def load_raws_observations(obs_file,glat,glon,grid_dist_km):
"""
Loads all of the RAWS observations valid at the time in question
and converts them to Observation objects.
"""
# load observations & register them to grid
orig_obs = []
if os.path.exists(obs_file):
orig_obs = np.loadtxt(obs_file,dtype=np.object,delimiter=',')
else:
print('WARN: no observation file found.')
obss = []
omin, omax = 0.6, 0.0
# format of file
# 0 1 2 3 4 5 6 7 8 9 10 11
# yyyy,mm,dd,hh,MM,ss,lat,lon,elevation,var_id,value,variance
for oo in orig_obs:
ts = datetime(int(oo[0]),int(oo[1]),int(oo[2]),int(oo[3]),int(oo[4]),int(oo[5]),tzinfo=pytz.timezone('GMT'))
lat, lon, elev = float(oo[6]), float(oo[7]), float(oo[8])
obs, ovar = float(oo[10]), float(oo[11])
i, j = find_closest_grid_point(lat,lon,glat,glon)
# compute distance to grid points
dist_grid_pt = great_circle_distance(lon,lat,glon[i,j],glat[i,j])
# check & remove nonsense zero-variance (or negative variance) observations
if ovar > 0 and dist_grid_pt < grid_dist_km / 2.0:
obss.append(Observation(ts,lat,lon,elev,oo[9],obs,ovar,(i,j)))
omin = min(omin, obs)
omax = max(omax, obs)
print('INFO: loaded %d observations in range %g to %g [%d available]' % (len(obss),omin,omax,len(obss)))
return obss
def register_to_grid(self, wrf_data):
"""
Find the nearest grid point to the current location.
"""
# only co-register to grid if required
mlon, mlat = wrf_data.get_lons(), wrf_data.get_lats()
self.grid_pt = find_closest_grid_point(self.lon, self.lat, mlon, mlat)
self.dist_grid_pt = great_circle_distance(self.lon, self.lat,
mlon[self.grid_pt], mlat[self.grid_pt])
def register_to_grid(self, wrf_data):
"""
Find the nearest grid point to the current location.
"""
# only co-register to grid if required
mlon, mlat = wrf_data.get_lons(), wrf_data.get_lats()
self.grid_pt = find_closest_grid_point(self.lon, self.lat, mlon, mlat)
self.dist_grid_pt = great_circle_distance(self.lon, self.lat,
mlon[self.grid_pt], mlat[self.grid_pt])
def get_grid_values_from_file(ncpath,varname,lat,lon): d = netCDF4.Dataset(ncpath) glat, glon = d.variables['Lat'][:,:], d.variables['Lon'][:,:] V = d.variables[varname] i, j = find_closest_grid_point(lon, lat, glon, glat) dist = great_circle_distance(glon[i,j],glat[i,j],lon,lat) vals = V[i,j,...] d.close() return dist, i, j, vals
def get_ngp(lat,lon,ncpath):
"""
Finds the closest grid point to lat/lon, prints the indices and shows the distance to the point.
"""
d = netCDF4.Dataset(ncpath)
# find closest grid point
glat,glon = d.variables['XLAT'][0,:,:], d.variables['XLONG'][0,:,:]
i,j = find_closest_grid_point(lon,lat,glon,glat)
dist = great_circle_distance(lon,lat,glon[i,j],glat[i,j])
print('GP closest to lat/lon %g,%g is %d,%d with distance %g km.' % (lat,lon,i,j,dist))
def get_ngp(lat, lon, ncpath):
"""
Finds the closest grid point to lat/lon, prints the indices and shows the distance to the point.
"""
d = netCDF4.Dataset(ncpath)
# find closest grid point
glat, glon = d.variables['XLAT'][0, :, :], d.variables['XLONG'][0, :, :]
i, j = find_closest_grid_point(lon, lat, glon, glat)
dist = great_circle_distance(lon, lat, glon[i, j], glat[i, j])
print('GP closest to lat/lon %g,%g is %d,%d with distance %g km.' %
(lat, lon, i, j, dist))
def load_raws_observations(obs_file, glat, glon, grid_dist_km):
"""
Loads all of the RAWS observations valid at the time in question
and converts them to Observation objects.
