles