def readThenReduce(grid, ndrift, fileloc='projects/driftcards/Deployment_data.txt'):
startdate, startlon, startlat, idnum = read(ndrift, fileloc)
# startdate, startlon, startlat, enddate, endlon, endlat, idnum = read(ndrift)
# If covering the whole domain, need to exclude points outside domain.
# Use info just inside domain so points aren't right at the edge.
xvert = np.hstack((np.flipud(grid['lonr'][1,:]),grid['lonr'][:,1],
grid['lonr'][-2,:],np.flipud(grid['lonr'][:,-2])))
yvert = np.hstack((np.flipud(grid['latr'][1,:]),grid['latr'][:,1],
grid['latr'][-2,:],np.flipud(grid['latr'][:,-2])))
verts = np.vstack((xvert,yvert))
# Form path
path = Path(verts.T)
# nan out all drifters that do not start within the TXLA model domain
for i in xrange(ndrift):
if not path.contains_point(np.vstack((startlon[i], startlat[i]))):
startlon[i] = np.nan
startlat[i] = np.nan
startdate[i] = np.nan
# endlon[i] = np.nan
# endlat[i] = np.nan
# # nan out drifters that have not been found yet
# for i in xrange(ndrift):
# if np.isnan(enddate[i]):
# startlon[i] = np.nan
# startlat[i] = np.nan
# startdate[i] = np.nan
# pdb.set_trace()
# Eliminate nan'ed entries (only save non-nan'ed entries)
ind = ~np.isnan(startlon)
startlon = startlon[ind]
startlat = startlat[ind]
startdate = startdate[ind]
idnum = idnum[ind]
return startdate, startlon, startlat, idnum
def check_points(lon0, lat0, grid, z0=None, nobays=False):
"""
Eliminate starting locations for drifters that are outside numerical
domain and that are masked out. If provided an array of starting vertical
locations in z0, it checks whether these points are at least 1m above the
bottom.
Args:
lon0,lat0: Starting locations for drifters in lon/lat
z0: Starting locations for drifters in z
grid: Grid made from readgrid.py
nobays: Whether to use points in bays or not. Default is False.
Returns:
* lon0,lat0 - Fixed lon0,lat0
* z0 - Fixed z0
"""
lonr = grid.lon_rho
latr = grid.lat_rho
# make interpolation function for water depth h.
# Used to check if float is above the bottom
if z0 is not None:
from scipy.interpolate import interp2d
h = grid.h
hint = interp2d(lonr[:, 1], latr[1, :], h.T, fill_value=np.nan)
# If covering the whole domain, need to exclude points outside domain.
# Use info just inside domain so points aren't right at the edge.
xvert = np.hstack(
(np.flipud(lonr[1, :]), lonr[:, 1], lonr[-2, :], np.flipud(lonr[:,
-2])))
yvert = np.hstack(
(np.flipud(latr[1, :]), latr[:, 1], latr[-2, :], np.flipud(latr[:,
-2])))
verts = np.vstack((xvert, yvert))
# Form path
path = Path(verts.T)
# Loop through particle tracks to eliminate drifters that start outside
# the domain
if lon0.ndim == 2:
for jd in range(lon0.shape[0]): # loop through drifters
for it in range(lon0.shape[1]):
# if drifter is not inside path, nan out this and all
# subsequent points
if not path.contains_point(
np.vstack((lon0[jd, it], lat0[jd, it]))):
lon0[jd, it] = np.nan
lat0[jd, it] = np.nan
if z0 is not None:
z0[jd, it] = np.nan
if z0 is not None:
# check that the drifter starts above the bottom
if z0[jd, it] <= -1 * hint(lon0[jd, it], lat0[jd, it]):
lon0[jd, it] = np.nan
lat0[jd, it] = np.nan
if z0 is not None:
z0[jd, it] = np.nan
elif lon0.ndim == 1:
for jd in range(lon0.shape[0]): # loop through drifters
# if drifter is not inside path, nan out this and all
# subsequent points
if not path.contains_point(np.vstack((lon0[jd], lat0[jd]))):
lon0[jd] = np.nan
lat0[jd] = np.nan
if z0 is not None:
z0[jd] = np.nan
if z0 is not None:
# check that the drifter starts above the bottom
if z0[jd] <= -1 * hint(lon0[jd], lat0[jd]):
lon0[jd] = np.nan
lat0[jd] = np.nan
z0[jd] = np.nan
# Also nan out points that are masked
fmask = mtri.LinearTriInterpolator(grid.trirllrho, grid.mask.flatten())
mask0 = fmask(lon0, lat0) # mask for lon0/lat0 points
ind1 = (mask0 == 1.) # indices select out where points are masked
# If nobays, eliminate points with shallow bathymetry
if nobays:
fh = grid.trirllrho.nn_interpolator(grid.h.flatten())
h0 = fh(lon0, lat0)
ind2 = (h0 > 10.)
else:
ind2 = np.ones(ind1.shape).astype(bool)
ind2 = ~np.isnan(lon0) * ind1 * ind2
L = len(ind2.ravel())
Lnan = sum(ind2.ravel())
print L - Lnan, '/', L, ' drifters NaN-ed out.'
lon0 = lon0[ind2].flatten()
lat0 = lat0[ind2].flatten()
if z0 is not None:
z0 = z0[ind2].flatten()
if z0 is None:
return lon0, lat0
else:
return lon0, lat0, z0
# continue
# Get necessary info from File
# d = np.load(File)
# xg = d['xg']; yg = d['yg']; tg = d['tg']
d = netCDF.Dataset(File)
xg = d.variables['xg'][:]
yg = d.variables['yg'][:]
tg = d.variables['tp'][:]
d.close()
# Load in coastline region info
# Mexico
dconn = np.load('calcs/MXpts.npz')
xp = np.asarray(dconn['MX'])[:,0]; yp = np.asarray(dconn['MX'])[:,1]
MXpath = Path(np.vstack((xp, yp)).T)
dconn.close()
# S Texas
dconn = np.load('calcs/STXpts.npz')
xp = np.asarray(dconn['STX'])[:,0]; yp = np.asarray(dconn['STX'])[:,1]
STXpath = Path(np.vstack((xp, yp)).T)
dconn.close()
# N Texas
dconn = np.load('calcs/NTXpts.npz')
xp = np.asarray(dconn['NTX'])[:,0]; yp = np.asarray(dconn['NTX'])[:,1]
NTXpath = Path(np.vstack((xp, yp)).T)
dconn.close()
# Chenier
dconn = np.load('calcs/CHpts.npz')
xp = np.asarray(dconn['CH'])[:,0]; yp = np.asarray(dconn['CH'])[:,1]
CHpath = Path(np.vstack((xp, yp)).T)
def init(name):
'''
Initialization for the simulation.
'''
# loc = 'http://barataria.tamu.edu:6060/thredds/dodsC/NcML/txla_nesting6.nc'
time_units = 'seconds since 1970-01-01'
# horizontal_diffusivity project showed that relative dispersion did not
# change between nsteps=25 and 50, but does between nsteps=5 and 25, and
# interim numbers have not been tested yet.
nsteps = 25 # in-between tracks: 12 # old tracks: 25
# Number of steps to divide model output for outputting drifter location
N = 1 # not interested in drifter locations
# Number of days
ndays = 30
# This is a backward-moving simulation
ff = -1
# Time between outputs
tseas = 4*3600 # 4 hours between outputs, in seconds, time between model outputs
ah = 0. # old tracks: 5.
av = 0. # m^2/s
# surface drifters
z0 = 's'
zpar = 29
# for 3d flag, do3d=0 makes the run 2d and do3d=1 makes the run 3d
do3d = 0
doturb = 0
# Flag for streamlines.
dostream = 1
# Initialize Tracpy class
tp = Tracpy(currents_filename, grid_filename=grid_filename, name=name, tseas=tseas, ndays=ndays, nsteps=nsteps, dostream=dostream, savell=False, doperiodic=0,
N=N, ff=ff, ah=ah, av=av, doturb=doturb, do3d=do3d, z0=z0, zpar=zpar,
time_units=time_units, usebasemap=True, grid=grid, vert_filename=vert_filename)
# tp._readgrid()
# Initial lon/lat locations for drifters
startindsfile = 'NTX-starting-inds.npz'
if not os.path.exists(startindsfile):
# get starting drifter locations for this region
loc_shelftransport = '/home/kthyng/projects/shelf_transport/'
NTX = np.load(loc_shelftransport + 'calcs/NTXpts.npz')['NTX']
xp = NTX[:,0]; yp = NTX[:,1]
gpath = Path(np.vstack((xp, yp)).T)
drifters = netCDF.Dataset(loc_shelftransport + 'tracks/2004-01-01T00gc.nc')
xg = drifters.variables['xg'][:]; yg = drifters.variables['yg'][:]
# Change to projected drifter locations now
nanind = np.isnan(xg) + (xg==-1) # indices where nans are location in xg, yg; for reinstitution of nans
# pdb.set_trace()
xp, yp, _ = tracpy.tools.interpolate2d(xg, yg, grid, 'm_ij2xy')
xp[nanind] = np.nan; yp[nanind] = np.nan
del(xg,yg) # don't need grid info anymore
# save indices of drifters that start in the coastal areas
inds = gpath.contains_points(np.vstack((xp[:,0].flat, yp[:,0].flat)).T).reshape(xp[:,0].shape)
np.savez(startindsfile, inds=inds)
else:
inds = np.load(startindsfile)['inds']
startptsfile = 'NTX-starting-ll.npz'
if not os.path.exists(startptsfile):
loc_shelftransport = '/home/kthyng/projects/shelf_transport/'
drifters = netCDF.Dataset(loc_shelftransport + 'tracks/2004-01-01T00gc.nc')
xg = drifters.variables['xg'][:]; yg = drifters.variables['yg'][:]
xg = xg[inds,0]; yg = yg[inds,0]
lon0, lat0, _ = tracpy.tools.interpolate2d(xg, yg, tp.grid, 'm_ij2ll')
np.savez(startptsfile, lon0=lon0, lat0=lat0)
else:
d = np.load(startptsfile)
lon0 = d['lon0']; lat0 = d['lat0']
d.close()
# equal weightings for drifters for transport.
T0 = np.ones(lon0.size, order='F')
U = np.ma.zeros(tp.grid['xu'].shape, order='F')
V = np.ma.zeros(tp.grid['xv'].shape, order='F')
# pdb.set_trace()
return tp, lon0, lat0, T0, U, V
def check_points(lon0, lat0, grid, z0=None, nobays=False):
"""
Eliminate starting locations for drifters that are outside numerical
domain and that are masked out. If provided an array of starting vertical
locations in z0, it checks whether these points are at least 1m above the
bottom.
Args:
lon0,lat0: Starting locations for drifters in lon/lat
z0: Starting locations for drifters in z
grid: Grid made from readgrid.py
nobays: Whether to use points in bays or not. Default is False.
Returns:
* lon0,lat0 - Fixed lon0,lat0
* z0 - Fixed z0
"""
lonr = grid.lon_rho
latr = grid.lat_rho
# make interpolation function for water depth h.
# Used to check if float is above the bottom
if z0 is not None:
from scipy.interpolate import interp2d
h = grid.h
hint = interp2d(lonr[:, 1], latr[1, :], h.T, fill_value=np.nan)
# If covering the whole domain, need to exclude points outside domain.
# Use info just inside domain so points aren't right at the edge.
xvert = np.hstack((np.flipud(lonr[1, :]), lonr[:, 1], lonr[-2, :],
np.flipud(lonr[:, -2])))
yvert = np.hstack((np.flipud(latr[1, :]), latr[:, 1], latr[-2, :],
np.flipud(latr[:, -2])))
verts = np.vstack((xvert, yvert))
# Form path
path = Path(verts.T)
# Loop through particle tracks to eliminate drifters that start outside
# the domain
if lon0.ndim == 2:
for jd in range(lon0.shape[0]): # loop through drifters
for it in range(lon0.shape[1]):
# if drifter is not inside path, nan out this and all
# subsequent points
if not path.contains_point(np.vstack((lon0[jd, it],
lat0[jd, it]))):
lon0[jd, it] = np.nan
lat0[jd, it] = np.nan
if z0 is not None:
z0[jd, it] = np.nan
if z0 is not None:
# check that the drifter starts above the bottom
if z0[jd, it] <= -1*hint(lon0[jd, it], lat0[jd, it]):
lon0[jd, it] = np.nan
lat0[jd, it] = np.nan
if z0 is not None:
z0[jd, it] = np.nan
elif lon0.ndim == 1:
for jd in range(lon0.shape[0]): # loop through drifters
# if drifter is not inside path, nan out this and all
# subsequent points
if not path.contains_point(np.vstack((lon0[jd], lat0[jd]))):
lon0[jd] = np.nan
lat0[jd] = np.nan
if z0 is not None:
z0[jd] = np.nan
if z0 is not None:
# check that the drifter starts above the bottom
if z0[jd] <= -1*hint(lon0[jd], lat0[jd]):
lon0[jd] = np.nan
lat0[jd] = np.nan
z0[jd] = np.nan
# Also nan out points that are masked
fmask = mtri.LinearTriInterpolator(grid.trirllrho, grid.mask.flatten())
mask0 = fmask(lon0, lat0) # mask for lon0/lat0 points
ind1 = (mask0 == 1.) # indices select out where points are masked
# If nobays, eliminate points with shallow bathymetry
if nobays:
fh = grid.trirllrho.nn_interpolator(grid.h.flatten())
h0 = fh(lon0, lat0)
ind2 = (h0 > 10.)
else:
ind2 = np.ones(ind1.shape).astype(bool)
ind2 = ~np.isnan(lon0)*ind1*ind2
L = len(ind2.ravel())
Lnan = sum(ind2.ravel())
print L - Lnan, '/', L, ' drifters NaN-ed out.'
lon0 = lon0[ind2].flatten()
lat0 = lat0[ind2].flatten()
if z0 is not None:
z0 = z0[ind2].flatten()
if z0 is None:
return lon0, lat0
else:
return lon0, lat0, z0
# Get necessary info from File
# d = np.load(File)
# xg = d['xg']; yg = d['yg']; tg = d['tg']
d = netCDF.Dataset(File)
xg = d.variables['xg'][:]
yg = d.variables['yg'][:]
tg = d.variables['tp'][:]
d.close()
# Load in coastline region info
# Mexico
dconn = np.load('calcs/MXpts.npz')
xp = np.asarray(dconn['MX'])[:, 0]
yp = np.asarray(dconn['MX'])[:, 1]
MXpath = Path(np.vstack((xp, yp)).T)
dconn.close()
# S Texas
dconn = np.load('calcs/STXpts.npz')
xp = np.asarray(dconn['STX'])[:, 0]
yp = np.asarray(dconn['STX'])[:, 1]
STXpath = Path(np.vstack((xp, yp)).T)
dconn.close()
# N Texas
dconn = np.load('calcs/NTXpts.npz')
xp = np.asarray(dconn['NTX'])[:, 0]
yp = np.asarray(dconn['NTX'])[:, 1]
NTXpath = Path(np.vstack((xp, yp)).T)
dconn.close()
# Chenier
dconn = np.load('calcs/CHpts.npz')
