"""
Dumping ground for general utilities
"""
import os
import shutil
import glob
import inspect
from collections import OrderedDict
import warnings
import itertools
import logging
import astropy.io.fits as pyfits
import astropy.wcs as pywcs
import astropy.table
import numpy as np
import astropy.units as u
from sregion import SRegion, patch_from_polygon
from . import GRIZLI_PATH
from .constants import JWST_DQ_FLAGS, KMS, FLAMBDA_CGS, FNU_CGS
# character to skip clearing line on STDOUT printing
NO_NEWLINE = "\x1b[1A\x1b[1M"
# R_V for Galactic extinction
MW_RV = 3.1
MPL_COLORS = {
"b": "#1f77b4",
"orange": "#ff7f0e",
"g": "#2ca02c",
"r": "#d62728",
"purple": "#9467bd",
"brown": "#8c564b",
"pink": "#e377c2",
"gray": "#7f7f7f",
"olive": "#bcbd22",
"cyan": "#17becf",
}
# sns.color_palette("husl", 8)
SNS_HUSL = {
"r": (0.9677975592919913, 0.44127456009157356, 0.5358103155058701),
"orange": (0.8087954113106306, 0.5634700050056693, 0.19502642696727285),
"olive": (0.5920891529639701, 0.6418467016378244, 0.1935069134991043),
"g": (0.19783576093349015, 0.6955516966063037, 0.3995301037444499),
"sea": (0.21044753832183283, 0.6773105080456748, 0.6433941168468681),
"b": (0.22335772267769388, 0.6565792317435265, 0.8171355503265633),
"purple": (0.6423044349219739, 0.5497680051256467, 0.9582651433656727),
"pink": (0.9603888539940703, 0.3814317878772117, 0.8683117650835491),
}
GRISM_COLORS = {
"G800L": (0.0, 0.4470588235294118, 0.6980392156862745),
"G102": (0.0, 0.6196078431372549, 0.45098039215686275),
"G141": (0.8352941176470589, 0.3686274509803922, 0.0),
"none": (0.8, 0.4745098039215686, 0.6549019607843137),
"G150": "k",
"F277W": (0.0, 0.6196078431372549, 0.45098039215686275),
"F356W": (0.8352941176470589, 0.3686274509803922, 0.0),
"F444W": (0.8, 0.4745098039215686, 0.6549019607843137),
"F250M": "lightblue",
"F300M": "steelblue",
"F335M": "cornflowerblue",
"F360M": "royalblue",
"F410M": (0.0, 0.4470588235294118, 0.6980392156862745),
"F430M": "sandybrown",
"F460M": "lightsalmon",
"F480M": "coral",
"G280": "purple",
"F090W": (0.0, 0.4470588235294118, 0.6980392156862745),
"F115W": (0.0, 0.6196078431372549, 0.45098039215686275),
"F150W": (0.8352941176470589, 0.3686274509803922, 0.0),
"F140M": (0.8352941176470589, 0.3686274509803922, 0.0),
"F158M": (0.8352941176470589, 0.3686274509803922, 0.0),
"F200W": (0.8, 0.4745098039215686, 0.6549019607843137),
"F140M": "orange",
"BLUE": "#1f77b4", # Euclid
"RED": "#d62728",
"CLEARP": "b",
}
GRISM_MAJOR = {
"G102": 0.1,
"G141": 0.1, # WFC3/IR
"G800L": 0.1, # ACS/WFC
"F090W": 0.1,
"F115W": 0.1,
"F150W": 0.1, # NIRISS
"F140M": 0.1,
"F158M": 0.1,
"F200W": 0.1,
"F277W": 0.2,
"F356W": 0.2,
"F444W": 0.2, # NIRCam
"F250M": 0.1,
"F300M": 0.1,
"F335M": 0.1,
"F360M": 0.1,
"F410M": 0.1,
"F430M": 0.1,
"F460M": 0.1,
"F480M": 0.1,
"BLUE": 0.1,
"RED": 0.1, # Euclid
"GRISM": 0.1,
"G150": 0.1, # Roman
}
GRISM_LIMITS = {
"G800L": [0.545, 1.02, 40.0], # ACS/WFC
"G280": [0.2, 0.4, 14], # WFC3/UVIS
"G102": [0.77, 1.18, 23.0], # WFC3/IR
"G141": [1.06, 1.73, 46.0],
"GRISM": [0.98, 1.98, 11.0], # WFIRST/Roman
"G150": [0.98, 1.98, 11.0],
"F090W": [0.76, 1.04, 45.0], # NIRISS
"F115W": [0.97, 1.32, 45.0],
"F140M": [1.28, 1.52, 45.0],
"F158M": [1.28, 1.72, 45.0],
"F150W": [1.28, 1.72, 45.0],
"F200W": [1.68, 2.30, 45.0],
"F140M": [1.20, 1.60, 45.0],
"CLEARP": [0.76, 2.3, 45.0],
"F277W": [2.5, 3.2, 20.0], # NIRCAM
"F356W": [3.05, 4.1, 20.0],
"F444W": [3.82, 5.08, 20],
"F250M": [2.4, 2.65, 20],
"F300M": [2.77, 3.23, 20],
"F335M": [3.1, 3.6, 20],
"F360M": [3.4, 3.85, 20],
"F410M": [3.8, 4.38, 20],
"F430M": [4.1, 4.45, 20],
"F460M": [4.5, 4.8, 20],
"F480M": [4.6, 5.05, 20],
"BLUE": [0.8, 1.2, 10.0], # Euclid
"RED": [1.1, 1.9, 14.0],
}
# DEFAULT_LINE_LIST = ['PaB', 'HeI-1083', 'SIII', 'OII-7325', 'ArIII-7138', 'SII', 'Ha+NII', 'OI-6302', 'HeI-5877', 'OIII', 'Hb', 'OIII-4363', 'Hg', 'Hd', 'H8','H9','NeIII-3867', 'OII', 'NeVI-3426', 'NeV-3346', 'MgII','CIV-1549', 'CIII-1908', 'OIII-1663', 'HeII-1640', 'NIII-1750', 'NIV-1487', 'NV-1240', 'Lya']
# Line species for determining individual line fluxes. See `load_templates`.
DEFAULT_LINE_LIST = ['BrA','BrB','BrG','PfG','PfD',
'PaA','PaB','PaG','PaD',
'HeI-1083', 'SIII', 'OII-7325', 'ArIII-7138',
'SII', 'Ha', 'OI-6302', 'HeI-5877', 'OIII', 'Hb',
'OIII-4363', 'Hg', 'Hd', 'H7', 'H8', 'H9', 'H10',
'NeIII-3867', 'OII', 'NeVI-3426', 'NeV-3346', 'MgII',
'CIV-1549', 'CIII-1906', 'CIII-1908', 'OIII-1663',
'HeII-1640', 'NIII-1750', 'NIV-1487', 'NV-1240', 'Lya']
LSTSQ_RCOND = None
# Clipping threshold for BKG extensions in drizzle_from_visit
# BKG_CLIP = [scale, percentile_lo, percentile_hi]
# BKG_CLIP = [2, 1, 99]
BKG_CLIP = None
[docs]def set_warnings(numpy_level="ignore", astropy_level="ignore"):
"""
Set global numpy and astropy warnings
Parameters
----------
numpy_level : {'ignore', 'warn', 'raise', 'call', 'print', 'log'}
Numpy error level (see `~numpy.seterr`).
astropy_level : {'error', 'ignore', 'always', 'default', 'module', 'once'}
Astropy error level (see `~warnings.simplefilter`).
"""
from astropy.utils.exceptions import AstropyWarning
np.seterr(all=numpy_level)
warnings.simplefilter(astropy_level, category=AstropyWarning)
JWST_TRANSLATE = {
"RA_TARG": "TARG_RA",
"DEC_TARG": "TARG_DEC",
"EXPTIME": "EFFEXPTM",
"PA_V3": "ROLL_REF",
}
[docs]def get_flt_info(
files=[],
columns=[
"FILE",
"FILTER",
"PUPIL",
"INSTRUME",
"DETECTOR",
"TARGNAME",
"DATE-OBS",
"TIME-OBS",
"EXPSTART",
"EXPTIME",
"PA_V3",
"RA_TARG",
"DEC_TARG",
"POSTARG1",
"POSTARG2",
],
translate=JWST_TRANSLATE,
defaults={"PUPIL": "---", "TARGNAME": "indef", "PA_V3": 0.0},
jwst_detector=True,
):
"""
Extract header information from a list of FLT files
Parameters
----------
files : list, optional
List of exposure filenames.
If not provided,
it will search for all "*flt.fits" files in the current directory.
columns : list, optional
List of header keywords to extract from the FITS files.
The default columns include:
- "FILE": Filename
- "FILTER": Filter used
- "PUPIL": Pupil element used
- "INSTRUME": Instrument name
- "DETECTOR": Detector name
- "TARGNAME": Target name
- "DATE-OBS": Observation date
- "TIME-OBS": Observation time
- "EXPSTART": Exposure start time
- "EXPTIME": Exposure time
- "PA_V3": Position angle of V3 axis
- "RA_TARG": Right ascension of target
- "DEC_TARG": Declination of target
- "POSTARG1": Post-slew offset in axis 1
- "POSTARG2": Post-slew offset in axis 2
translate : dict, optional
Dictionary mapping header keywords to their corresponding FITS keywords.
Default is JWST_TRANSLATE.
defaults : dict, optional
Dictionary of default values for header keywords
that are not present in the FITS files. Default values include:
- "PUPIL": "---"
- "TARGNAME": "indef"
- "PA_V3": 0.0
jwst_detector : bool, optional
Flag indicating whether the FITS files are from JWST detectors.
Default is True.
Returns
-------
tab : `~astropy.table.Table`
Table containing header keywords extracted from the FITS files.
"""
import astropy.io.fits as pyfits
from astropy.table import Table
if not files:
files = glob.glob("*flt.fits")
N = len(files)
data = []
for c in columns[2:]:
if c not in translate:
translate[c] = "xxxxxxxxxxxxxx"
targprop = []
for i in range(N):
line = [os.path.basename(files[i]).split(".gz")[0]]
if files[i].endswith(".gz"):
im = pyfits.open(files[i])
h = im[0].header.copy()
im.close()
else:
h = pyfits.Header().fromfile(files[i])
if os.path.basename(files[i]).startswith("jw0"):
with pyfits.open(files[i]) as _im:
h1 = _im["SCI"].header
if "PA_V3" in h1:
h["PA_V3"] = h1["PA_V3"]
if "TARGPROP" in h:
targprop.append(h["TARGPROP"].lower())
else:
targprop.append("indef")
else:
targprop.append("indef")
filt = parse_filter_from_header(h, jwst_detector=jwst_detector)
line.append(filt)
has_columns = ["FILE", "FILTER"]
for key in columns[2:]:
has_columns.append(key)
if key in h:
line.append(h[key])
elif translate[key] in h:
line.append(h[translate[key]])
else:
if key in defaults:
line.append(defaults[key])
else:
line.append(np.nan)
continue
data.append(line)
tab = Table(rows=data, names=has_columns)
if "TARGNAME" in tab.colnames:
miss = tab["TARGNAME"] == ""
targs = [t.replace(" ", "-") for t in tab["TARGNAME"]]
if miss.sum() > 0:
for i in np.where(miss)[0]:
targs[i] = targprop[i] #'indef'
tab["TARGNAME"] = targs
return tab
[docs]def radec_to_targname(
ra=0,
dec=0,
round_arcsec=(4, 60),
precision=2,
targstr="j{rah}{ram}{ras}{sign}{ded}{dem}",
header=None,
):
"""
Turn decimal degree coordinates into a string with rounding.
Parameters
----------
ra, dec : float
Sky coordinates in decimal degrees
round_arcsec : (scalar, scalar)
Round the coordinates to nearest value of `round`, in arcseconds.
precision : int
Sub-arcsecond precision, in `~astropy.coordinates.SkyCoord.to_string`.
targstr : string
Build `targname` with this parent string. Arguments
`rah, ram, ras, rass, sign, ded, dem, des, dess` are computed from the
(rounded) target coordinates (`ra`, `dec`) and passed to
`targstr.format`.
header : `~astropy.io.fits.Header`, None
Try to get `ra`, `dec` from header keywords, first `CRVAL` and then
`RA_TARG`, `DEC_TARG`.
Returns
-------
targname : str
Target string, see the example above.
Examples
--------
>>> # Test dec: -10d10m10.10s
>>> dec = -10. - 10./60. - 10.1/3600
>>> # Test ra: 02h02m02.20s
>>> cosd = np.cos(dec/180*np.pi)
>>> ra = 2*15 + 2./60*15 + 2.2/3600.*15
>>> # Round to nearest arcmin
>>> from grizli.utils import radec_to_targname
>>> print(radec_to_targname(ra=ra, dec=dec, round_arcsec=(4,60),
targstr='j{rah}{ram}{ras}{sign}{ded}{dem}'))
j020204m1010 # (rounded to 4 arcsec in RA)
>>> # Full precision
>>> targstr = 'j{rah}{ram}{ras}.{rass}{sign}{ded}{dem}{des}.{dess}'
>>> print(radec_to_targname(ra, dec,round_arcsec=(0.0001, 0.0001),
precision=3, targstr=targstr))
j020202.200m101010.100
"""
import astropy.coordinates
import astropy.units as u
import re
import numpy as np
if header is not None:
if "CRVAL1" in header:
ra, dec = header["CRVAL1"], header["CRVAL2"]
else:
if "RA_TARG" in header:
ra, dec = header["RA_TARG"], header["DEC_TARG"]
cosd = np.cos(dec / 180 * np.pi)
scl = np.array(round_arcsec) / 3600 * np.array([360 / 24, 1])
dec_scl = int(np.round(dec / scl[1])) * scl[1]
ra_scl = int(np.round(ra / scl[0])) * scl[0]
coo = astropy.coordinates.SkyCoord(
ra=ra_scl * u.deg, dec=dec_scl * u.deg, frame="icrs"
)
cstr = re.split("[hmsd.]", coo.to_string("hmsdms", precision=precision))
# targname = ('j{0}{1}'.format(''.join(cstr[0:3]), ''.join(cstr[4:7])))
# targname = targname.replace(' ', '').replace('+','p').replace('-','m')
rah, ram, ras, rass = cstr[0:4]
ded, dem, des, dess = cstr[4:8]
sign = "p" if ded[1] == "+" else "m"
targname = targstr.format(
rah=rah,
ram=ram,
ras=ras,
rass=rass,
ded=ded[2:],
dem=dem,
des=des,
dess=dess,
sign=sign,
)
return targname
[docs]def blot_nearest_exact(
in_data,
in_wcs,
out_wcs,
verbose=True,
stepsize=-1,
scale_by_pixel_area=False,
wcs_mask=True,
fill_value=0,
):
"""
Own blot function for blotting exact pixels without rescaling for input
and output pixel size
Parameters
----------
in_data : `~numpy.ndarray`
Input data to blot.
in_wcs : `~astropy.wcs.WCS`
Input WCS. Must have _naxis1, _naxis2 or pixel_shape attributes.
out_wcs : `~astropy.wcs.WCS`
Output WCS. Must have _naxis1, _naxis2 or pixel_shape attributes.
verbose : bool, optional
If True, print information about the overlap. Default is True.
stepsize : int, optional
Step size for interpolation. If set to <=1, the function will use the explicit
wcs mapping ``out_wcs.all_pix2world > in_wcs.all_world2pix``. If > 1,
will use
``astrodrizzle.DefaultWCSMapping(out_wcs, in_wcs, nx, ny, stepsize)``.
scale_by_pixel_area : bool
If True, then scale the output image by the square of the image pixel
scales (out**2/in**2), i.e., the pixel areas.
wcs_mask : bool
Use fast WCS masking. If False, use ``regions``.
fill_value : int/float
Value in ``out_data`` not covered by ``in_data``.
Returns
-------
out_data : `~numpy.ndarray`
Blotted data.
"""
from regions import Regions
from shapely.geometry import Polygon
import scipy.ndimage as nd
from drizzlepac import cdriz
try:
from .utils_numba.interp import pixel_map_c
except ImportError:
from grizli.utils_numba.interp import pixel_map_c
# Shapes, in numpy array convention (y, x)
if hasattr(in_wcs, "pixel_shape"):
in_sh = in_wcs.pixel_shape[::-1]
elif hasattr(in_wcs, "array_shape"):
in_sh = in_wcs.array_shape
else:
in_sh = (in_wcs._naxis2, in_wcs._naxis1)
if hasattr(out_wcs, "pixel_shape"):
out_sh = out_wcs.pixel_shape[::-1]
elif hasattr(out_wcs, "array_shape"):
out_sh = out_wcs.array_shape
else:
out_sh = (out_wcs._naxis2, out_wcs._naxis1)
in_px = in_wcs.calc_footprint()
in_poly = Polygon(in_px).buffer(5.0 / 3600.0)
out_px = out_wcs.calc_footprint()
out_poly = Polygon(out_px).buffer(5.0 / 3600)
olap = in_poly.intersection(out_poly)
if olap.area == 0:
if verbose:
print("No overlap")
return np.zeros(out_sh)
# Region mask for speedup
if np.isclose(olap.area, out_poly.area, 0.01):
mask = np.ones(out_sh, dtype=bool)
elif wcs_mask:
# Use wcs / Path
from matplotlib.path import Path
out_xy = out_wcs.all_world2pix(np.array(in_poly.exterior.xy).T, 0) - 0.5
out_xy_path = Path(out_xy)
yp, xp = np.indices(out_sh)
pts = np.array([xp.flatten(), yp.flatten()]).T
mask = out_xy_path.contains_points(pts).reshape(out_sh)
else:
olap_poly = np.array(olap.exterior.xy)
poly_reg = (
"fk5\npolygon("
+ ",".join(["{0}".format(p + 1) for p in olap_poly.T.flatten()])
+ ")\n"
)
reg = Regions.parse(poly_reg, format="ds9")[0]
mask = reg.to_mask().to_image(shape=out_sh)
# yp, xp = np.indices(in_data.shape)
# xi, yi = xp[mask], yp[mask]
yo, xo = np.where(mask > 0)
if stepsize <= 1:
rd = out_wcs.all_pix2world(xo, yo, 0)
xf, yf = in_wcs.all_world2pix(rd[0], rd[1], 0)
else:
# Seems backwards and doesn't quite agree with above
blot_wcs = out_wcs
source_wcs = in_wcs
if hasattr(blot_wcs, "pixel_shape"):
nx, ny = blot_wcs.pixel_shape
else:
nx, ny = int(blot_wcs._naxis1), int(blot_wcs._naxis2)
mapping = cdriz.DefaultWCSMapping(blot_wcs, source_wcs, nx, ny, stepsize)
xf, yf = mapping(xo, yo)
xi, yi = np.asarray(np.round(xf),dtype=int), np.asarray(np.round(yf),dtype=int)
m2 = (xi >= 0) & (yi >= 0) & (xi < in_sh[1]) & (yi < in_sh[0])
xi, yi, xf, yf, xo, yo = xi[m2], yi[m2], xf[m2], yf[m2], xo[m2], yo[m2]
out_data = np.ones(out_sh, dtype=np.float64)*fill_value
status = pixel_map_c(np.asarray(in_data,dtype=np.float64), xi, yi, out_data, xo, yo)
# Fill empty
func = nd.maximum_filter
fill = out_data == 0
filtered = func(out_data, size=5)
out_data[fill] = filtered[fill]
if scale_by_pixel_area:
in_scale = get_wcs_pscale(in_wcs)
out_scale = get_wcs_pscale(out_wcs)
out_data *= out_scale ** 2 / in_scale ** 2
return out_data.astype(in_data.dtype)
def _slice_ndfilter(data, filter_func, slices, args, size, footprint, kwargs):
"""
Helper function passing image slices to `scipy.ndimage` filters that is
pickleable for threading with `multiprocessing`
Parameters
----------
data, filter_func, args, size, footprint :
See `multiprocessing_ndfilter`
slices : (slice, slice, slice, slice)
Array slices for insert a cutout back into a larger parent array
Returns
-------
filtered : array-like
Filtered data
slices : tuple
`slices` as input
"""
filtered = filter_func(data, *args, size=size, footprint=footprint, **kwargs)
return filtered, slices
[docs]def multiprocessing_ndfilter(
data,
filter_func,
filter_args=(),
size=None,
footprint=None,
cutout_size=256,
n_proc=4,
timeout=90,
mask=None,
verbose=True,
**kwargs,
):
"""
Cut up a large array and send slices to `scipy.ndimage` filters
Parameters
----------
data : array-like
Main image array
filter_func : function
Filtering function, e.g., `scipy.ndimage.median_filter`
filter_args : tuple
Arguments to pass to `filter_func`
size, footprint : int, array-like
Filter size or footprint, see, e.g., `scipy.ndimage.median_filter`
cutout_size : int
Size of subimage cutouts
n_proc : int
Number of `multiprocessing` processes to use
timeout : float
`multiprocessing` timeout (seconds)
mask : array-like
Array multiplied to `data` that can zero-out regions to ignore
verbose : bool
Print status messages
kwargs : dict
Keyword arguments passed through to `filter_func`
Returns
-------
filtered : array-like
Filtered version of `data`
Examples
--------
>>> import time
>>> import numpy as np
>>> import scipy.ndimage as nd
>>> from grizli.utils import multiprocessing_ndfilter
>>> rnd = np.random.normal(size=(512,512))
>>> t0 = time.time()
>>> f_serial = nd.median_filter(rnd, size=10)
>>> t1 = time.time()
>>> f_mp = multiprocessing_ndfilter(rnd, nd.median_filter, size=10,
>>> cutout_size=256, n_proc=4)
>>> t2 = time.time()
>>> np.allclose(f_serial, f_mp)
True
>>> print(f' serial: {(t1-t0)*1000:.1f} ms')
>>> print(f'parallel: {(t2-t1)*1000:.1f} ms')
serial: 573.9 ms
parallel: 214.8 ms
"""
import multiprocessing as mp
try:
from tqdm import tqdm
except ImportError:
verbose = False
sh = data.shape
msg = None
if cutout_size > np.max(sh):
msg = f"cutout_size={cutout_size} greater than image dimensions, run "
msg += f"`{filter_func}` directly"
elif n_proc == 0:
msg = f"n_proc = 0, run in a single command"
if msg is not None:
if verbose:
print(msg)
filtered = filter_func(data, *filter_args, size=size, footprint=footprint)
return filtered
# Grid size
nx = data.shape[1] // cutout_size + 1
ny = data.shape[0] // cutout_size + 1
# Padding
if footprint is not None:
fpsh = footprint.shape
pad = np.max(fpsh)
elif size is not None:
pad = size
else:
raise ValueError("Either size or footprint must be specified")
if n_proc < 0:
n_proc = mp.cpu_count()
n_proc = np.minimum(n_proc, mp.cpu_count())
pool = mp.Pool(processes=n_proc)
jobs = []
if mask is not None:
data_mask = data * mask
else:
data_mask = data
# Make image slices
for i in range(nx):
xmi = np.maximum(0, i * cutout_size - pad)
xma = np.minimum(sh[1], (i + 1) * cutout_size + pad)
# print(i, xmi, xma)
if i == 0:
slx = slice(0, cutout_size)
x0 = 0
elif i < nx - 1:
slx = slice(pad, cutout_size + pad)
x0 = i * cutout_size
else:
slx = slice(pad, cutout_size + 1)
x0 = xmi + pad
nxs = slx.stop - slx.start
oslx = slice(x0, x0 + nxs)
for j in range(ny):
ymi = np.maximum(0, j * cutout_size - pad)
yma = np.minimum(sh[0], (j + 1) * cutout_size + pad)
if j == 0:
sly = slice(0, cutout_size)
y0 = 0
elif j < ny - 1:
sly = slice(pad, cutout_size + pad)
y0 = j * cutout_size
else:
sly = slice(pad, cutout_size + 1)
y0 = ymi + pad
nys = sly.stop - sly.start
osly = slice(y0, y0 + nys)
cut = data_mask[ymi:yma, xmi:xma]
if cut.max() == 0:
# print(f'Skip {xmi} {xma} {ymi} {yma}')
continue
# Make jobs for filtering the image slices
slices = (osly, oslx, sly, slx)
_args = (cut, filter_func, slices, filter_args, size, footprint, kwargs)
jobs.append(pool.apply_async(_slice_ndfilter, _args))
# Collect results
pool.close()
filtered = np.zeros_like(data)
if verbose:
_iter = tqdm(jobs)
else:
_iter = jobs
for res in _iter:
filtered_i, slices = res.get(timeout=timeout)
filtered[slices[:2]] += filtered_i[slices[2:]]
return filtered
[docs]def parse_flt_files(
files=[],
info=None,
uniquename=False,
use_visit=False,
get_footprint=False,
translate={
"AEGIS-": "aegis-",
"COSMOS-": "cosmos-",
"GNGRISM": "goodsn-",
"GOODS-SOUTH-": "goodss-",
"UDS-": "uds-",
},
visit_split_shift=1.5,
max_dt=1e9,
path="../RAW",
):
"""
Read header information from a list of exposures and parse out groups
based on filter/target/orientation.
Parameters
----------
files : list, optional
List of exposure filenames. If not specified, will use ``*flt.fits``.
info : None or `~astropy.table.Table`, optional
Output from `~grizli.utils.get_flt_info`.
uniquename : bool, optional
If True, then split everything by program ID and visit name. If
False, then just group by targname/filter/pa_v3.
use_visit : bool, optional
For parallel observations with ``targname='ANY'``, use the filename
up to the visit ID as the target name. For example:
>>> flc = 'jbhj64d8q_flc.fits'
>>> visit_targname = flc[:6]
>>> print(visit_targname)
jbhj64
If False, generate a targname for parallel observations based on the
pointing coordinates using `radec_to_targname`. Use this keyword
for dithered parallels like 3D-HST / GLASS but set to False for
undithered parallels like WISP. Should also generally be used with
``uniquename=False`` otherwise generates names that are a bit
redundant:
+--------------+---------------------------+
| `uniquename` | Output Targname |
+==============+===========================+
| True | jbhj45-bhj-45-180.0-F814W |
+--------------+---------------------------+
| False | jbhj45-180.0-F814W |
+--------------+---------------------------+
get_footprint : bool, optional
If True, get the visit footprint from FLT WCS.
translate : dict, optional
Translation dictionary to modify TARGNAME keywords to some other
value. Used like:
>>> targname = 'GOODS-SOUTH-10'
>>> translate = {'GOODS-SOUTH-': 'goodss-'}
>>> for k in translate:
>>> targname = targname.replace(k, translate[k])
>>> print(targname)
goodss-10
visit_split_shift : float, optional
Separation in ``arcmin`` beyond which exposures in a group are split
into separate visits.
max_dt : float, optional
Maximum time separation between exposures in a visit, in seconds.
path : str, optional
PATH to search for `flt` files if ``info`` not provided
Returns
-------
output_list : dict
Dictionary split by target/filter/pa_v3. Keys are derived visit
product names and values are lists of exposure filenames corresponding
to that set. Keys are generated with the formats like:
>>> targname = 'macs1149+2223'
>>> pa_v3 = 32.0
>>> filter = 'f140w'
>>> flt_filename = 'ica521naq_flt.fits'
>>> propstr = flt_filename[1:4]
>>> visit = flt_filename[4:6]
>>> # uniquename = False
>>> print('{0}-{1:05.1f}-{2}'.format(targname, pa_v3, filter))
macs1149.6+2223-032.0-f140w
>>> # uniquename = True
>>> print('{0}-{1:3s}-{2:2s}-{3:05.1f}-{4:s}'.format(targname, propstr, visit, pa_v3, filter))
macs1149.6+2223-ca5-21-032.0-f140w
filter_list : dict
Nested dictionary split by filter and then PA_V3. This shouldn't
be used if exposures from completely disjoint pointings are stored
in the same working directory.
"""
if info is None:
if not files:
files = glob.glob(os.path.join(path), "*flt.fits")
if len(files) == 0:
return False
info = get_flt_info(files)
else:
info = info.copy()
for c in info.colnames:
if not c.islower():
info.rename_column(c, c.lower())
if "expstart" not in info.colnames:
info["expstart"] = info["exptime"] * 0.0
so = np.argsort(info["expstart"])
info = info[so]
# pa_v3 = np.round(info['pa_v3']*10)/10 % 360.
pa_v3 = np.round(np.round(info["pa_v3"], decimals=1)) % 360.0
target_list = []
for i in range(len(info)):
# Replace ANY targets with JRhRmRs-DdDmDs
if info["targname"][i] == "ANY":
if use_visit:
new_targname = info["file"][i][:6]
else:
new_targname = "par-" + radec_to_targname(
ra=info["ra_targ"][i], dec=info["dec_targ"][i]
)
target_list.append(new_targname.lower())
else:
target_list.append(info["targname"][i])
target_list = np.array(target_list)
_prog_ids = []
visits = []
for file in info["file"]:
bfile = os.path.basename(file)
if bfile.startswith("jw"):
_prog_ids.append(bfile[2:7])
visits.append(bfile[7:10])
else:
_prog_ids.append(bfile[1:4])
visits.append(bfile[4:6])
visits = np.array(visits)
info["progIDs"] = _prog_ids
progIDs = np.unique(info["progIDs"])
dates = np.array(["".join(date.split("-")[1:]) for date in info["date-obs"]])
targets = np.unique(target_list)
output_list = [] # OrderedDict()
filter_list = OrderedDict()
for filter in np.unique(info["filter"]):
filter_list[filter] = OrderedDict()
angles = np.unique(pa_v3[(info["filter"] == filter)])
for angle in angles:
filter_list[filter][angle] = []
for target in targets:
# 3D-HST targname translations
target_use = target
for key in translate.keys():
target_use = target_use.replace(key, translate[key])
# pad i < 10 with zero
for key in translate.keys():
if translate[key] in target_use:
spl = target_use.split("-")
try:
if (int(spl[-1]) < 10) & (len(spl[-1]) == 1):
spl[-1] = "{0:02d}".format(int(spl[-1]))
target_use = "-".join(spl)
except:
pass
for filter in np.unique(info["filter"][(target_list == target)]):
angles = np.unique(
pa_v3[(info["filter"] == filter) & (target_list == target)]
)
for angle in angles:
exposure_list = []
exposure_start = []
product = "{0}-{1:05.1f}-{2}".format(target_use, angle, filter)
visit_match = np.unique(
visits[(target_list == target) & (info["filter"] == filter)]
)
this_progs = []
this_visits = []
for visit in visit_match:
ix = (visits == visit) & (target_list == target)
ix &= info["filter"] == filter
# this_progs.append(info['progIDs'][ix][0])
# print visit, ix.sum(), np.unique(info['progIDs'][ix])
new_progs = list(np.unique(info["progIDs"][ix]))
this_visits.extend([visit] * len(new_progs))
this_progs.extend(new_progs)
for visit, prog in zip(this_visits, this_progs):
visit_list = []
visit_start = []
_vstr = "{0}-{1}-{2}-{3:05.1f}-{4}"
visit_product = _vstr.format(target_use, prog, visit, angle, filter)
use = target_list == target
use &= info["filter"] == filter
use &= visits == visit
use &= pa_v3 == angle
use &= info["progIDs"] == prog
if use.sum() == 0:
continue
for tstart, file in zip(info["expstart"][use], info["file"][use]):
f = file.split(".gz")[0]
if f not in exposure_list:
visit_list.append(str(f))
visit_start.append(tstart)
exposure_list = np.append(exposure_list, visit_list)
exposure_start.extend(visit_start)
filter_list[filter][angle].extend(visit_list)
if uniquename:
print(visit_product, len(visit_list))
so = np.argsort(visit_start)
exposure_list = np.array(visit_list)[so]
# output_list[visit_product.lower()] = visit_list
d = OrderedDict(
product=str(visit_product.lower()),
files=list(np.array(visit_list)[so]),
)
output_list.append(d)
if not uniquename:
print(product, len(exposure_list))
so = np.argsort(exposure_start)
exposure_list = np.array(exposure_list)[so]
# output_list[product.lower()] = exposure_list
d = OrderedDict(
product=str(product.lower()),
files=list(np.array(exposure_list)[so]),
)
output_list.append(d)
# Split large shifts
if visit_split_shift > 0:
split_list = []
for o in output_list:
_spl = split_visit(
o, path=path, max_dt=max_dt, visit_split_shift=visit_split_shift
)
split_list.extend(_spl)
output_list = split_list
# Get visit footprint from FLT WCS
if get_footprint:
from shapely.geometry import Polygon
N = len(output_list)
for i in range(N):
for j in range(len(output_list[i]["files"])):
flt_file = output_list[i]["files"][j]
if not os.path.exists(flt_file):
for gzext in ["", ".gz"]:
_flt_file = os.path.join(path, flt_file + gzext)
if os.path.exists(_flt_file):
flt_file = _flt_file
break
flt_j = pyfits.open(flt_file)
h = flt_j[0].header
_ext = 0
if h["INSTRUME"] == "WFC3":
_ext = 1
if h["DETECTOR"] == "IR":
wcs_j = pywcs.WCS(flt_j["SCI", 1])
else:
wcs_j = pywcs.WCS(flt_j["SCI", 1], fobj=flt_j)
elif h["INSTRUME"] == "WFPC2":
_ext = 1
wcs_j = pywcs.WCS(flt_j["SCI", 1])
else:
_ext = 1
wcs_j = pywcs.WCS(flt_j["SCI", 1], fobj=flt_j)
if (wcs_j.pixel_shape is None) & ("NPIX1" in flt_j["SCI", 1].header):
_h = flt_j["SCI", 1].header
wcs_j.pixel_shape = (_h["NPIX1"], _h["NPIX2"])
fp_j = Polygon(wcs_j.calc_footprint())
if j == 0:
fp_i = fp_j.buffer(1.0 / 3600)
else:
fp_i = fp_i.union(fp_j.buffer(1.0 / 3600))
flt_j.close()
output_list[i]["footprint"] = fp_i
return output_list, filter_list
[docs]def split_visit(visit, visit_split_shift=1.5, max_dt=6.0 / 24, path="../RAW"):
"""
Check if files in a visit have large shifts and split them otherwise
Parameters
----------
visit : dict
The visit dictionary containing information about the visit.
visit_split_shift : float, optional
The threshold for splitting the visit if shifts are larger than
``visit_split_shift`` arcmin. Default is 1.5.
max_dt : float, optional
The maximum time difference between exposures in days.
Default is 6.0 / 24.
path : str, optional
The path to the directory containing the visit files.
Default is "../RAW".
Returns
-------
list of dict
A list of visit dictionaries, each representing a split visit.
"""
ims = []
for file in visit["files"]:
for gzext in ["", ".gz"]:
_file = os.path.join(path, file) + gzext
if os.path.exists(_file):
ims.append(pyfits.open(_file))
break
#ims = [pyfits.open(os.path.join(path, file)) for file in visit['files']]
crval1 = np.array([im[1].header['CRVAL1'] for im in ims])
crval2 = np.array([im[1].header['CRVAL2'] for im in ims])
expstart = np.array([im[0].header['EXPSTART'] for im in ims])
dt = np.asarray((expstart-expstart[0])/max_dt,dtype=int)
for im in ims:
im.close()
dx = (crval1 - crval1[0]) * 60 * np.cos(crval2[0] / 180 * np.pi)
dy = (crval2 - crval2[0]) * 60
dxi = np.asarray(np.round(dx/visit_split_shift),dtype=int)
dyi = np.asarray(np.round(dy/visit_split_shift),dtype=int)
keys = dxi*100+dyi+1000*dt
un = np.unique(keys)
if len(un) == 1:
return [visit]
else:
spl = visit["product"].split("-")
isJWST = spl[-1].lower().startswith("clear")
isJWST |= spl[-1].lower() in ["gr150r", "gr150c", "grismr", "grismc"]
if isJWST:
spl.insert(-2, "")
else:
spl.insert(-1, "")
visits = []
for i in range(len(un)):
ix = keys == un[i]
if isJWST:
spl[-3] = "abcdefghijklmnopqrsuvwxyz"[i]
else:
spl[-2] = "abcdefghijklmnopqrsuvwxyz"[i]
new_visit = {
"files": list(np.array(visit["files"])[ix]),
"product": "-".join(spl),
}
if "footprints" in visit:
new_visit["footprints"] = list(np.array(visit["footprints"])[ix])
visits.append(new_visit)
return visits
[docs]def parse_visit_overlaps(visits, buffer=15.0):
"""Find overlapping visits/filters to make combined mosaics
Parameters
----------
visits : list
Output list of visit information from `~grizli.utils.parse_flt_files`.
The script looks for files like `visits[i]['product']+'_dr?_sci.fits'`
to compute the WCS footprint of a visit. These are produced, e.g., by
`~grizli.prep.process_direct_grism_visit`.
buffer : float
Buffer, in `~astropy.units.arcsec`, to add around visit footprints to
look for overlaps.
Returns
-------
exposure_groups : list
List of overlapping visits, with similar format as input `visits`.
"""
import copy
from shapely.geometry import Polygon
N = len(visits)
exposure_groups = []
used = np.arange(len(visits)) < 0
for i in range(N):
f_i = visits[i]["product"].split("-")[-1]
if used[i]:
continue
if "footprint" in visits[i]:
fp_i = visits[i]["footprint"].buffer(buffer / 3600.0)
else:
_products = visits[i]["product"] + "_dr?_sci.fits"
im_i = pyfits.open(glob.glob(_products)[0])
wcs_i = pywcs.WCS(im_i[0])
fp_i = Polygon(wcs_i.calc_footprint()).buffer(buffer / 3600.0)
im_i.close()
exposure_groups.append(copy.deepcopy(visits[i]))
for j in range(i + 1, N):
f_j = visits[j]["product"].split("-")[-1]
if (f_j != f_i) | (used[j]):
continue
#
if "footprint" in visits[j]:
fp_j = visits[j]["footprint"].buffer(buffer / 3600.0)
else:
_products = visits[j]["product"] + "_dr?_sci.fits"
im_j = pyfits.open(glob.glob(_products)[0])
wcs_j = pywcs.WCS(im_j[0])
fp_j = Polygon(wcs_j.calc_footprint()).buffer(buffer / 3600.0)
im_j.close()
olap = fp_i.intersection(fp_j)
if olap.area > 0:
used[j] = True
fp_i = fp_i.union(fp_j)
exposure_groups[-1]["footprint"] = fp_i
exposure_groups[-1]["files"].extend(visits[j]["files"])
for i in range(len(exposure_groups)):
flt_i = pyfits.open(exposure_groups[i]["files"][0])
product = flt_i[0].header["TARGNAME"].lower()
if product == "any":
product = "par-" + radec_to_targname(header=flt_i["SCI", 1].header)
f_i = exposure_groups[i]["product"].split("-")[-1]
product += "-" + f_i
exposure_groups[i]["product"] = product
flt_i.close()
return exposure_groups
DIRECT_ORDER = {
"G102": [
"F105W","F110W","F098M","F125W","F140W","F160W","F127M","F139M","F153M",
"F132N","F130N","F128N","F126N","F164N","F167N",
],
"G141": [
"F140W","F160W","F125W","F105W","F110W","F098M","F127M","F139M","F153M",
"F132N","F130N","F128N","F126N","F164N","F167N",
],
"G800L": [
"F814W","F606W","F850LP","F775W","F435W","F105W","F110W","F098M","F125W",
"F140W","F160W","F127M","F139M","F153M","F132N","F130N","F128N","F126N",
"F164N","F167N",
],
"GR150C": ["F115W", "F150W", "F200W"],
"GR150R": ["F115W", "F150W", "F200W"],
}
[docs]def parse_grism_associations(exposure_groups, info,
best_direct=DIRECT_ORDER,
get_max_overlap=True):
"""Get associated lists of grism and direct exposures
Parameters
----------
exposure_grups : list
Output list of overlapping visits from
`~grizli.utils.parse_visit_overlaps`.
best_direct : dict
Dictionary of the preferred direct imaging filters to use with a
particular grism.
Returns
-------
grism_groups : list
List of dictionaries with associated 'direct' and 'grism' entries.
"""
N = len(exposure_groups)
grism_groups = []
for i in range(N):
_espi = exposure_groups[i]['product'].split('-')
if _espi[-2][0] in 'fg':
pupil_i = _espi[-2]
f_i = _espi[-1]
root_i = '-'.join(_espi[:-2])
else:
pupil_i = None
f_i = _espi[-1]
root_i = '-'.join(_espi[:-1])
if f_i.startswith('g'):
group = OrderedDict(grism=exposure_groups[i],
direct=None)
else:
continue
fp_i = exposure_groups[i]['footprint']
olap_i = 0.
d_i = f_i
d_idx = 10
for j in range(N):
_espj = exposure_groups[j]['product'].split('-')
if _espj[-2][0] in 'fg':
pupil_j = _espj[-2]
f_j = _espj[-1]
root_j = '-'.join(_espj[:-2])
else:
f_j = _espj[-1]
root_j = '-'.join(_espj[:-1])
pupil_j = None
if f_j.startswith('g'):
continue
fp_j = exposure_groups[j]['footprint']
olap = fp_i.intersection(fp_j)
if (root_j == root_i):
if pupil_i is not None:
if pupil_j == pupil_i:
group['direct'] = exposure_groups[j]
else:
continue
else:
if f_j.upper() not in best_direct[f_i.upper()]:
continue
if best_direct[f_i.upper()].index(f_j.upper()) < d_idx:
d_idx = best_direct[f_i.upper()].index(f_j.upper())
group['direct'] = exposure_groups[j]
olap_i = olap.area
d_i = f_j
grism_groups.append(group)
return grism_groups
[docs]def get_hst_filter(header, **kwargs):
"""
Deprecated: use `grizli.utils.parse_filter_from_header`
"""
result = parse_filter_from_header(header, **kwargs)
return result
EE_RADII = [0.1, 0.15, 0.2, 0.25, 0.3, 0.4, 0.5, 0.8, 1.0, 1.5, 2.0]
[docs]def get_filter_obsmode(
filter="f160w", acs_chip="wfc1", uvis_chip="uvis2", aper=np.inf, case=str.lower
):
"""
Derive `~pysynphot` obsmode keyword from a filter name, where UVIS filters
end in 'u'
Parameters
----------
filter : str, optional
The name of the filter. Default is "f160w".
acs_chip : str, optional
The ACS chip. Default is "wfc1".
uvis_chip : str, optional
The UVIS chip. Default is "uvis2".
aper : float, optional
The aperture size. Set to np.inf by default.
case : function, optional
The case conversion function. Default is str.lower.
Returns
-------
str
The `~pysynphot` obsmode keyword derived from the filter name.
"""
if filter.lower()[:2] in ["f0", "f1", "g1"]:
inst = "wfc3,ir"
else:
if filter.lower().endswith("u"):
inst = f"wfc3,{uvis_chip}"
else:
inst = f"acs,{acs_chip}"
obsmode = inst + "," + filter.strip("u").lower()
if np.isfinite(aper):
obsmode += f",aper#{aper:4.2f}"
return case(obsmode)
[docs]def tabulate_encircled_energy(aper_radii=EE_RADII, norm_radius=4.0):
"""
Tabulated encircled energy for different aperture radii
and normalization radius.
Parameters
----------
aper_radii : list, optional
List of aperture radii in arcseconds.
Default is [0.1, 0.15, 0.2, 0.25, 0.3, 0.4, 0.5, 0.8, 1.0, 1.5, 2.0].
norm_radius : float, optional
Normalization radius in arcseconds. Default is 4.0.
"""
import pysynphot as S
from .pipeline import default_params
# Default spectrum
sp = S.FlatSpectrum(25, fluxunits="ABMag")
tab = GTable()
tab["radius"] = aper_radii * u.arcsec
tab.meta["RNORM"] = norm_radius, "Normalization radius, arcsec"
# IR
for f in default_params.IR_M_FILTERS + default_params.IR_W_FILTERS:
obsmode = "wfc3,ir," + f.lower()
print(obsmode)
tab[obsmode] = synphot_encircled_energy(
obsmode=obsmode, sp=sp, aper_radii=aper_radii, norm_radius=norm_radius
)
tab.meta["ZP_{0}".format(obsmode)] = synphot_zeropoint(
obsmode=obsmode, radius=norm_radius
)
# Optical. Wrap in try/except to catch missing filters
for inst in ["acs,wfc1,", "wfc3,uvis2,"]:
for f in (
default_params.OPT_M_FILTERS
+ default_params.OPT_W_FILTERS
+ default_params.UV_M_FILTERS
+ default_params.UV_W_FILTERS
):
obsmode = inst + f.lower()
try:
tab[obsmode] = synphot_encircled_energy(
obsmode=obsmode,
sp=sp,
aper_radii=aper_radii,
norm_radius=norm_radius,
)
print(obsmode)
tab.meta["ZP_{0}".format(obsmode)] = synphot_zeropoint(
obsmode=obsmode, radius=norm_radius
)
except:
# Failed because obsmode not available in synphot
continue
tab.meta["PSYNVER"] = S.__version__, "Pysynphot version"
tab.write("hst_encircled_energy.fits", overwrite=True)
[docs]def synphot_zeropoint(obsmode="wfc3,ir,f160w", radius=4.0):
"""
Compute synphot for a specific aperture.
Parameters
----------
obsmode : str, optional
The observation mode string. Default is "wfc3,ir,f160w".
radius : float, optional
The radius of the aperture in arcseconds. Default is 4.0.
Returns
-------
ZP : float
The zero point magnitude calculated using synphot.
"""
import pysynphot as S
sp = S.FlatSpectrum(25, fluxunits="ABMag")
if np.isfinite(radius):
bp = S.ObsBandpass(obsmode + ",aper#{0:.2f}".format(radius))
else:
bp = S.ObsBandpass(obsmode)
obs = S.Observation(sp, bp)
ZP = 25 + 2.5 * np.log10(obs.countrate())
return ZP
[docs]def synphot_encircled_energy(
obsmode="wfc3,ir,f160w", sp="default", aper_radii=EE_RADII, norm_radius=4.0
):
"""
Compute encircled energy curves with pysynphot
Parameters
----------
obsmode : str
The observation mode string specifying the
instrument, detector, and filter.
sp : `pysynphot.spectrum.SourceSpectrum` or None, optional
The source spectrum to use for the calculation.
If None, a flat spectrum with a magnitude of 25 AB mag is used.
aper_radii : array-like, optional
The array of aperture radii in arcseconds at which to compute
the encircled energy.
Default is [0.1, 0.15, 0.2, 0.25, 0.3, 0.4, 0.5, 0.8, 1.0, 1.5, 2.0].
norm_radius : float, optional
The normalization radius in arcseconds.
The encircled energy at this radius will be used to normalize
the encircled energy curve. If set to np.inf, the normalization
will be performed using the full aperture. Default is 4.0.
Returns
-------
counts : array-like
The array of encircled energy counts normalized to the counts
at the normalization radius.
"""
import pysynphot as S
if sp == "default":
sp = S.FlatSpectrum(25, fluxunits="ABMag")
# Normalization
if np.isfinite(norm_radius):
bp = S.ObsBandpass(obsmode + ",aper#{0:.2f}".format(norm_radius))
else:
bp = S.ObsBandpass(obsmode)
obs = S.Observation(sp, bp)
norm_counts = obs.countrate()
counts = np.ones_like(aper_radii)
for i, r_aper in enumerate(aper_radii):
# print(obsmode, r_aper)
bp = S.ObsBandpass(obsmode + ",aper#{0:.2f}".format(r_aper))
obs = S.Observation(sp, bp)
counts[i] = obs.countrate()
return counts / norm_counts
[docs]def photfnu_from_photflam(photflam, photplam):
"""
Compute PHOTFNU from PHOTFLAM and PHOTPLAM, e.g., for ACS/WFC
Parameters
----------
photflam : float
The PHOTFLAM value from the FITS header.
photplam : float
The PHOTPLAM value from the FITS header.
Returns
-------
photfnu : float
The computed PHOTFNU value.
Examples
--------
>>> ZP = -2.5 * np.log10(photflam) - 21.10 - 5 * np.log10(photplam) + 18.6921
>>> photfnu = 10 ** (-0.4 * (ZP - 23.9)) * 1.0e-6
"""
ZP = -2.5 * np.log10(photflam) - 21.10 - 5 * np.log10(photplam) + 18.6921
photfnu = 10 ** (-0.4 * (ZP - 23.9)) * 1.0e-6
return photfnu
DEFAULT_PRIMARY_KEYS = [
"FILENAME",
"INSTRUME",
"INSTRUME",
"DETECTOR",
"FILTER",
"FILTER1",
"FILTER2",
"EXPSTART",
"DATE-OBS",
"EXPTIME",
"IDCTAB",
"NPOLFILE",
"D2IMFILE",
"PA_V3",
"FGSLOCK",
"GYROMODE",
"PROPOSID",
]
# For grism
DEFAULT_EXT_KEYS = [
"EXTNAME",
"EXTVER",
"MDRIZSKY",
"CRPIX1",
"CRPIX2",
"CRVAL1",
"CRVAL2",
"CD1_1",
"CD1_2",
"CD2_1",
"CD2_2",
"PC1_1",
"PC1_2",
"PC2_1",
"PC2_2",
"CDELT1",
"CDELT2",
"CUNIT1",
"CUNIT2",
"CTYPE1",
"CTYPE2",
"RADESYS",
"LONPOLE",
"LATPOLE",
"IDCTAB",
"D2IMEXT",
"WCSNAME",
"PHOTMODE",
"ORIENTAT",
"CCDCHIP",
]
[docs]def flt_to_dict(
fobj,
primary_keys=DEFAULT_PRIMARY_KEYS,
extensions=[("SCI", i + 1) for i in range(2)],
ext_keys=DEFAULT_EXT_KEYS,
):
"""
Parse basic elements from a FLT/FLC header to a dictionary
Parameters
----------
fobj : `~astropy.io.fits.HDUList`
FITS object
primary_keys : list
Keywords to extract from the primary extension (0).
extensions : list
List of additional extension names / indices.
ext_keys : list
Keywords to extract from the extension headers.
Returns
-------
flt_dict : dict
"""
import astropy.time
flt_dict = OrderedDict()
flt_dict["timestamp"] = astropy.time.Time.now().iso
h0 = fobj[0].header
# Primary keywords
for k in primary_keys:
if k in h0:
flt_dict[k] = h0[k]
# Grism keys
for k in h0:
if k.startswith("GSKY"):
flt_dict[k] = h0[k]
# WCS, etc. keywords from SCI extensions
flt_dict["extensions"] = OrderedDict()
count = 0
for ext in extensions:
if ext in fobj:
d_i = OrderedDict()
h_i = fobj[ext].header
for k in ext_keys:
if k in h_i:
d_i[k] = h_i[k]
# Grism keys
for k in h_i:
if k.startswith("GSKY"):
d_i[k] = h_i[k]
count += 1
flt_dict["extensions"][count] = d_i
return flt_dict
[docs]def mod_dq_bits(value, okbits=32 + 64 + 512, badbits=0, verbose=False):
"""
Modify bit flags from a DQ array
For WFC3/IR, the following DQ bits can usually be unset:
32, 64: these pixels usually seem OK
512: blobs not relevant for grism exposures
Parameters
----------
value : int, `~numpy.ndarray`
Input DQ value
okbits : int
Bits to unset
badbits : int
Bits to set
verbose : bool
Print some information
Returns
-------
new_value : int, `~numpy.ndarray`
"""
if verbose:
print(f"Unset bits: {np.binary_repr(okbits)}")
print(f"Set bits: {np.binary_repr(badbits)}")
return (value & ~okbits) | badbits
[docs]def detect_with_photutils(sci, err=None, dq=None, seg=None, detect_thresh=2.,
npixels=8, grow_seg=5, gauss_fwhm=2., gsize=3,
wcs=None, save_detection=False, root='mycat',
background=None, gain=None, AB_zeropoint=0.,
rename_columns={'xcentroid': 'x_flt',
'ycentroid': 'y_flt',
'ra_icrs_centroid': 'ra',
'dec_icrs_centroid': 'dec'},
overwrite=True, verbose=True):
r"""
Use `~photutils` to detect objects and make segmentation map
.. note::
Deprecated in favor of sep catalogs in `~grizli.prep`.
Parameters
----------
sci : `~numpy.ndarray`
err, dq, seg : TBD
detect_thresh : float
Detection threshold, in :math:`\sigma`
grow_seg : int
Number of pixels to grow around the perimeter of detected objects
witha maximum filter
gauss_fwhm : float
FWHM of Gaussian convolution kernel that smoothes the detection
image.
verbose : bool
Print logging information to the terminal
save_detection : bool
Save the detection images and catalogs
wcs : `~astropy.wcs.WCS`
WCS object passed to `photutils.source_properties` used to compute
sky coordinates of detected objects.
Returns
-------
catalog : `~astropy.table.Table`
Object catalog with the default parameters.
"""
import scipy.ndimage as nd
from photutils import detect_threshold, detect_sources, SegmentationImage
from photutils import source_properties
import astropy.io.fits as pyfits
from astropy.table import Column
from astropy.stats import sigma_clipped_stats, gaussian_fwhm_to_sigma
from astropy.convolution import Gaussian2DKernel
# DQ masks
mask = sci == 0
if dq is not None:
mask |= dq > 0
# Detection threshold
if err is None:
threshold = detect_threshold(sci, snr=detect_thresh, mask=mask)
else:
threshold = (detect_thresh * err) * (~mask)
threshold[mask] = np.median(threshold[~mask])
if seg is None:
# Run the source detection and create the segmentation image
# Gaussian kernel
sigma = gauss_fwhm * gaussian_fwhm_to_sigma # FWHM = 2.
kernel = Gaussian2DKernel(sigma, x_size=gsize, y_size=gsize)
kernel.normalize()
if verbose:
print(
"{0}: photutils.detect_sources (detect_thresh={1:.1f}, grow_seg={2:d}, gauss_fwhm={3:.1f}, ZP={4:.1f})".format(
root, detect_thresh, grow_seg, gauss_fwhm, AB_zeropoint
)
)
# Detect sources
segm = detect_sources(
sci * (~mask), threshold, npixels=npixels, filter_kernel=kernel
)
grow = nd.maximum_filter(segm.data, grow_seg)
seg = np.asarray(grow,dtype=np.float32)
else:
# Use the supplied segmentation image
segm = SegmentationImage(seg)
# Source properties catalog
if verbose:
print("{0}: photutils.source_properties".format(root))
props = source_properties(
sci,
segm,
error=threshold / detect_thresh,
mask=mask,
background=background,
wcs=wcs,
)
catalog = props.to_table()
# Mag columns
mag = AB_zeropoint - 2.5 * np.log10(catalog["source_sum"])
mag._name = "mag"
catalog.add_column(mag)
try:
logscale = 2.5 / np.log(10)
mag_err = logscale * catalog["source_sum_err"] / catalog["source_sum"]
except:
mag_err = np.zeros_like(mag) - 99
mag_err._name = "mag_err"
catalog.add_column(mag_err)
# Rename some catalog columns
for key in rename_columns.keys():
if key not in catalog.colnames:
continue
catalog.rename_column(key, rename_columns[key])
if verbose:
print("Rename column: {0} -> {1}".format(key, rename_columns[key]))
# Done!
if verbose:
print(
NO_NEWLINE
+ (
"{0}: photutils.source_properties - {1:d} objects".format(
root, len(catalog)
)
)
)
# Save outputs?
if save_detection:
seg_file = root + ".detect_seg.fits"
seg_cat = root + ".detect.cat"
if verbose:
print("{0}: save {1}, {2}".format(root, seg_file, seg_cat))
if wcs is not None:
header = wcs.to_header(relax=True)
else:
header = None
pyfits.writeto(seg_file, data=seg, header=header, overwrite=overwrite)
if os.path.exists(seg_cat) & overwrite:
os.remove(seg_cat)
catalog.write(seg_cat, format="ascii.commented_header")
return catalog, seg
[docs]def safe_invert(arr):
"""
Version-safe matrix inversion using `numpy.linalg` or `numpy.matrix.I`
Parameters
----------
arr : array_like
The input array to be inverted.
Returns
-------
_inv : ndarray
The inverted array.
"""
try:
from numpy.linalg import inv
_inv = inv(arr)
except:
_inv = np.matrix(arr).I.A
return _inv
[docs]def nmad(data):
"""
Normalized NMAD = 1.4826022 * `~.astropy.stats.median_absolute_deviation`
Parameters
----------
data: array-like
The input data array.
Returns
-------
nmad: float
The normalized median absolute deviation of the input data.
"""
import astropy.stats
return 1.4826022 * astropy.stats.median_absolute_deviation(data)
[docs]def get_line_wavelengths():
"""
Get a dictionary of common emission line wavelengths and line ratios
Returns
-------
line_wavelengths, line_ratios : dict
Keys are common to both dictionaries and are simple names for lines
and line complexes. Values are lists of line wavelengths and line
ratios.
>>> from grizli.utils import get_line_wavelengths
>>> line_wavelengths, line_ratios = get_line_wavelengths()
>>> print(line_wavelengths['Ha'], line_ratios['Ha'])
[6564.61] [1.0]
>>> print(line_wavelengths['OIII'], line_ratios['OIII'])
[5008.24, 4960.295] [2.98, 1]
Includes some additional combined line complexes useful for redshift
fits:
>>> from grizli.utils import get_line_wavelengths
>>> line_wavelengths, line_ratios = get_line_wavelengths()
>>> key = 'Ha+SII+SIII+He'
>>> print(line_wavelengths[key], '\\n', line_ratios[key])
[6564.61, 6718.29, 6732.67, 9068.6, 9530.6, 10830.0]
[1.0, 0.1, 0.1, 0.05, 0.122, 0.04]
"""
line_wavelengths = OrderedDict()
line_ratios = OrderedDict()
# Paschen: https://www.gemini.edu/sciops/instruments/nearir-resources/astronomical-lines/h-lines
# Rh = 0.0010967757
# k = Rh * (1/n0**2 - 1/n**2)
# wave = 1./k # Angstroms
# Pfund n0=5
line_wavelengths["PfA"] = [74598.8]
line_ratios["PfA"] = [1.0]
line_wavelengths["PfB"] = [46537.9]
line_ratios["PfB"] = [1.0]
line_wavelengths["PfG"] = [37405.7]
line_ratios["PfG"] = [1.0]
line_wavelengths["PfD"] = [32970.0]
line_ratios["PfD"] = [1.0]
line_wavelengths["PfE"] = [30392.1]
line_ratios["PfE"] = [1.0]
# Brackett n0=4
line_wavelengths["BrA"] = [40522.8]
line_ratios["BrA"] = [1.0]
line_wavelengths["BrB"] = [26258.8]
line_ratios["BrB"] = [1.0]
line_wavelengths["BrG"] = [21661.3]
line_ratios["BrG"] = [1.0]
line_wavelengths["BrD"] = [19451.0]
line_ratios["BrD"] = [1.0]
line_wavelengths["BrE"] = [18179.2]
line_ratios["BrE"] = [1.0]
line_wavelengths["BrF"] = [17366.9]
line_ratios["BrF"] = [1.0]
# Paschen n0=3
line_wavelengths["PaA"] = [18756.3]
line_ratios["PaA"] = [1.0]
line_wavelengths["PaB"] = [12821.7]
line_ratios["PaB"] = [1.0]
line_wavelengths["PaG"] = [10941.2]
line_ratios["PaG"] = [1.0]
line_wavelengths["PaD"] = [10052.2]
line_ratios["PaD"] = [1.0]
line_wavelengths["Pa8"] = [9548.65]
line_ratios["Pa8"] = [1.0]
line_wavelengths["Pa9"] = [9231.60]
line_ratios["Pa9"] = [1.0]
line_wavelengths["Pa10"] = [9017.44]
line_ratios["Pa10"] = [1.0]
# Balmer n0=2
line_wavelengths["Ha"] = [6564.697]
line_ratios["Ha"] = [1.0]
line_wavelengths["Hb"] = [4862.738]
line_ratios["Hb"] = [1.0]
line_wavelengths["Hg"] = [4341.731]
line_ratios["Hg"] = [1.0]
line_wavelengths["Hd"] = [4102.936]
line_ratios["Hd"] = [1.0]
line_wavelengths["H7"] = [3971.236]
line_ratios["H7"] = [1.0]
line_wavelengths["H8"] = [3890.191]
line_ratios["H8"] = [1.0]
line_wavelengths["H9"] = [3836.511]
line_ratios["H9"] = [1.0]
line_wavelengths["H10"] = [3799.014]
line_ratios["H10"] = [1.0]
line_wavelengths["H11"] = [3771.739]
line_ratios["H11"] = [1.0]
line_wavelengths["H12"] = [3751.255]
line_ratios["H12"] = [1.0]
# Groves et al. 2011, Table 1
# Osterbrock table 4.4 for H7 to H10
# line_wavelengths['Balmer 10kK'] = [6564.61, 4862.68, 4341.68, 4101.73]
# line_ratios['Balmer 10kK'] = [2.86, 1.0, 0.468, 0.259]
line_wavelengths["Balmer 10kK"] = [
6564.61,
4862.68,
4341.68,
4101.73,
3971.198,
3890.166,
3836.485,
3798.987,
]
line_ratios["Balmer 10kK"] = [2.86, 1.0, 0.468, 0.259, 0.159, 0.105, 0.0731, 0.0530]
# Paschen from Osterbrock, e.g., Pa-beta relative to H-gamma
line_wavelengths["Balmer 10kK"] += (
line_wavelengths["PaA"]
+ line_wavelengths["PaB"]
+ line_wavelengths["PaG"]
+ line_wavelengths["PaD"]
)
line_ratios["Balmer 10kK"] += [
0.348 * line_ratios["Balmer 10kK"][i] for i in [1, 2, 3, 4]
]
# Osterbrock table 4.4 for H7 to H10
line_wavelengths["Balmer 10kK + MgII"] = line_wavelengths["Balmer 10kK"] + [
2799.117
]
line_ratios["Balmer 10kK + MgII"] = line_ratios["Balmer 10kK"] + [3.0]
# # Paschen from Osterbrock, e.g., Pa-beta relative to H-gamma
# line_wavelengths['Balmer 10kK + MgII'] += line_wavelengths['PaA'] + line_wavelengths['PaB'] + line_wavelengths['PaG']
# line_ratios['Balmer 10kK + MgII'] += [0.348 * line_ratios['Balmer 10kK + MgII'][i] for i in [1,2,3]]
# With Paschen lines & He 10830 from Glikman 2006
# https://iopscience.iop.org/article/10.1086/500098/pdf
# line_wavelengths['Balmer 10kK + MgII'] = [6564.61, 4862.68, 4341.68, 4101.73, 3971.198, 2799.117, 12821.6, 10941.1]
# line_ratios['Balmer 10kK + MgII'] = [2.86, 1.0, 0.468, 0.259, 0.16, 3., 2.86*4.8/100, 2.86*1.95/100]
# Redden with Calzetti00
if False:
from extinction import calzetti00
Av = 1.0
Rv = 3.1
waves = line_wavelengths["Balmer 10kK + MgII"]
ratios = line_ratios["Balmer 10kK + MgII"]
for Av in [0.5, 1.0, 2.0]:
mred = calzetti00(np.array(waves), Av, Rv)
fred = 10 ** (-0.4 * mred)
key = "Balmer 10kK + MgII Av={0:.1f}".format(Av)
line_wavelengths[key] = [w for w in waves]
line_ratios[key] = [ratios[i] * fred[i] for i in range(len(waves))]
line_wavelengths["Balmer 10kK + MgII Av=0.5"] = [
6564.61,
4862.68,
4341.68,
4101.73,
3971.198,
2799.117,
12821.6,
10941.1,
]
line_ratios["Balmer 10kK + MgII Av=0.5"] = [
2.009811938798515,
0.5817566641521459,
0.25176970824566913,
0.1338409369665902,
0.08079209880749984,
1.1739297839690317,
0.13092553990513178,
0.05033866127477651,
]
line_wavelengths["Balmer 10kK + MgII Av=1.0"] = [
6564.61,
4862.68,
4341.68,
4101.73,
3971.198,
2799.117,
12821.6,
10941.1,
]
line_ratios["Balmer 10kK + MgII Av=1.0"] = [
1.4123580522157504,
0.33844081628543266,
0.13544441450878067,
0.0691636926953466,
0.04079602018575511,
0.4593703792298591,
0.12486521707058751,
0.045436270735820045,
]
line_wavelengths["Balmer 10kK + MgII Av=2.0"] = [
6564.61,
4862.68,
4341.68,
4101.73,
3971.198,
2799.117,
12821.6,
10941.1,
]
line_ratios["Balmer 10kK + MgII Av=2.0"] = [
0.6974668768037302,
0.11454218612794999,
0.03919912269578289,
0.018469561340758073,
0.010401970393728362,
0.0703403817712615,
0.11357315292894044,
0.03701729780130422,
]
###########
# Reddened with Kriek & Conroy dust, tau_V=0.5
line_wavelengths["Balmer 10kK t0.5"] = [6564.61, 4862.68, 4341.68, 4101.73]
line_ratios["Balmer 10kK t0.5"] = [
2.86 * 0.68,
1.0 * 0.55,
0.468 * 0.51,
0.259 * 0.48,
]
# Reddened with Kriek & Conroy dust, tau_V=1
line_wavelengths["Balmer 10kK t1"] = [6564.61, 4862.68, 4341.68, 4101.73]
line_ratios["Balmer 10kK t1"] = [
2.86 * 0.46,
1.0 * 0.31,
0.468 * 0.256,
0.259 * 0.232,
]
line_wavelengths["OIII-4363"] = [4364.436]
line_ratios["OIII-4363"] = [1.0]
line_wavelengths["OIII"] = [5008.240, 4960.295]
line_ratios["OIII"] = [2.98, 1]
# Split doublet, if needed
line_wavelengths["OIII-4959"] = [4960.295]
line_ratios["OIII-4959"] = [1]
line_wavelengths["OIII-5007"] = [5008.240]
line_ratios["OIII-5007"] = [1]
line_wavelengths["OII"] = [3727.092, 3729.875]
line_ratios["OII"] = [1, 1.0]
line_wavelengths["OI-5578"] = [5578.89]
line_ratios["OI-5578"] = [1]
line_wavelengths["OI-6302"] = [6302.046, 6365.535]
line_ratios["OI-6302"] = [1, 0.33]
line_wavelengths["OI-7776"] = [7776.3]
line_ratios["OI-7776"] = [1]
line_wavelengths["OI-8448"] = [8448.7]
line_ratios["OI-8448"] = [1]
line_wavelengths["OI-11290"] = [11290.0]
line_ratios["OI-11290"] = [1]
# Auroral OII
# lines roughly taken from https://arxiv.org/pdf/1610.06939.pdf
line_wavelengths["OII-7325"] = [7321.9, 7332.21]
line_ratios["OII-7325"] = [1.2, 1.0]
line_wavelengths["OII-7323"] = [7321.9]
line_ratios["OII-7323"] = [1.0]
line_wavelengths["OII-7332"] = [7332.21]
line_ratios["OII-7332"] = [1.0]
# Weak Ar III in SF galaxies
line_wavelengths["ArIII-7138"] = [7137.77]
line_ratios["ArIII-7138"] = [1.0]
line_wavelengths["ArIII-7753"] = [7753.19]
line_ratios["ArIII-7753"] = [1.0]
line_wavelengths["NeIII-3867"] = [3869.87]
line_ratios["NeIII-3867"] = [1.0]
line_wavelengths["NeIII-3968"] = [3968.59]
line_ratios["NeIII-3968"] = [1.0]
line_wavelengths["NeV-3346"] = [3343.5]
line_ratios["NeV-3346"] = [1.0]
line_wavelengths["NeVI-3426"] = [3426.85]
line_ratios["NeVI-3426"] = [1.0]
line_wavelengths["SIII"] = [9071.1, 9533.2][::-1]
line_ratios["SIII"] = [1, 2.44][::-1]
# Split doublet, if needed
line_wavelengths["SIII-9068"] = [9071.1]
line_ratios["SIII-9068"] = [1]
line_wavelengths["SIII-9531"] = [9533.2]
line_ratios["SIII-9531"] = [1]
line_wavelengths["SIII-6314"] = [6313.81]
line_ratios["SIII-6314"] = [1.0]
line_wavelengths["SII"] = [6718.29, 6732.67]
line_ratios["SII"] = [1.0, 1.0]
line_wavelengths["SII-6717"] = [6718.29]
line_ratios["SII-6717"] = [1.0]
line_wavelengths["SII-6731"] = [6732.67]
line_ratios["SII-6731"] = [1.0]
line_wavelengths["SII-4075"] = [4069.75, 4077.5]
line_ratios["SII-4075"] = [1.0, 1.0]
line_wavelengths["SII-4070"] = [4069.75]
line_ratios["SII-4075"] = [1.0]
line_wavelengths["SII-4078"] = [4077.5]
line_ratios["SII-4078"] = [1.0]
line_wavelengths["HeII-4687"] = [4687.5]
line_ratios["HeII-4687"] = [1.0]
line_wavelengths["HeII-5412"] = [5412.5]
line_ratios["HeII-5412"] = [1.0]
line_wavelengths["HeII-16923"] = [1.69230e4]
line_ratios["HeII-16923"] = [1.0]
line_wavelengths["HeI-5877"] = [5877.249]
line_ratios["HeI-5877"] = [1.0]
line_wavelengths["HeI-3889"] = [3889.75]
line_ratios["HeI-3889"] = [1.0]
line_wavelengths["HeI-1083"] = [10832.057, 10833.306]
line_ratios["HeI-1083"] = [1.0, 1.0]
line_wavelengths["HeI-3820"] = [3820.7]
line_ratios["HeI-3820"] = [1.0]
line_wavelengths["HeI-4027"] = [4027.3]
line_ratios["HeI-4027"] = [1.0]
line_wavelengths["HeI-4472"] = [4472.7]
line_ratios["HeI-4472"] = [1.0]
line_wavelengths["HeI-6680"] = [6679.995]
line_ratios["HeI-6680"] = [1.0]
line_wavelengths["HeI-7065"] = [7067.1]
line_ratios["HeI-7065"] = [1.0]
line_wavelengths["HeI-8446"] = [8446.7]
line_ratios["HeI-8446"] = [1.0]
# From CAFE
# https://github.com/GOALS-survey/CAFE/blob/master/CAFE/tables/
line_wavelengths["FeII-11128"] = [1.11286e4]
line_ratios["FeII-11128"] = [1.0]
line_wavelengths["FeII-12570"] = [1.25702e4]
line_ratios["FeII-12570"] = [1.0]
line_wavelengths["FeII-16440"] = [1.64400e4]
line_ratios["FeII-16440"] = [1.0]
line_wavelengths["FeII-16877"] = [1.68778e4]
line_ratios["FeII-16877"] = [1.0]
line_wavelengths["FeII-17418"] = [1.74188e4]
line_ratios["FeII-17418"] = [1.0]
line_wavelengths["FeII-17418"] = [1.74188e4]
line_ratios["FeII-17418"] = [1.0]
line_wavelengths["FeII-18362"] = [1.83624e4]
line_ratios["FeII-18362"] = [1.0]
line_wavelengths["SiVI-19634"] = [1.9634e4]
line_ratios["SiVI-19634"] = [1.0]
# AGN line?
# https://academic.oup.com/pasj/article/63/1/L7/1460068#431992120
line_wavelengths["PII-11886"] = [1.188610e4]
line_ratios["PII-11886"] = [1.0]
# Osterbrock Table 4.5
# -> N=4
line_wavelengths["HeI-series"] = [
4472.7,
5877.2,
4027.3,
3820.7,
7067.1,
10833.2,
3889.7,
3188.7,
]
line_ratios["HeI-series"] = [1.0, 2.75, 0.474, 0.264, 0.330, 4.42, 2.26, 0.916]
line_wavelengths["MgII"] = [2799.117]
line_ratios["MgII"] = [1.0]
line_wavelengths["CIV-1549"] = [1549.480]
line_ratios["CIV-1549"] = [1.0]
line_wavelengths["CIII-1906"] = [1906.683]
line_ratios["CIII-1906"] = [1.0]
line_wavelengths["CIII-1908"] = [1908.734]
line_ratios["CIII-1908"] = [1.0]
line_wavelengths["CI-9580"] = [9850.26] # leave typo for back compatibility
line_ratios["CI-9580"] = [1.0]
line_wavelengths["CI-9850"] = [9850.26]
line_ratios["CI-9850"] = [1.0]
# Sodium D I lines from Davies 2023
# https://arxiv.org/abs/2310.17939v2
line_wavelengths["NaDI"] = [5891.0, 5897.0]
line_ratios["NaDI"] = [1.0, 1.0]
# Hutchinson
# https://iopscience.iop.org/article/10.3847/1538-4357/ab22a2
line_wavelengths["CIII-1906x"] = [1906.683, 1908.734]
line_ratios["CIII-1906x"] = [1.5, 1.0]
line_wavelengths["OIII-1663"] = [1665.85]
line_ratios["OIII-1663"] = [1.0]
line_wavelengths["HeII-1640"] = [1640.4]
line_ratios["HeII-1640"] = [1.0]
line_wavelengths["SiIV+OIV-1398"] = [1398.0]
line_ratios["SiIV+OIV-1398"] = [1.0]
# Weak line in LEGA-C spectra
line_wavelengths["NI-5199"] = [5199.4, 5201.76]
line_ratios["NI-5199"] = [1.0, 1.0]
line_wavelengths["NII"] = [6549.86, 6585.27][::-1]
line_ratios["NII"] = [1.0, 3.0][::-1]
line_wavelengths["NII-6549"] = [6549.86]
line_ratios["NII-6549"] = [1.0]
line_wavelengths["NII-6584"] = [6585.27]
line_ratios["NII-6584"] = [1.0]
line_wavelengths["NIII-1750"] = [1750.0]
line_ratios["NIII-1750"] = [1.0]
line_wavelengths["NIV-1487"] = [1487.0]
line_ratios["NIV-1487"] = [1.0]
line_wavelengths["NV-1240"] = [1240.81]
line_ratios["NV-1240"] = [1.0]
line_wavelengths["Lya"] = [1215.4]
line_ratios["Lya"] = [1.0]
line_wavelengths["QSO-UV-lines"] = [
line_wavelengths[k][0]
for k in [
"Lya",
"CIV-1549",
"CIII-1906",
"CIII-1908",
"OIII-1663",
"HeII-1640",
"SiIV+OIV-1398",
"NV-1240",
"NIII-1750",
]
]
line_ratios["QSO-UV-lines"] = [1.0, 0.5, 0.1, 0.1, 0.008, 0.09, 0.1, 0.3, 0.05]
line_wavelengths["QSO-Narrow-lines"] = [
line_wavelengths[k][0]
for k in [
"OII",
"OIII-5007",
"OIII-4959",
"SII-6717",
"SII-6731",
"OI-6302",
"NeIII-3867",
"NeVI-3426",
"NeV-3346",
]
]
line_ratios["QSO-Narrow-lines"] = [
0.2,
1.6,
1.6 / 2.98,
0.1,
0.1,
0.01,
0.5,
0.2,
0.02,
]
# redder lines
line_wavelengths["QSO-Narrow-lines"] += line_wavelengths["SIII"]
line_ratios["QSO-Narrow-lines"] += [lr * 0.05 for lr in line_ratios["SIII"]]
line_wavelengths["QSO-Narrow-lines"] += line_wavelengths["HeI-1083"]
line_ratios["QSO-Narrow-lines"] += [0.2]
line_wavelengths["Lya+CIV"] = [1215.4, 1549.49]
line_ratios["Lya+CIV"] = [1.0, 0.1]
line_wavelengths["Gal-UV-lines"] = [
line_wavelengths[k][0]
for k in [
"Lya",
"CIV-1549",
"CIII-1906",
"CIII-1908",
"OIII-1663",
"HeII-1640",
"SiIV+OIV-1398",
"NV-1240",
"NIII-1750",
"MgII",
]
]
line_ratios["Gal-UV-lines"] = [
1.0,
0.15,
0.1,
0.1,
0.008,
0.09,
0.1,
0.05,
0.05,
0.1,
]
line_wavelengths["Ha+SII"] = [6564.61, 6718.29, 6732.67]
line_ratios["Ha+SII"] = [1.0, 1.0 / 10, 1.0 / 10]
line_wavelengths["Ha+SII+SIII+He"] = [
6564.61,
6718.29,
6732.67,
9068.6,
9530.6,
10830.0,
]
line_ratios["Ha+SII+SIII+He"] = [
1.0,
1.0 / 10,
1.0 / 10,
1.0 / 20,
2.44 / 20,
1.0 / 25.0,
]
line_wavelengths["Ha+NII+SII+SIII+He"] = [
6564.61,
6549.86,
6585.27,
6718.29,
6732.67,
9068.6,
9530.6,
10830.0,
]
line_ratios["Ha+NII+SII+SIII+He"] = [
1.0,
1.0 / (4.0 * 4),
3.0 / (4 * 4),
1.0 / 10,
1.0 / 10,
1.0 / 20,
2.44 / 20,
1.0 / 25.0,
]
line_wavelengths["Ha+NII+SII+SIII+He+PaB"] = [
6564.61,
6549.86,
6585.27,
6718.29,
6732.67,
9068.6,
9530.6,
10830.0,
12821,
]
line_ratios["Ha+NII+SII+SIII+He+PaB"] = [
1.0,
1.0 / (4.0 * 4),
3.0 / (4 * 4),
1.0 / 10,
1.0 / 10,
1.0 / 20,
2.44 / 20,
1.0 / 25.0,
1.0 / 10,
]
line_wavelengths["Ha+NII+SII+SIII+He+PaB+PaG"] = [
6564.61,
6549.86,
6585.27,
6718.29,
6732.67,
9068.6,
9530.6,
10830.0,
12821,
10941.1,
]
line_ratios["Ha+NII+SII+SIII+He+PaB+PaG"] = [
1.0,
1.0 / (4.0 * 4),
3.0 / (4 * 4),
1.0 / 10,
1.0 / 10,
1.0 / 20,
2.44 / 20,
1.0 / 25.0,
1.0 / 10,
1.0 / 10 / 2.86,
]
line_wavelengths["Ha+NII"] = [6564.61, 6549.86, 6585.27]
n2ha = 1.0 / 3 # log NII/Ha ~ -0.6, Kewley 2013
line_ratios["Ha+NII"] = [1.0, 1.0 / 4.0 * n2ha, 3.0 / 4.0 * n2ha]
line_wavelengths["OIII+Hb"] = [5008.240, 4960.295, 4862.68]
line_ratios["OIII+Hb"] = [2.98, 1, 3.98 / 6.0]
# Include more balmer lines
line_wavelengths["OIII+Hb+Hg+Hd"] = (
line_wavelengths["OIII"] + line_wavelengths["Balmer 10kK"][1:]
)
line_ratios["OIII+Hb+Hg+Hd"] = line_ratios["OIII"] + line_ratios["Balmer 10kK"][1:]
# o3hb = 1./6
# for i in range(2, len(line_ratios['Balmer 10kK'])-1):
# line_ratios['OIII+Hb+Hg+Hd'][i] *= 3.98*o3hb
# Compute as O3/Hb
o3hb = 6
for i in range(2):
line_ratios["OIII+Hb+Hg+Hd"][i] *= 1.0 / 3.98 * o3hb
line_wavelengths["OIII+Hb+Ha"] = [5008.240, 4960.295, 4862.68, 6564.61]
line_ratios["OIII+Hb+Ha"] = [2.98, 1, 3.98 / 10.0, 3.98 / 10.0 * 2.86]
line_wavelengths["OIII+Hb+Ha+SII"] = [
5008.240,
4960.295,
4862.68,
6564.61,
6718.29,
6732.67,
]
line_ratios["OIII+Hb+Ha+SII"] = [
2.98,
1,
3.98 / 10.0,
3.98 / 10.0 * 2.86 * 4,
3.98 / 10.0 * 2.86 / 10.0 * 4,
3.98 / 10.0 * 2.86 / 10.0 * 4,
]
line_wavelengths["OIII+OII"] = [5008.240, 4960.295, 3729.875]
line_ratios["OIII+OII"] = [2.98, 1, 3.98 / 4.0]
line_wavelengths["OII+Ne"] = [3729.875, 3869]
line_ratios["OII+Ne"] = [1, 1.0 / 5]
# Groups of all lines
line_wavelengths["full"] = [w for w in line_wavelengths["Balmer 10kK"]]
line_ratios["full"] = [w for w in line_ratios["Balmer 10kK"]]
line_wavelengths["full"] += line_wavelengths["NII"]
line_ratios["full"] += [
1.0 / 5 / 3.0 * line_ratios["Balmer 10kK"][1] * r for r in line_ratios["NII"]
]
line_wavelengths["full"] += line_wavelengths["SII"]
line_ratios["full"] += [
1.0 / 3.8 / 2 * line_ratios["Balmer 10kK"][1] * r for r in line_ratios["SII"]
]
# Lines from Hagele 2006, low-Z HII galaxies
# SDSS J002101.03+005248.1
line_wavelengths["full"] += line_wavelengths["SIII"]
line_ratios["full"] += [
401.0 / 1000 / 2.44 * line_ratios["Balmer 10kK"][1] * r
for r in line_ratios["SIII"]
]
# HeI
line_wavelengths["full"] += line_wavelengths["HeI-series"]
he5877_hb = 127.0 / 1000 / line_ratios["HeI-series"][1]
line_ratios["full"] += [he5877_hb * r for r in line_ratios["HeI-series"]]
# NeIII
line_wavelengths["full"] += line_wavelengths["NeIII-3867"]
line_ratios["full"] += [388.0 / 1000 for r in line_ratios["NeIII-3867"]]
line_wavelengths["full"] += line_wavelengths["NeIII-3968"]
line_ratios["full"] += [290.0 / 1000 for r in line_ratios["NeIII-3968"]]
# Add UV lines: MgII/Hb = 3
line_wavelengths["full"] += line_wavelengths["Gal-UV-lines"]
line_ratios["full"] += [
r * 3 / line_ratios["Gal-UV-lines"][-1] for r in line_ratios["Gal-UV-lines"]
]
# High O32 - low metallicity
o32, r23 = 4, 8
o3_hb = r23 / (1 + 1 / o32)
line_wavelengths["highO32"] = [w for w in line_wavelengths["full"]]
line_ratios["highO32"] = [r for r in line_ratios["full"]]
line_wavelengths["highO32"] += line_wavelengths["OIII"]
line_ratios["highO32"] += [r * o3_hb / 3.98 for r in line_ratios["OIII"]]
line_wavelengths["highO32"] += line_wavelengths["OII"]
line_ratios["highO32"] += [r * o3_hb / 2 / o32 for r in line_ratios["OII"]]
# Low O32 - low metallicity
o32, r23 = 0.3, 4
o3_hb = r23 / (1 + 1 / o32)
line_wavelengths["lowO32"] = [w for w in line_wavelengths["full"]]
line_ratios["lowO32"] = [r for r in line_ratios["full"]]
line_wavelengths["lowO32"] += line_wavelengths["OIII"]
line_ratios["lowO32"] += [r * o3_hb / 3.98 for r in line_ratios["OIII"]]
line_wavelengths["lowO32"] += line_wavelengths["OII"]
line_ratios["lowO32"] += [r * o3_hb / 2 / o32 for r in line_ratios["OII"]]
return line_wavelengths, line_ratios
[docs]def emission_line_templates():
"""
Testing FSPS line templates
"""
import numpy as np
import matplotlib.pyplot as plt
from grizli import utils
import fsps
sp = fsps.StellarPopulation(imf_type=1, zcontinuous=1)
sp_params = {}
sp_params["starburst"] = {
"sfh": 4,
"tau": 0.3,
"tage": 0.1,
"logzsol": -1,
"gas_logz": -1,
"gas_logu": -2.5,
}
sp_params["mature"] = {
"sfh": 4,
"tau": 0.2,
"tage": 0.9,
"logzsol": -0.2,
"gas_logz": -0.2,
"gas_logu": -2.5,
}
line_templates = {}
for t in sp_params:
pset = sp_params[t]
header = "wave flux\n\n"
for p in pset:
header += "{0} = {1}\n".format(p, pset[p])
if p == "tage":
continue
print(p, pset[p])
sp.params[p] = pset[p]
spec = {}
for neb in [True, False]:
sp.params["add_neb_emission"] = neb
sp.params["add_neb_continuum"] = neb
wave, spec[neb] = sp.get_spectrum(tage=pset["tage"], peraa=True)
# plt.plot(wave, spec[neb], alpha=0.5)
neb_only = spec[True] - spec[False]
neb_only = neb_only / neb_only.max()
neb_only = spec[True] / spec[True].max()
plt.plot(wave, neb_only, label=t, alpha=0.5)
neb_only[neb_only < 1.0e-4] = 0
np.savetxt(
"fsps_{0}_lines.txt".format(t),
np.array([wave, neb_only]).T,
fmt="%.5e",
header=header,
)
line_templates[t] = utils.SpectrumTemplate(
wave=wave, flux=neb_only, name="fsps_{0}_lines".format(t)
)
[docs]def pah33(wave_grid):
"""
Set of 3.3 micron PAH lines from Li et al. 2020
Parameters
----------
wave_grid : array-like
Wavelength grid for the templates.
Returns
-------
pah_templates : list
List of `~grizli.utils.SpectrumTemplate` templates for three components
around 3.3 microns
"""
pah_templates = {}
for lc, lw in zip([3.29, 3.40, 3.47], [0.043, 0.031, 0.100]):
ti = pah_line_template(wave_grid, center_um=lc, fwhm=lw)
pah_templates[ti.name] = ti
return pah_templates
[docs]def pah_line_template(wave_grid, center_um=3.29, fwhm=0.043):
"""
Make a template for a broad PAH line with a Drude profile
Default parameters in Lai et al. 2020
https://iopscience.iop.org/article/10.3847/1538-4357/abc002/pdf
from Tokunaga et al. 1991
Drude equation and normalization from Yamada et al. 2013
Parameters
----------
wave_grid : array-like
Wavelength grid in angstroms
center_um : float
Central wavelength in microns
fwhm : float
Drude profile FWHM in microns
Returns
-------
pah_templ : `~grizli.utils.SpectrumTemplate`
Template with the PAH feature
"""
br = 1.0
gamma_width = fwhm / center_um
Iv = br * gamma_width ** 2
Iv /= (
wave_grid / 1.0e4 / center_um - center_um * 1.0e4 / wave_grid
) ** 2 + gamma_width ** 2
Inorm = np.pi * 2.99e14 / 2.0 * br * gamma_width / center_um
Iv *= 1 / Inorm
# Flambda
Ilam = Iv * 2.99e18 / (wave_grid) ** 2
pah_templ = SpectrumTemplate(
wave=wave_grid, flux=Ilam, name=f"line PAH-{center_um:.2f}"
)
return pah_templ
[docs]class SpectrumTemplate(object):
def __init__(
self,
wave=None,
flux=None,
central_wave=None,
fwhm=None,
velocity=False,
fluxunits=FLAMBDA_CGS,
waveunits=u.angstrom,
name="template",
lorentz=False,
err=None,
):
r"""
Container for template spectra.
Parameters
----------
wave : array-like
Wavelength
In `astropy.units.Angstrom`.
flux : float array-like
If float, then the integrated flux of a Gaussian line. If
array, then f-lambda flux density.
central_wave, fwhm : float
Initialize the template with a Gaussian at this wavelength (in
`astropy.units.Angstrom`.) that has an integrated flux of `flux`
and `fwhm` in `astropy.units.Angstrom` or `km/s` for
`velocity=True`.
velocity : bool
``fwhm`` is a velocity in `km/s`.
fluxunits : astropy.units.Unit
Units of the flux.
Default is `FLAMBDA_CGS` (1e-17 erg/s/cm^2/Angstrom).
waveunits : astropy.units.Unit
Units of the wavelength. Default is Angstrom.
name : str
Name of the template. Default is "template".
lorentz : bool
Make a Lorentzian line instead of a Gaussian.
err : float array-like, optional
Error on the flux.
Attributes
----------
wave, flux : array-like
Passed from the input parameters or generated/modified later.
Methods
-------
__add__, __mul__ : Addition and multiplication of templates.
Examples
--------
.. plot::
:include-source:
import matplotlib.pyplot as plt
ha = SpectrumTemplate(central_wave=6563., fwhm=10)
plt.plot(ha.wave, ha.flux)
ha_z = ha.zscale(0.1)
plt.plot(ha_z.wave, ha_z.flux, label='z=0.1')
plt.legend()
plt.xlabel(r'$\lambda$')
plt.xlim(6000, 7500)
plt.show()
"""
self.wave = wave
if wave is not None:
self.wave = np.asarray(wave,dtype=np.float64)
self.flux = flux
if flux is not None:
self.flux = np.asarray(flux,dtype=np.float64)
if err is not None:
self.err = np.asarray(err,dtype=np.float64)
else:
self.err = None
self.fwhm = None
self.velocity = None
self.fluxunits = fluxunits
self.waveunits = waveunits
self.name = name
if (central_wave is not None) & (fwhm is not None):
self.fwhm = fwhm
self.velocity = velocity
self.wave, self.flux = self.make_gaussian(
central_wave,
fwhm,
wave_grid=wave,
velocity=velocity,
max_sigma=50,
lorentz=lorentz,
)
self.fnu_units = FNU_CGS
self.to_fnu()
[docs] @staticmethod
def make_gaussian(
central_wave,
fwhm,
max_sigma=5,
step=0.1,
wave_grid=None,
velocity=False,
clip=1.0e-6,
lorentz=False,
):
"""
Make Gaussian template
Parameters
----------
central_wave, fwhm : None or float or array-like
Central wavelength and FWHM of the desired Gaussian
velocity : bool
`fwhm` is a velocity.
max_sigma, step : float
Generated wavelength array is
>>> rms = fwhm/2.35
>>> xgauss = np.arange(-max_sigma, max_sigma, step)*rms+central_wave
clip : float
Clip values where the value of the gaussian function is less than
`clip` times its maximum (i.e., `1/sqrt(2*pi*sigma**2)`).
lorentz : bool
Make a Lorentzian line instead of a Gaussian.
Returns
-------
wave, flux : array-like
Wavelength and flux of a Gaussian line
"""
import astropy.constants as const
from astropy.modeling.models import Lorentz1D
if hasattr(fwhm, "unit"):
rms = fwhm.value / 2.35
velocity = u.physical.get_physical_type(fwhm.unit) == "speed"
if velocity:
rms = central_wave * (fwhm / const.c.to(KMS)).value / 2.35
else:
rms = fwhm.value / 2.35
else:
if velocity:
rms = central_wave * (fwhm / const.c.to(KMS).value) / 2.35
else:
rms = fwhm / 2.35
if wave_grid is None:
# print('xxx line', central_wave, max_sigma, rms)
wave_grid = np.arange(-max_sigma, max_sigma, step) * rms
wave_grid += central_wave
wave_grid = np.hstack([91.0, wave_grid, 1.0e8])
if lorentz:
if velocity:
use_fwhm = central_wave * (fwhm / const.c.to(KMS).value)
else:
use_fwhm = fwhm
lmodel = Lorentz1D(amplitude=1, x_0=central_wave, fwhm=use_fwhm)
line = lmodel(wave_grid)
line[0:2] = 0
line[-2:] = 0
line /= np.trapz(line, wave_grid)
peak = line.max()
else:
# Gaussian
line = np.exp(-((wave_grid - central_wave) ** 2) / 2 / rms ** 2)
peak = np.sqrt(2 * np.pi * rms ** 2)
line *= 1.0 / peak # np.sqrt(2*np.pi*rms**2)
line[line < 1.0 / peak * clip] = 0
return wave_grid, line
# self.wave = xgauss
# self.flux = gaussian
[docs] def zscale(self, z, scalar=1, apply_igm=True):
"""
Redshift the template and multiply by a scalar.
Parameters
----------
z : float
Redshift to use.
scalar : float
Multiplicative factor. Additional factor of 1/(1+z) is implicit.
apply_igm : bool
Apply the intergalactic medium (IGM) attenuation correction.
Returns
-------
new_spectrum : `~grizli.utils.SpectrumTemplate`
Redshifted and scaled spectrum.
"""
if apply_igm:
try:
import eazy.igm
igm = eazy.igm.Inoue14()
igmz = igm.full_IGM(z, self.wave * (1 + z))
except:
igmz = 1.0
else:
igmz = 1.0
return SpectrumTemplate(
wave=self.wave * (1 + z), flux=self.flux * scalar / (1 + z) * igmz
)
def __add__(self, spectrum):
"""
Add two templates together
The new wavelength array is the union of both input spectra and each
input spectrum is linearly interpolated to the final grid.
Parameters
----------
spectrum : `~grizli.utils.SpectrumTemplate`
Returns
-------
new_spectrum : `~grizli.utils.SpectrumTemplate`
"""
new_wave = np.unique(np.append(self.wave, spectrum.wave))
new_wave.sort()
new_flux = np.interp(new_wave, self.wave, self.flux)
new_flux += np.interp(new_wave, spectrum.wave, spectrum.flux)
out = SpectrumTemplate(wave=new_wave, flux=new_flux)
out.fwhm = spectrum.fwhm
return out
def __mul__(self, scalar):
"""
Multiply spectrum by a scalar value
Parameters
----------
scalar : float
Factor to multipy to `self.flux`.
Returns
-------
new_spectrum : `~grizli.utils.SpectrumTemplate`
"""
out = SpectrumTemplate(wave=self.wave, flux=self.flux * scalar)
out.fwhm = self.fwhm
return out
[docs] def to_fnu(self, fnu_units=FNU_CGS):
"""
Make fnu version of the template.
Sets the `flux_fnu` attribute, assuming that the wavelength is given
in Angstrom and the flux is given in flambda:
>>> flux_fnu = self.flux * self.wave**2 / 3.e18
"""
# import astropy.constants as const
# flux_fnu = self.flux * self.wave**2 / 3.e18
# flux_fnu = (self.flux*self.fluxunits*(self.wave*self.waveunits)**2/const.c).to(FNU_CGS) #,
if (FNU_CGS.__str__() == "erg / (cm2 Hz s)") & (
self.fluxunits.__str__() == "erg / (Angstrom cm2 s)"
):
# Faster
flux_fnu = self.flux * self.wave ** 2 / 2.99792458e18 * fnu_units
if self.err is not None:
err_fnu = self.err * self.wave ** 2 / 2.99792458e18 * fnu_units
else:
# Use astropy conversion
flux_fnu = (self.flux * self.fluxunits).to(
fnu_units, equivalencies=u.spectral_density(self.wave * self.waveunits)
)
if self.err is not None:
err_fnu = (self.err * self.fluxunits).to(
fnu_units,
equivalencies=u.spectral_density(self.wave * self.waveunits),
)
self.fnu_units = fnu_units
self.flux_fnu = flux_fnu.value
if self.err is not None:
self.err_fnu = err_fnu.value
else:
self.err_fnu = None
[docs] def integrate_filter(self, filter, abmag=False, use_wave="filter"):
"""
Integrate the template through an `~eazy.FilterDefinition` filter
object.
Parameters
----------
filter : `~pysynphot.ObsBandpass`
Or any object that has `wave` and `throughput` attributes, with
the former in the same units as the input spectrum.
abmag : bool
Return AB magnitude rather than fnu flux
use_wave : str, optional
Determines whether to interpolate the template to the filter
wavelengths or the spectrum wavelengths. Default is 'filter'.
Returns
-------
temp_flux : float
Examples
--------
Compute the WFC3/IR F140W AB magnitude of a pure emission line at the
5-sigma 3D-HST line detection limit (5e-17 erg/s/cm2):
>>> import numpy as np
>>> from grizli.utils import SpectrumTemplate
>>> from eazy.filters import FilterDefinition
>>> import pysynphot as S
>>> line = SpectrumTemplate(central_wave=1.4e4, fwhm=150.,
velocity=True)*5.e-17
>>> filter = FilterDefinition(bp=S.ObsBandpass('wfc3,ir,f140w'))
>>> fnu = line.integrate_filter(filter)
>>> print('AB mag = {0:.3f}'.format(-2.5*np.log10(fnu)-48.6))
AB mag = 26.619
"""
INTEGRATOR = np.trapz
try:
from .utils_numba.interp import interp_conserve_c
interp = interp_conserve_c
except ImportError:
interp = np.interp
# wz = self.wave*(1+z)
nonzero = filter.throughput > 0
if (
(filter.wave[nonzero].min() > self.wave.max())
| (filter.wave[nonzero].max() < self.wave.min())
| (filter.wave[nonzero].min() < self.wave.min())
):
if self.err is None:
return 0.0
else:
return 0.0, 0.0
if use_wave == "filter":
# Interpolate to filter wavelengths
integrate_wave = filter.wave
integrate_templ = interp(
filter.wave.astype(np.float64),
self.wave,
self.flux_fnu,
left=0,
right=0,
)
if self.err is not None:
templ_ivar = (
1.0
/ interp(filter.wave.astype(np.float64), self.wave, self.err_fnu)
** 2
)
templ_ivar[~np.isfinite(templ_ivar)] = 0
integrate_weight = (
filter.throughput / filter.wave * templ_ivar / filter.norm
)
else:
integrate_weight = filter.throughput / filter.wave
else:
# Interpolate to spectrum wavelengths
integrate_wave = self.wave
integrate_templ = self.flux_fnu
# test = nonzero
test = np.isfinite(filter.throughput)
interp_thru = interp(
integrate_wave,
filter.wave[test],
filter.throughput[test],
left=0,
right=0,
)
if self.err is not None:
templ_ivar = 1 / self.err_fnu ** 2
templ_ivar[~np.isfinite(templ_ivar)] = 0
integrate_weight = (
interp_thru / integrate_wave * templ_ivar / filter.norm
)
else:
integrate_weight = interp_thru / integrate_wave # /templ_err**2
if hasattr(filter, "norm") & (self.err is None):
filter_norm = filter.norm
else:
# e.g., pysynphot bandpass
filter_norm = INTEGRATOR(integrate_weight, integrate_wave)
# f_nu/lam dlam == f_nu d (ln nu)
temp_flux = (
INTEGRATOR(integrate_templ * integrate_weight, integrate_wave) / filter_norm
)
if self.err is not None:
temp_err = 1 / np.sqrt(filter_norm)
if abmag:
temp_mag = -2.5 * np.log10(temp_flux) - 48.6
return temp_mag
else:
if self.err is not None:
return temp_flux, temp_err
else:
return temp_flux
@property
def eazy(self):
"""
Convert to `eazy.template.Template` object
"""
import eazy.templates
templ = eazy.templates.Template(arrays=(self.wave, self.flux), name=self.name)
return templ
[docs]def load_templates(
fwhm=400,
line_complexes=True,
stars=False,
full_line_list=DEFAULT_LINE_LIST,
continuum_list=None,
fsps_templates=False,
alf_template=False,
lorentz=False,
):
"""
Generate a list of templates for fitting to the grism spectra
The different sets of continuum templates are stored in
>>> temp_dir = os.path.join(GRIZLI_PATH, 'templates')
Parameters
----------
fwhm : float, optional
FWHM of a Gaussian, in km/s, that is convolved with the emission
line templates. If too narrow, then can see pixel effects in the
fits as a function of redshift. Default is 400.
line_complexes : bool, optional
Generate line complex templates with fixed flux ratios rather than
individual lines. This is useful for the redshift fits where there
would be redshift degeneracies if the line fluxes for individual
lines were allowed to vary completely freely. See the list of
available lines and line groups in
`~grizli.utils.get_line_wavelengths`. Currently,
`line_complexes=True` generates the following groups:
Ha+NII+SII+SIII+He
OIII+Hb
OII+Ne
stars : bool, optional
Get stellar templates rather than galaxies + lines. Default is False.
full_line_list : None or list, optional
Full set of lines to try. The default is set in the global variable
`~grizli.utils.DEFAULT_LINE_LIST`.
The full list of implemented lines is in `~grizli.utils.get_line_wavelengths`.
continuum_list : None or list, optional
Override the default continuum templates if None.
fsps_templates : bool, optional
If True, get the FSPS NMF templates. Default is False.
alf_template : bool, optional
If True, include Alf templates. Default is False.
lorentz : bool, optional
If True, use Lorentzian line profiles instead of Gaussian.
Default is False.
Returns
-------
temp_list : dictionary of `~grizli.utils.SpectrumTemplate` objects
Output template list
"""
if stars:
# templates = glob.glob('%s/templates/Pickles_stars/ext/*dat' %(GRIZLI_PATH))
# templates = []
# for t in 'obafgkmrw':
# templates.extend( glob.glob('%s/templates/Pickles_stars/ext/uk%s*dat' %(os.getenv('THREEDHST'), t)))
# templates.extend(glob.glob('%s/templates/SPEX/spex-prism-M*txt' %(os.getenv('THREEDHST'))))
# templates.extend(glob.glob('%s/templates/SPEX/spex-prism-[LT]*txt' %(os.getenv('THREEDHST'))))
#
# #templates = glob.glob('/Users/brammer/Downloads/templates/spex*txt')
# templates = glob.glob('bpgs/*ascii')
# info = catIO.Table('bpgs/bpgs.info')
# type = np.array([t[:2] for t in info['type']])
# templates = []
# for t in 'OBAFGKM':
# test = type == '-%s' %(t)
# so = np.argsort(info['type'][test])
# templates.extend(info['file'][test][so])
#
# temp_list = OrderedDict()
# for temp in templates:
# #data = np.loadtxt('bpgs/'+temp, unpack=True)
# data = np.loadtxt(temp, unpack=True)
# #data[0] *= 1.e4 # spex
# scl = np.interp(5500., data[0], data[1])
# name = os.path.basename(temp)
# #ix = info['file'] == temp
# #name='%5s %s' %(info['type'][ix][0][1:], temp.split('.as')[0])
# print(name)
# temp_list[name] = utils.SpectrumTemplate(wave=data[0],
# flux=data[1]/scl)
# np.save('stars_bpgs.npy', [temp_list])
# tall = np.load(os.path.join(GRIZLI_PATH,
# 'templates/stars.npy'))[0]
#
# return tall
#
# temp_list = OrderedDict()
# for k in tall:
# if k.startswith('uk'):
# temp_list[k] = tall[k]
#
# return temp_list
#
# for t in 'MLT':
# for k in tall:
# if k.startswith('spex-prism-'+t):
# temp_list[k] = tall[k]
#
# return temp_list
# return temp_list
templates = [
"M6.5.txt",
"M8.0.txt",
"L1.0.txt",
"L3.5.txt",
"L6.0.txt",
"T2.0.txt",
"T6.0.txt",
"T7.5.txt",
]
templates = ["stars/" + t for t in templates]
else:
# Intermediate and very old
# templates = ['templates/EAZY_v1.0_lines/eazy_v1.0_sed3_nolines.dat',
# 'templates/cvd12_t11_solar_Chabrier.extend.skip10.dat']
templates = ["eazy_intermediate.dat", "cvd12_t11_solar_Chabrier.dat"]
# Post starburst
# templates.append('templates/UltraVISTA/eazy_v1.1_sed9.dat')
templates.append("post_starburst.dat")
# Very blue continuum
# templates.append('templates/YoungSB/erb2010_continuum.dat')
templates.append("erb2010_continuum.dat")
# Test new templates
# templates = ['templates/erb2010_continuum.dat',
# 'templates/fsps/tweak_fsps_temp_kc13_12_006.dat',
# 'templates/fsps/tweak_fsps_temp_kc13_12_008.dat']
if fsps_templates:
# templates = ['templates/fsps/tweak_fsps_temp_kc13_12_0{0:02d}.dat'.format(i+1) for i in range(12)]
templates = [
"fsps/fsps_QSF_12_v3_nolines_0{0:02d}.dat".format(i + 1)
for i in range(12)
]
# templates = ['fsps/fsps_QSF_7_v3_nolines_0{0:02d}.dat'.format(i+1) for i in range(7)]
if alf_template:
templates.append("alf_SSP.dat")
if continuum_list is not None:
templates = continuum_list
temp_list = OrderedDict()
for temp in templates:
data = np.loadtxt(os.path.join(GRIZLI_PATH, "templates", temp), unpack=True)
# scl = np.interp(5500., data[0], data[1])
scl = 1.0
name = temp # os.path.basename(temp)
temp_list[name] = SpectrumTemplate(wave=data[0], flux=data[1] / scl, name=name)
temp_list[name].name = name
if stars:
return temp_list
# Emission lines:
line_wavelengths, line_ratios = get_line_wavelengths()
if line_complexes:
# line_list = ['Ha+SII', 'OIII+Hb+Ha', 'OII']
# line_list = ['Ha+SII', 'OIII+Hb', 'OII']
# line_list = ['Ha+NII+SII+SIII+He+PaB', 'OIII+Hb', 'OII+Ne', 'Lya+CIV']
# line_list = ['Ha+NII+SII+SIII+He+PaB', 'OIII+Hb+Hg+Hd', 'OII+Ne', 'Lya+CIV']
line_list = [
"Ha+NII+SII+SIII+He+PaB",
"OIII+Hb+Hg+Hd",
"OII+Ne",
"Gal-UV-lines",
]
else:
if full_line_list is None:
line_list = DEFAULT_LINE_LIST
else:
line_list = full_line_list
# line_list = ['Ha', 'SII']
# Use FSPS grid for lines
wave_grid = None
# if fsps_templates:
# wave_grid = data[0]
# else:
# wave_grid = None
for li in line_list:
scl = line_ratios[li] / np.sum(line_ratios[li])
for i in range(len(scl)):
if ("O32" in li) & (np.abs(line_wavelengths[li][i] - 2799) < 2):
fwhm_i = 2500
lorentz_i = True
else:
fwhm_i = fwhm
lorentz_i = lorentz
line_i = SpectrumTemplate(
wave=wave_grid,
central_wave=line_wavelengths[li][i],
flux=None,
fwhm=fwhm_i,
velocity=True,
lorentz=lorentz_i,
)
if i == 0:
line_temp = line_i * scl[i]
else:
line_temp = line_temp + line_i * scl[i]
name = "line {0}".format(li)
line_temp.name = name
temp_list[name] = line_temp
return temp_list
[docs]def load_beta_templates(wave=np.arange(400, 2.5e4), betas=[-2, -1, 0]):
"""
Step-function templates with f_lambda ~ (wave/1216.)**beta
Parameters
----------
wave: array_like
The wavelength grid.
beta: float
The power-law index.
Returns
-------
t0: dict
A dictionary containing the step-function templates.
"""
t0 = {}
for beta in betas:
key = "beta {0}".format(beta)
t0[key] = SpectrumTemplate(wave=wave, flux=(wave / 1216.0) ** beta)
return t0
[docs]def load_quasar_templates(
broad_fwhm=2500,
narrow_fwhm=1200,
broad_lines=[
"HeI-5877",
"MgII",
"Lya",
"CIV-1549",
"CIII-1906",
"CIII-1908",
"OIII-1663",
"HeII-1640",
"SiIV+OIV-1398",
"NIV-1487",
"NV-1240",
"PaB",
"PaG",
],
narrow_lines=[
"NIII-1750",
"OII",
"OIII",
"SII",
"OI-6302",
"OIII-4363",
"NeIII-3867",
"NeVI-3426",
"NeV-3346",
"OII-7325",
"ArIII-7138",
"SIII",
"HeI-1083",
],
include_feii=True,
slopes=[-2.8, 0, 2.8],
uv_line_complex=True,
fixed_narrow_lines=False,
t1_only=False,
nspline=13,
Rspline=30,
betas=None,
include_reddened_balmer_lines=False,
):
"""
Make templates suitable for fitting broad-line quasars
Parameters
----------
broad_fwhm : float, optional
Full width at half maximum of the broad lines. Default is 2500.
narrow_fwhm : float, optional
Full width at half maximum of the narrow lines. Default is 1200.
broad_lines : list, optional
List of broad lines to include in the templates.
narrow_lines : list, optional
List of narrow lines to include in the templates.
include_feii : bool, optional
Whether to include Fe II templates. Default is True.
slopes : list, optional
List of slopes for linear continua. Default is [-2.8, 0, 2.8].
uv_line_complex : bool, optional
Whether to include UV line complex templates. Default is True.
fixed_narrow_lines : bool, optional
Whether to fix the narrow lines. Default is False.
t1_only : bool, optional
Whether to only include t1 templates. Default is False.
nspline : int, optional
Number of spline continua templates. Default is 13.
Rspline : int, optional
Resolution of the spline continua templates. Default is 30.
betas : list, optional
List of beta values for beta templates. Default is None.
include_reddened_balmer_lines : bool, optional
Whether to include reddened Balmer lines. Default is False.
Returns
-------
t0 : OrderedDict
Dictionary of templates for t0.
t1 : OrderedDict
Dictionary of templates for t1.