"""
# load observations & register them to grid
orig_obs = []
if os.path.exists(obs_file):
orig_obs = np.loadtxt(obs_file, dtype=np.object, delimiter=',')
else:
print('WARN: no observation file found.')
obss = []
omin, omax = 0.6, 0.0
# format of file
# 0 1 2 3 4 5 6 7 8 9 10 11
# yyyy,mm,dd,hh,MM,ss,lat,lon,elevation,var_id,value,variance
for oo in orig_obs:
ts = datetime(int(oo[0]),
int(oo[1]),
int(oo[2]),
int(oo[3]),
int(oo[4]),
int(oo[5]),
tzinfo=pytz.timezone('GMT'))
lat, lon, elev = float(oo[6]), float(oo[7]), float(oo[8])
obs, ovar = float(oo[10]), float(oo[11])
i, j = find_closest_grid_point(lat, lon, glat, glon)
# compute distance to grid points
dist_grid_pt = great_circle_distance(lon, lat, glon[i, j], glat[i, j])
# check & remove nonsense zero-variance (or negative variance) observations
if ovar > 0 and dist_grid_pt < grid_dist_km / 2.0:
obss.append(
Observation(ts, lat, lon, elev, oo[9], obs, ovar, (i, j)))
omin = min(omin, obs)
omax = max(omax, obs)
print('INFO: loaded %d observations in range %g to %g [%d available]' %
(len(obss), omin, omax, len(obss)))
return obss
def extract_nearest_time_series(varnames, lat, lon, ncpath):
"""
Extract the time series in varnames for the grid point that is closest to the given lat/lon
and store them in a CSV file in the same order as varnames.
This assumes that the time dimension is first. Each variable name is either a name or a tuple
with name and index (for 4d vars).
"""
d = netCDF4.Dataset(ncpath)
# find closest grid point
glat, glon = d.variables['XLAT'][0, :, :], d.variables['XLONG'][0, :, :]
i, j = find_closest_grid_point(lon, lat, glon, glat)
# retrieve all time series
tss = []
for vi in varnames:
if isinstance(vi, tuple):
vn, ndx = vi
tss.append(d.variables[vn][:, ndx, i, j])
else:
tss.append(d.variables[vi][:, i, j])
# extract time strings (example 2012-09-14_23:00:00) & convert to datetime
esmft = d.variables['Times'][:, :]
times = map(lambda x: ''.join(x), esmft)
dts = map(
lambda x: datetime(int(x[:4]), int(x[5:7]), int(x[8:10]), int(x[
11:13]), int(x[14:16]), 0), times)
d.close()
for i in range(len(dts)):
dt = dts[i]
toks = map(lambda x: str(x),
[dt.year, dt.month, dt.day, dt.hour, dt.minute, 0])
for ts in tss:
toks.append(str(ts[i]))
print(','.join(toks))
def load_raws_observations(obs_file,glat,glon):
"""
Loads all of the RAWS observations valid at the time in question
and converts them to Observation objects.
"""
# load observations & register them to grid
orig_obs = []
if os.path.exists(obs_file):
orig_obs = np.loadtxt(obs_file,delimiter=',')
else:
print('WARN: no observation file found, only performing model time step.')
obss = []
omin, omax = 0.6, 0.0
for oo in orig_obs:
i, j = find_closest_grid_point(oo[6],oo[7],glat,glon)
obst = datetime(int(oo[0]),int(oo[1]),int(oo[2]),int(oo[3]),int(oo[4]),int(oo[5]))
# check & remove non-sense zero observations
if oo[8] > 0:
obss.append(Observation(obst,oo[6],oo[7],oo[8],oo[9],(i,j)))
omin = min(omin, oo[8])
omax = max(omax, oo[8])
print('INFO: loaded %d observations in range %g to %g' % (len(obss),omin,omax))
return obss
def extract_nearest_time_series(varnames,lat,lon,ncpath):
"""
Extract the time series in varnames for the grid point that is closest to the given lat/lon
and store them in a CSV file in the same order as varnames.
This assumes that the time dimension is first. Each variable name is either a name or a tuple
with name and index (for 4d vars).
"""
d = netCDF4.Dataset(ncpath)
# find closest grid point
glat,glon = d.variables['XLAT'][0,:,:], d.variables['XLONG'][0,:,:]
i,j = find_closest_grid_point(lon,lat,glon,glat)
# retrieve all time series
tss = []
for vi in varnames:
if isinstance(vi,tuple):
vn,ndx = vi
tss.append(d.variables[vn][:,ndx,i,j])
else:
tss.append(d.variables[vi][:,i,j])
# extract time strings (example 2012-09-14_23:00:00) & convert to datetime
esmft = d.variables['Times'][:,:]
times = map(lambda x: ''.join(x), esmft)
dts = map(lambda x: datetime(int(x[:4]),int(x[5:7]),int(x[8:10]),int(x[11:13]), int(x[14:16]), 0), times)
d.close()
for i in range(len(dts)):
dt = dts[i]
toks = map(lambda x: str(x), [dt.year,dt.month,dt.day,dt.hour,dt.minute,0])
for ts in tss:
toks.append(str(ts[i]))
print(','.join(toks))