"""
from collections import OrderedDict
import scipy.ndimage as nd
t0 = OrderedDict()
t1 = OrderedDict()
broad1 = load_templates(
fwhm=broad_fwhm,
line_complexes=False,
stars=False,
full_line_list=["Ha", "Hb", "Hg", "Hd", "H7", "H8", "H9", "H10"] + broad_lines,
continuum_list=[],
fsps_templates=False,
alf_template=False,
lorentz=True,
)
narrow1 = load_templates(
fwhm=400,
line_complexes=False,
stars=False,
full_line_list=narrow_lines,
continuum_list=[],
fsps_templates=False,
alf_template=False,
)
if fixed_narrow_lines:
if t1_only:
narrow0 = narrow1
else:
narrow0 = load_templates(
fwhm=narrow_fwhm,
line_complexes=False,
stars=False,
full_line_list=["QSO-Narrow-lines"],
continuum_list=[],
fsps_templates=False,
alf_template=False,
)
else:
narrow0 = load_templates(
fwhm=narrow_fwhm,
line_complexes=False,
stars=False,
full_line_list=narrow_lines,
continuum_list=[],
fsps_templates=False,
alf_template=False,
)
if t1_only:
broad0 = broad1
else:
if uv_line_complex:
full_line_list = ["Balmer 10kK + MgII Av=0.5", "QSO-UV-lines"]
else:
full_line_list = ["Balmer 10kK + MgII Av=0.5"]
if include_reddened_balmer_lines:
line_wavelengths, line_ratios = get_line_wavelengths()
if "Balmer 10kK + MgII Av=1.0" in line_wavelengths:
full_line_list += ["Balmer 10kK + MgII"]
full_line_list += ["Balmer 10kK + MgII Av=1.0"]
full_line_list += ["Balmer 10kK + MgII Av=2.0"]
broad0 = load_templates(
fwhm=broad_fwhm,
line_complexes=False,
stars=False,
full_line_list=full_line_list,
continuum_list=[],
fsps_templates=False,
alf_template=False,
lorentz=True,
)
for k in broad0:
t0[k] = broad0[k]
for k in broad1:
t1[k] = broad1[k]
for k in narrow0:
t0[k] = narrow0[k]
for k in narrow1:
t1[k] = narrow1[k]
# Fe II
if include_feii:
feii_wave, feii_flux = np.loadtxt(
os.path.dirname(__file__) + "/data/templates/FeII_VeronCetty2004.txt",
unpack=True,
)
# smoothing, in units of input velocity resolution
feii_kern = broad_fwhm / 2.3548 / 75.0
feii_sm = nd.gaussian_filter(feii_flux, feii_kern)
t0["FeII-VC2004"] = t1["FeII-VC2004"] = SpectrumTemplate(
wave=feii_wave, flux=feii_sm, name="FeII-VC2004"
)
# Linear continua
# cont_wave = np.arange(400, 2.5e4)
# for slope in slopes:
# key = 'slope {0}'.format(slope)
# t0[key] = t1[key] = SpectrumTemplate(wave=cont_wave, flux=(cont_wave/6563.)**slope)
if Rspline is not None:
wspline = np.arange(4200, 2.5e4, 10)
df_spl = log_zgrid(zr=[wspline[0], wspline[-1]], dz=1.0 / Rspline)
bsplines = bspline_templates(wspline, df=len(df_spl) + 2, log=True, clip=0.0001)
for key in bsplines:
t0[key] = t1[key] = bsplines[key]
elif nspline > 0:
# Spline continua
cont_wave = np.arange(5000, 2.4e4)
bsplines = bspline_templates(cont_wave, df=nspline, log=True)
for key in bsplines:
t0[key] = t1[key] = bsplines[key]
elif betas is not None:
btemp = load_beta_templates(wave=np.arange(400, 2.5e4), betas=betas)
for key in btemp:
t0[key] = t1[key] = btemp[key]
else:
# Observed frame steps
onedR = -nspline
wlim = [5000, 18000.0]
bin_steps, step_templ = step_templates(wlim=wlim, R=onedR, round=10)
for key in step_templ:
t0[key] = t1[key] = step_templ[key]
# t0['blue'] = t1['blue'] = SpectrumTemplate(wave=cont_wave, flux=(cont_wave/6563.)**-2.8)
# t0['mid'] = t1['mid'] = SpectrumTemplate(wave=cont_wave, flux=(cont_wave/6563.)**0)
# t0['red'] = t1['mid'] = SpectrumTemplate(wave=cont_wave, flux=(cont_wave/6563.)**2.8)
return t0, t1
PHOENIX_LOGG_FULL = [3.0, 3.5, 4.0, 4.5, 5.0, 5.5]
PHOENIX_LOGG = [4.0, 4.5, 5.0, 5.5]
PHOENIX_TEFF_FULL = [400.0, 420.0, 450.0, 500.0, 550.0, 600.0, 650.0, 700.0, 750.0,
800.0, 850.0, 900.0, 950.0, 1000.0, 1050.0, 1100.0, 1150.0,
1200.0, 1250.0, 1300.0, 1350.0, 1400.0, 1450.0, 1500.0, 1550.0,
1600.0, 1650.0, 1700.0, 1750.0, 1800.0, 1850.0, 1900.0, 1950.0,
2000.0, 2100.0, 2200.0, 2300.0, 2400.0, 2500.0, 2600.0, 2700.0,
2800.0, 2900.0, 3000.0, 3100.0, 3200.0, 3300.0, 3400.0, 3500.0,
3600.0, 3700.0, 3800.0, 3900.0, 4000.0, 4100.0, 4200.0, 4300.0,
4400.0, 4500.0, 4600.0, 4700.0, 4800.0, 4900.0, 5000.0]
PHOENIX_TEFF = [400.0, 420.0, 450.0, 500.0, 550.0, 600.0, 650.0, 700.0, 750.0,
800.0, 850.0, 900.0, 950.0, 1000.0, 1050.0, 1100.0, 1150.0,
1200.0, 1300.0, 1400.0, 1500.0, 1600.0, 1700.0, 1800.0, 1900.0,
2000.0, 2100.0, 2200.0, 2300.0, 2400.0, 2500.0, 2600.0, 2700.0,
2800.0, 2900.0, 3000.0, 3100.0, 3200.0, 3300.0, 3400.0, 3500.0,
3600.0, 3700.0, 3800.0, 3900.0, 4000.0, 4200.0, 4400.0, 4600.0,
4800.0, 5000.0, 5500.0, 5500, 6000.0, 6500.0, 7000.0]
PHOENIX_ZMET_FULL = [-2.5, -2.0, -1.5, -1.0, -0.5, -0.0, 0.5]
PHOENIX_ZMET = [-1.0, -0.5, -0.0]
[docs]def load_phoenix_stars(
logg_list=PHOENIX_LOGG,
teff_list=PHOENIX_TEFF,
zmet_list=PHOENIX_ZMET,
add_carbon_star=True,
file="bt-settl_t400-7000_g4.5.fits",
):
"""
Load Phoenix stellar templates
Parameters
----------
logg_list : list, optional
List of log(g) values for the templates to load.
teff_list : list, optional
List of effective temperature values for the templates to load.
zmet_list : list, optional
List of metallicity values for the templates to load.
add_carbon_star : bool, optional
Whether to include a carbon star template.
file : str, optional
Name of the FITS file containing the templates.
Default is "bt-settl_t400-7000_g4.5.fits".
Returns
-------
tstars : OrderedDict
Dictionary of SpectrumTemplate objects, with the template names as keys.
"""
from collections import OrderedDict
try:
from urllib.request import urlretrieve
except:
from urllib import urlretrieve
# file='bt-settl_t400-5000_g4.5.fits'
# file='bt-settl_t400-3500_z0.0.fits'
try:
hdu = pyfits.open(os.path.join(GRIZLI_PATH, "templates/stars/", file))
except:
# url = 'https://s3.amazonaws.com/grizli/CONF'
# url = 'https://erda.ku.dk/vgrid/Gabriel%20Brammer/CONF'
url = "https://raw.githubusercontent.com/gbrammer/" + "grizli-config/master"
print("Fetch {0}/{1}".format(url, file))
# os.system('wget -O /tmp/{1} {0}/{1}'.format(url, file))
res = urlretrieve(
"{0}/{1}".format(url, file), filename=os.path.join("/tmp", file)
)
hdu = pyfits.open(os.path.join("/tmp/", file))
tab = GTable.gread(hdu[1])
hdu.close()
tstars = OrderedDict()
N = tab["flux"].shape[1]
for i in range(N):
teff = tab.meta["TEFF{0:03d}".format(i)]
logg = tab.meta["LOGG{0:03d}".format(i)]
try:
met = tab.meta["ZMET{0:03d}".format(i)]
except:
met = 0.0
if (logg not in logg_list) | (teff not in teff_list) | (met not in zmet_list):
# print('Skip {0} {1}'.format(logg, teff))
continue
label = "bt-settl_t{0:05.0f}_g{1:3.1f}_m{2:.1f}".format(teff, logg, met)
tstars[label] = SpectrumTemplate(
wave=tab["wave"], flux=tab["flux"][:, i], name=label
)
if add_carbon_star:
cfile = os.path.join(GRIZLI_PATH, "templates/stars/carbon_star.txt")
sp = read_catalog(cfile)
if add_carbon_star > 1:
import scipy.ndimage as nd
cflux = nd.gaussian_filter(sp["flux"], add_carbon_star)
else:
cflux = sp["flux"]
tstars["bt-settl_t05000_g0.0_m0.0"] = SpectrumTemplate(
wave=sp["wave"], flux=cflux, name="carbon-lancon2002"
)
return tstars
[docs]def load_sdss_pca_templates(file="spEigenQSO-55732.fits", smooth=3000):
"""
Load SDSS eigen PCA templates
Parameters
----------
file : str, optional
The name of the FITS file containing the templates.
Default is "spEigenQSO-55732.fits".
smooth : float, optional
The smoothing parameter for the templates. Default is 3000.
Returns
-------
temp_list : OrderedDict
A dictionary of SpectrumTemplate objects representing the SDSS eigen templates.
"""
from collections import OrderedDict
import scipy.ndimage as nd
im = pyfits.open(os.path.join(GRIZLI_PATH, "templates", file))
h = im[0].header
log_wave = np.arange(h["NAXIS1"]) * h["COEFF1"] + h["COEFF0"]
wave = 10 ** log_wave
name = file.split(".fits")[0]
if smooth > 0:
dv_in = h["COEFF1"] * 3.0e5
n = smooth / dv_in
data = nd.gaussian_filter1d(im[0].data, n, axis=1).astype(np.float64)
skip = int(n / 2.5)
wave = wave[::skip]
data = data[:, ::skip]
else:
data = im[0].data.astype(np.float64)
N = h["NAXIS2"]
temp_list = OrderedDict()
for i in range(N):
temp_list["{0} {1}".format(name, i + 1)] = SpectrumTemplate(
wave=wave, flux=data[i, :]
)
im.close()
return temp_list
[docs]def cheb_templates(
wave, order=6, get_matrix=False, log=False, clip=1.0e-4, minmax=None
):
"""
Chebyshev polynomial basis functions
Parameters
----------
wave : array-like
The wavelength array.
order : int
The order of the Chebyshev polynomial.
get_matrix : bool, optional
If True, return array data rather than template objects.
Default is False.
log : bool, optional
If True, use the logarithm of the wavelength array. Default is False.
clip : float, optional
The clipping threshold for wavelengths outside the range.
Default is 1.0e-4.
minmax : array-like, optional
The minimum and maximum values of the wavelength range.
If not provided, the minimum and maximum values of
the input wavelength array are used.
Returns
-------
templates : OrderedDict
The Chebyshev polynomial templates.
"""
from numpy.polynomial.chebyshev import chebval, chebvander
if minmax is None:
mi = wave.min()
ma = wave.max()
else:
mi, ma = np.squeeze(minmax) * 1.0
if log:
xi = np.log(wave)
mi = np.log(mi)
ma = np.log(ma)
else:
xi = wave * 1
x = (xi - mi) * 2 / (ma - mi) - 1
n_bases = order + 1
basis = chebvander(x, order)
# for i in range(n_bases):
out_of_range = (xi < mi) | (xi > ma)
basis[out_of_range, :] = 0
if get_matrix:
return basis
temp = OrderedDict()
for i in range(n_bases):
key = f"cheb o{i}"
temp[key] = SpectrumTemplate(wave, basis[:, i])
temp[key].name = key
return temp
[docs]def bspline_templates(
wave, degree=3, df=6, get_matrix=False, log=False, clip=1.0e-4, minmax=None
):
"""
B-spline basis functions, modeled after `~patsy.splines`
Parameters
----------
wave : array-like
The wavelength array.
degree : int
The degree of the B-spline basis functions.
df : int
The number of degrees of freedom.
get_matrix : bool
If True, return the basis function matrix.
log : bool
If True, use the logarithm of the wavelength array.
clip : float
The threshold for clipping the basis functions.
minmax : tuple
The minimum and maximum values for the wavelength array.
Returns
-------
basis : array-like
The B-spline basis functions.
temp : OrderedDict
The B-spline templates.
"""
from scipy.interpolate import splev
order = degree + 1
n_inner_knots = df - order
inner_knots = np.linspace(0, 1, n_inner_knots + 2)[1:-1]
norm_knots = np.concatenate(([0, 1] * order, inner_knots))
norm_knots.sort()
if log:
xspl = np.log(wave)
else:
xspl = wave * 1
if minmax is None:
mi = xspl.min()
ma = xspl.max()
else:
mi, ma = minmax
width = ma - mi
all_knots = norm_knots * width + mi
n_bases = len(all_knots) - (degree + 1)
basis = np.empty((xspl.shape[0], n_bases), dtype=float)
coefs = np.identity(n_bases)
basis = splev(xspl, (all_knots, coefs, degree))
for i in range(n_bases):
out_of_range = (xspl < mi) | (xspl > ma)
basis[i][out_of_range] = 0
wave_peak = np.round(wave[np.argmax(basis, axis=1)])
maxval = np.max(basis, axis=1)
for i in range(n_bases):
basis[i][basis[i] < clip * maxval[i]] = 0
if get_matrix:
return np.vstack(basis).T
temp = OrderedDict()
for i in range(n_bases):
key = "bspl {0} {1:.0f}".format(i, wave_peak[i])
temp[key] = SpectrumTemplate(wave, basis[i])
temp[key].name = key
temp[key].wave_peak = wave_peak[i]
temp.knots = all_knots
temp.degree = degree
temp.xspl = xspl
return temp
[docs]def eval_bspline_templates(wave, bspl, coefs):
"""
Evaluate B-spline templates at given wavelengths.
Parameters
----------
wave : array-like
The wavelengths at which to evaluate the B-spline templates.
bspl : scipy.interpolate.BSpline
The B-spline object defining the basis functions.
coefs : array-like
The coefficients of the B-spline basis functions.
Returns
-------
array-like:
The evaluated B-spline templates at the given wavelengths.
"""
from scipy.interpolate import splev
xspl = np.log(wave)
basis = splev(xspl, (bspl.knots, coefs, bspl.degree))
return np.array(basis)
[docs]def split_spline_template(templ, wavelength_range=[5000, 2.4e4], Rspline=10, log=True):
"""
Multiply a single template by spline bases to effectively generate a
spline multiplicative correction that can be fit with linear least
squares.
Parameters
----------
templ : `~grizli.utils.SpectrumTemplate`
Template to split.
wavelength_range : [float, float]
Limit the splines to this wavelength range
Rspline : float
Effective resolution, R=dlambda/lambda, of the spline correction
function.
log : bool
Log-spaced splines
Returns
-------
stemp : dict
Dictionary of spline-component templates, with additional attributes:
wspline = wavelength of the templates / spline correction
tspline = matrix of the spline corrections
knots = peak wavelenghts of each spline component
"""
from collections import OrderedDict
from grizli import utils
if False:
stars = utils.load_templates(stars=True)
templ = stars["stars/L1.0.txt"]
wspline = templ.wave
clip = (wspline > wavelength_range[0]) & (wspline < wavelength_range[1])
df_spl = len(utils.log_zgrid(zr=wavelength_range, dz=1.0 / Rspline))
tspline = utils.bspline_templates(
wspline[clip], df=df_spl + 2, log=log, clip=0.0001, get_matrix=True
)
ix = np.argmax(tspline, axis=0)
knots = wspline[clip][ix]
N = tspline.shape[1]
stemp = OrderedDict()
for i in range(N):
name = "{0} {1:.2f}".format(templ.name, knots[i] / 1.0e4)
stemp[name] = utils.SpectrumTemplate(
wave=wspline[clip], flux=templ.flux[clip] * tspline[:, i], name=name
)
stemp[name].knot = knots[i]
stemp.wspline = wspline[clip]
stemp.tspline = tspline
stemp.knots = knots
return stemp
[docs]def step_templates(
wlim=[5000, 1.8e4],
bin_steps=None,
R=30,
round=10,
rest=False,
special=None,
order=0,
):
"""
Step-function templates for easy binning
Parameters
----------
wlim : list
The wavelength range for the templates.
bin_steps : ndarray
The array of bin steps for the templates.
R : int
The resolution of the templates.
round : int
The rounding factor for the bin steps.
rest : bool
Flag indicating whether the templates are in the rest frame.
special : str
Special template type. Options are 'D4000', 'Dn4000', and None.
order : int
The order of the step function templates.
Returns
-------
bin_steps : ndarray
The array of bin steps for the templates.
step_templ : dict
Dictionary of step function templates.
"""
if special == "Dn4000":
rest = True
bin_steps = np.hstack(
[
np.arange(850, 3849, 100),
[3850, 3950, 4000, 4100],
np.arange(4200, 1.7e4, 100),
]
)
elif special == "D4000":
rest = True
bin_steps = np.hstack(
[
np.arange(850, 3749, 200),
[3750, 3950, 4050, 4250],
np.arange(4450, 1.7e4, 200),
]
)
elif special not in ["D4000", "Dn4000", None]:
print(
"step_templates: {0} not recognized (options are 'D4000', 'Dn4000', and None)".format(
special
)
)
return {}
if bin_steps is None:
bin_steps = np.round(log_zgrid(wlim, 1.0 / R) / round) * round
else:
wlim = [bin_steps[0], bin_steps[-1]]
ds = np.diff(bin_steps)
xspec = np.arange(wlim[0] - ds[0], wlim[1] + ds[-1])
bin_mid = bin_steps[:-1] + ds / 2.0
step_templ = {}
for i in range(len(bin_steps) - 1):
yspec = ((xspec >= bin_steps[i]) & (xspec < bin_steps[i + 1])) * 1
for o in range(order + 1):
label = "step {0:.0f}-{1:.0f} {2}".format(bin_steps[i], bin_steps[i + 1], o)
if rest:
label = "r" + label
flux = ((xspec - bin_mid[i]) / ds[i]) ** o * (yspec > 0)
step_templ[label] = SpectrumTemplate(wave=xspec, flux=flux, name=label)
return bin_steps, step_templ
[docs]def polynomial_templates(wave, ref_wave=1.0e4, order=0, line=False):
"""
Generate polynomial templates based on the input parameters.
If `line` is True, the method generates line templates by applying
a sign to the polynomial. Otherwise, it generates polynomial templates by
raising the wavelength ratio to the power of the polynomial order.
Each template is stored in the returned dictionary with a key in the format
"poly {i}", where i is the polynomial order.
Parameters
----------
wave : array-like
The wavelength array.
ref_wave : float, optional
The reference wavelength. Default is 1.0e4.
order : int, optional
The order of the polynomial. Default is 0.
line : bool, optional
Whether to generate line templates. Default is False.
Returns
-------
temp: OrderedDict
A dictionary of SpectrumTemplate objects
representing the polynomial templates.
"""
temp = OrderedDict()
if line:
for sign in [1, -1]:
key = "poly {0}".format(sign)
temp[key] = SpectrumTemplate(wave, sign * (wave / ref_wave - 1) + 1)
temp[key].name = key
return temp
for i in range(order + 1):
key = "poly {0}".format(i)
temp[key] = SpectrumTemplate(wave, (wave / ref_wave - 1) ** i)
temp[key].name = key
temp[key].ref_wave = ref_wave
return temp
[docs]def split_poly_template(templ, ref_wave=1.0e4, order=3):
"""
Multiply a single template by polynomial bases to effectively generate a
polynomial multiplicative correction that can be fit with linear least
squares.
Parameters
----------
templ : `~grizli.utils.SpectrumTemplate`
Template to split.
ref_wave : float
Wavelength where to normalize the polynomials.
Order : int
Polynomial order. Returns order+1 templates.
Returns
-------
ptemp : dict
Dictionary of polynomial-component templates, with additional
attributes:
ref_wave = wavelength where polynomials normalized
"""
from collections import OrderedDict
from grizli import utils
tspline = polynomial_templates(
templ.wave, ref_wave=ref_wave, order=order, line=False
)
ptemp = OrderedDict()
for i, t in enumerate(tspline):
name = "{0} poly {1}".format(templ.name, i)
ptemp[name] = utils.SpectrumTemplate(
wave=templ.wave, flux=templ.flux * tspline[t].flux, name=name
)
ptemp[name].ref_wave = ref_wave
ptemp.ref_wave = ref_wave
return ptemp
[docs]def dot_templates(coeffs, templates, z=0, max_R=5000, apply_igm=True):
"""
Compute template sum analogous to `np.dot(coeffs, templates)`.
Parameters
----------
coeffs : array-like
Coefficients for each template.
templates : list of `~grizli.utils.SpectrumTemplate`
List of templates.
z : float, optional
Redshift to apply to the templates (default is 0).
max_R : float, optional
Maximum spectral resolution to apply to the templates (default is 5000).
apply_igm : bool, optional
Apply intergalactic medium (IGM) attenuation to the templates.
Returns
-------
tc : `~grizli.utils.SpectrumTemplate`
Continuum template.
tl : `~grizli.utils.SpectrumTemplate`
Full template (including lines).
"""
if len(coeffs) != len(templates):
raise ValueError(
"shapes of coeffs ({0}) and templates ({1}) don't match".format(
len(coeffs), len(templates)
)
)
wave, flux_arr, is_line = array_templates(
templates, max_R=max_R, z=z, apply_igm=apply_igm
)
# Continuum
cont = np.dot(coeffs * (~is_line), flux_arr)
tc = SpectrumTemplate(wave=wave, flux=cont).zscale(z, apply_igm=False)
# Full template
line = np.dot(coeffs, flux_arr)
tl = SpectrumTemplate(wave=wave, flux=line).zscale(z, apply_igm=False)
return tc, tl
[docs]def array_templates(templates, wave=None, max_R=5000, z=0, apply_igm=False):
"""
Return an array version of the templates
that have all been interpolated to the same grid.
Parameters
----------
templates : dictionary of `~grizli.utils.SpectrumTemplate` objects
Output template list with `NTEMP` templates.
max_R : float
Maximum spectral resolution of the regridded templates.
z : float
Redshift where to evaluate the templates. But note that this is only
used to shift templates produced by `bspline_templates`, which are
defined in the observed frame.
wave : `~numpy.ndarray`, dimensions `(NL,)`, optional
Array containing unique wavelengths. If not provided, the wavelengths
will be determined from the templates.
flux_arr : `~numpy.ndarray`, dimensions `(NTEMP, NL)`, optional
Array containing the template fluxes interpolated at `wave`. If not
provided, the fluxes will be computed from the templates.
is_line : `~numpy.ndarray`, optional
Boolean array indicating emission line templates (the key in the
template dictionary starts with "line ").
Returns
-------
wave : `~numpy.ndarray`, dimensions `(NL,)`
Array containing unique wavelengths.
flux_arr : `~numpy.ndarray`, dimensions `(NTEMP, NL)`
Array containing the template fluxes interpolated at `wave`.
is_line : `~numpy.ndarray`
Boolean array indicating emission line templates (the key in the
template dictionary starts with "line ").
"""
from grizli.utils_numba.interp import interp_conserve_c
if wave is None:
wstack = []
for t in templates:
if t.split()[0] in ["bspl", "step", "poly"]:
wstack.append(templates[t].wave / (1 + z))
else:
wstack.append(templates[t].wave)
wave = np.unique(np.hstack(wstack))
clipsum, iter = 1, 0
while (clipsum > 0) & (iter < 10):
clip = np.gradient(wave) / wave < 1 / max_R
idx = np.arange(len(wave))[clip]
wave[idx[::2]] = np.nan
wave = wave[np.isfinite(wave)]
iter += 1
clipsum = clip.sum()
# print(iter, clipsum)
NTEMP = len(templates)
flux_arr = np.zeros((NTEMP, len(wave)))
for i, t in enumerate(templates):
if t.split()[0] in ["bspl", "step", "poly"]:
flux_arr[i, :] = interp_conserve_c(
wave, templates[t].wave / (1 + z), templates[t].flux * (1 + z)
)
else:
if hasattr(templates[t], "flux_flam"):
# Redshift-dependent eazy-py Template
flux_arr[i, :] = interp_conserve_c(
wave, templates[t].wave, templates[t].flux_flam(z=z)
)
else:
flux_arr[i, :] = interp_conserve_c(
wave, templates[t].wave, templates[t].flux
)
is_line = np.array([t.startswith("line ") for t in templates])
# IGM
if apply_igm:
try:
import eazy.igm
IGM = eazy.igm.Inoue14()
lylim = wave < 1250
igmz = np.ones_like(wave)
igmz[lylim] = IGM.full_IGM(z, wave[lylim] * (1 + z))
except:
igmz = 1.0
else:
igmz = 1.0
obsnames = ["bspl", "step", "poly"]
is_obsframe = np.array([t.split()[0] in obsnames for t in templates])
flux_arr[~is_obsframe, :] *= igmz
# Multiply spline?
for i, t in enumerate(templates):
if "spline" in t:
for j, tj in enumerate(templates):
if is_obsframe[j]:
ma = flux_arr[j, :].sum()
ma = ma if ma > 0 else 1
ma = 1
flux_arr[j, :] *= flux_arr[i, :] / ma
return wave, flux_arr, is_line
[docs]def compute_equivalent_widths(
templates, coeffs, covar, max_R=5000, Ndraw=1000, seed=0, z=0, observed_frame=False
):
"""
Compute template-fit emission line equivalent widths
Parameters
----------
templates : dictionary of `~grizli.utils.SpectrumTemplate` objects
Output template list with `NTEMP` templates.
coeffs : `~numpy.ndarray`, dimensions (`NTEMP`)
Fit coefficients
covar : `~numpy.ndarray`, dimensions (`NTEMP`, `NTEMP`)
Covariance matrix
max_R : float
Maximum spectral resolution of the regridded templates.
Ndraw : int
Number of random draws to extract from the covariance matrix
seed : positive int
Random number seed to produce repeatible results. If `None`, then
use default state.
z : float
Redshift where the fit is evaluated
observed_framme : bool
If true, then computed EWs are observed frame, otherwise they are
rest frame at redshift `z`.
Returns
-------
EWdict : dict
Dictionary of [16, 50, 84th] percentiles of the line EW distributions.
"""
# Array versions of the templates
wave, flux_arr, is_line = array_templates(templates, max_R=max_R, z=z)
keys = np.array(list(templates.keys()))
EWdict = OrderedDict()
for key in keys[is_line]:
EWdict[key] = (0.0, 0.0, 0.0)
# Only worry about templates with non-zero coefficients, which should
# be accounted for in the covariance array (with get_uncertainties=2)
clip = coeffs != 0
# No valid lines
if (is_line & clip).sum() == 0:
return EWdict
# Random draws from the covariance matrix
covar_clip = covar[clip, :][:, clip]
if seed is not None:
np.random.seed(seed)
draws = np.random.multivariate_normal(coeffs[clip], covar_clip, size=Ndraw)
# Evaluate the continuum fits from the draws
continuum = np.dot(draws * (~is_line[clip]), flux_arr[clip, :])
# Compute the emission line EWs
tidx = np.where(is_line[clip])[0]
for ix in tidx:
key = keys[clip][ix]
# Line template
line = np.dot(draws[:, ix][:, None], flux_arr[clip, :][ix, :][None, :])
# Where line template non-zero
mask = flux_arr[clip, :][ix, :] > 0
ew_i = np.trapz(
(line / continuum)[:, mask], wave[mask] * (1 + z * observed_frame), axis=1
)
EWdict[key] = np.percentile(ew_i, [16.0, 50.0, 84.0])
return EWdict
#####################
# Photometry from Vizier tables
# CFHTLS
CFHTLS_W_VIZIER = "II/317/cfhtls_w"
CFHTLS_W_BANDS = OrderedDict(
[
("cfht_mega_u", ["umag", "e_umag"]),
("cfht_mega_g", ["gmag", "e_gmag"]),
("cfht_mega_r", ["rmag", "e_rmag"]),
("cfht_mega_i", ["imag", "e_imag"]),
("cfht_mega_z", ["zmag", "e_zmag"]),
]
)
CFHTLS_D_VIZIER = "II/317/cfhtls_d"
CFHTLS_D_BANDS = OrderedDict(
[
("cfht_mega_u", ["umag", "e_umag"]),
("cfht_mega_g", ["gmag", "e_gmag"]),
("cfht_mega_r", ["rmag", "e_rmag"]),
("cfht_mega_i", ["imag", "e_imag"]),
("cfht_mega_z", ["zmag", "e_zmag"]),
]
)
# SDSS DR12
SDSS_DR12_VIZIER = "V/147/sdss12"
SDSS_DR12_BANDS = OrderedDict(
[
("SDSS/u", ["umag", "e_umag"]),
("SDSS/g", ["gmag", "e_gmag"]),
("SDSS/r", ["rmag", "e_rmag"]),
("SDSS/i", ["imag", "e_imag"]),
("SDSS/z", ["zmag", "e_zmag"]),
]
)
# PanStarrs
PS1_VIZIER = "II/349/ps1"
PS1_BANDS = OrderedDict(
[
("PS1.g", ["gKmag", "e_gKmag"]),
("PS1.r", ["rKmag", "e_rKmag"]),
("PS1.i", ["iKmag", "e_iKmag"]),
("PS1.z", ["zKmag", "e_zKmag"]),
("PS1.y", ["yKmag", "e_yKmag"]),
]
)
# KIDS DR3
KIDS_DR3_VIZIER = "II/347/kids_dr3"
KIDS_DR3_BANDS = OrderedDict(
[
("OCam.sdss.u", ["umag", "e_umag"]),
("OCam.sdss.g", ["gmag", "e_gmag"]),
("OCam.sdss.r", ["rmag", "e_rmag"]),
("OCam.sdss.i", ["imag", "e_imag"]),
]
)
# WISE all-sky
WISE_VIZIER = "II/328/allwise"
WISE_BANDS = OrderedDict(
[("WISE/RSR-W1", ["W1mag", "e_W1mag"]), ("WISE/RSR-W2", ["W2mag", "e_W2mag"])]
)
# ('WISE/RSR-W3', ['W3mag', 'e_W3mag']),
# ('WISE/RSR-W4', ['W4mag', 'e_W4mag'])])
# VIKING VISTA
VIKING_VIZIER = "II/343/viking2"
VIKING_BANDS = OrderedDict(
[
("SDSS/z", ["Zpmag", "e_Zpmag"]),
("VISTA/Y", ["Ypmag", "e_Ypmag"]),
("VISTA/J", ["Jpmag", "e_Jpmag"]),
("VISTA/H", ["Hpmag", "e_Hpmag"]),
("VISTA/Ks", ["Kspmag", "e_Kspmag"]),
]
)
# UKIDSS wide surveys
UKIDSS_LAS_VIZIER = "II/319/las9"
UKIDSS_LAS_BANDS = OrderedDict(
[
("WFCAM_Y", ["Ymag", "e_Ymag"]),
("WFCAM_J", ["Jmag1", "e_Jmag1"]),
("WFCAM_J", ["Jmag2", "e_Jmag2"]),
("WFCAM_H", ["Hmag", "e_Hmag"]),
("WFCAM_K", ["Kmag", "e_Kmag"]),
]
)
UKIDSS_DXS_VIZIER = "II/319/dxs9"
UKIDSS_DXS_BANDS = OrderedDict(
[("WFCAM_J", ["Jmag", "e_Jmag"]), ("WFCAM_K", ["Kmag", "e_Kmag"])]
)
# GALEX
GALEX_MIS_VIZIER = "II/312/mis"
GALEX_MIS_BANDS = OrderedDict([("FUV", ["FUV", "e_FUV"]), ("NUV", ["NUV", "e_NUV"])])
GALEX_AIS_VIZIER = "II/312/ais"
GALEX_AIS_BANDS = OrderedDict([("FUV", ["FUV", "e_FUV"]), ("NUV", ["NUV", "e_NUV"])])
# Combined Dict
VIZIER_BANDS = OrderedDict()
VIZIER_BANDS[CFHTLS_W_VIZIER] = CFHTLS_W_BANDS
VIZIER_BANDS[CFHTLS_D_VIZIER] = CFHTLS_D_BANDS
VIZIER_BANDS[SDSS_DR12_VIZIER] = SDSS_DR12_BANDS
VIZIER_BANDS[PS1_VIZIER] = PS1_BANDS
VIZIER_BANDS[KIDS_DR3_VIZIER] = KIDS_DR3_BANDS
VIZIER_BANDS[WISE_VIZIER] = WISE_BANDS
VIZIER_BANDS[VIKING_VIZIER] = VIKING_BANDS
VIZIER_BANDS[UKIDSS_LAS_VIZIER] = UKIDSS_LAS_BANDS
VIZIER_BANDS[UKIDSS_DXS_VIZIER] = UKIDSS_DXS_BANDS
VIZIER_BANDS[GALEX_MIS_VIZIER] = GALEX_MIS_BANDS
VIZIER_BANDS[GALEX_AIS_VIZIER] = GALEX_AIS_BANDS
VIZIER_VEGA = OrderedDict()
VIZIER_VEGA[CFHTLS_W_VIZIER] = False
VIZIER_VEGA[CFHTLS_D_VIZIER] = False
VIZIER_VEGA[SDSS_DR12_VIZIER] = False
VIZIER_VEGA[PS1_VIZIER] = False
VIZIER_VEGA[KIDS_DR3_VIZIER] = False
VIZIER_VEGA[WISE_VIZIER] = True
VIZIER_VEGA[VIKING_VIZIER] = True
VIZIER_VEGA[UKIDSS_LAS_VIZIER] = True
VIZIER_VEGA[UKIDSS_DXS_VIZIER] = True
VIZIER_VEGA[GALEX_MIS_VIZIER] = False
VIZIER_VEGA[GALEX_AIS_VIZIER] = False
[docs]def get_Vizier_photometry(
ra,
dec,
templates=None,
radius=2,
vizier_catalog=PS1_VIZIER,
bands=PS1_BANDS,
filter_file="/usr/local/share/eazy-photoz/filters/FILTER.RES.latest",
MW_EBV=0,
convert_vega=False,
raw_query=False,
verbose=True,
timeout=300,
rowlimit=50000,
):
"""
Fetch photometry from a Vizier catalog
Requires eazypy/eazy
Parameters
----------
ra : float
Right ascension of the target in degrees.
dec : float
Declination of the target in degrees.
templates : dict, optional
Dictionary of templates to be used for photometric redshift fitting.
radius : float, optional
Search radius around the target position in arcseconds.
vizier_catalog : str or list, optional
Name of the Vizier catalog(s) to query or a list of catalog names.
bands : dict, optional
Dictionary of band names and corresponding column names in
the Vizier catalog.
filter_file : str, optional
Path to the filter file.
MW_EBV : float, optional
Milky Way E(B-V) reddening value.
convert_vega : bool, optional
Flag indicating whether to convert the photometry
from Vega to AB magnitude system.
raw_query : bool, optional
Flag indicating whether to return the raw query result.
verbose : bool, optional
Flag indicating whether to print verbose output.
timeout : int, optional
Timeout value for the Vizier query in seconds.
rowlimit : int, optional
Maximum number of rows to retrieve from the Vizier catalog.
Returns
-------
phot : OrderedDict
Dictionary containing the retrieved photometry and related information.
"""
from collections import OrderedDict
import astropy.units as u
from astroquery.vizier import Vizier
Vizier.ROW_LIMIT = rowlimit
Vizier.TIMEOUT = timeout
# print('xxx', Vizier.ROW_LIMIT, Vizier.TIMEOUT)
import astropy.coordinates as coord
import astropy.units as u
# import pysynphot as S
from eazy.templates import Template
from eazy.filters import FilterFile
from eazy.photoz import TemplateGrid
from eazy.filters import FilterDefinition
res = FilterFile(filter_file)
coo = coord.SkyCoord(ra=ra, dec=dec, unit=(u.deg, u.deg), frame="icrs")
columns = ["*"]
# columns = []
if isinstance(vizier_catalog, list):
for c in [VIKING_VIZIER]:
for b in VIZIER_BANDS[c]:
columns += VIZIER_BANDS[c][b]
columns = list(np.unique(columns))
# print("xxx columns", columns)
else:
for b in bands:
columns += bands[b]
if isinstance(vizier_catalog, list):
v = Vizier(catalog=VIKING_VIZIER, columns=["+_r"] + columns)
else:
v = Vizier(catalog=vizier_catalog, columns=["+_r"] + columns)
v.ROW_LIMIT = rowlimit
v.TIMEOUT = timeout
# query_catalog = vizier_catalog
try:
tabs = v.query_region(
coo, radius="{0}s".format(radius), catalog=vizier_catalog
) # [0]
if raw_query:
return tabs
tab = tabs[0]
if False:
for t in tabs:
bands = VIZIER_BANDS[t.meta["name"]]
for b in bands:
for c in bands[b]:
print(t.meta["name"], c, c in t.colnames) # c = bands[b][0]
ix = np.argmin(tab["_r"])
tab = tab[ix]
except:
tab = None
return None
viz_tables = ", ".join([t.meta["name"] for t in tabs])
if verbose:
print("Photometry from vizier catalogs: {0}".format(viz_tables))
pivot = [] # OrderedDict()
flam = []
eflam = []
filters = []
for tab in tabs:
# Downweight PS1 if have SDSS ? For now, do nothing
if (tab.meta["name"] == PS1_VIZIER) & (SDSS_DR12_VIZIER in viz_tables):
# continue
err_scale = 1
else:
err_scale = 1
# Only use one CFHT catalog
if (tab.meta["name"] == CFHTLS_W_VIZIER) & (CFHTLS_D_VIZIER in viz_tables):
continue
if tab.meta["name"] == UKIDSS_LAS_VIZIER:
flux_scale = 1.33
else:
flux_scale = 1.0
convert_vega = VIZIER_VEGA[tab.meta["name"]]
bands = VIZIER_BANDS[tab.meta["name"]]
# if verbose:
# print(tab.colnames)
# filters += [res.filters[res.search(b, verbose=False)[0]] for b in bands]
to_flam = 10 ** (-0.4 * (48.6)) * 3.0e18 # / pivot(Ang)**2
for ib, b in enumerate(bands):
filt = res.filters[res.search(b, verbose=False)[0]]
filters.append(filt)
if convert_vega:
to_ab = filt.ABVega()
else:
to_ab = 0.0
fcol, ecol = bands[b]
pivot.append(filt.pivot())
flam.append(
10 ** (-0.4 * (tab[fcol][0] + to_ab)) * to_flam / pivot[-1] ** 2
)
flam[-1] *= flux_scale
eflam.append(tab[ecol][0] * np.log(10) / 2.5 * flam[-1] * err_scale)
for i in range(len(filters))[::-1]:
if np.isscalar(flam[i]) & np.isscalar(eflam[i]):
continue
else:
flam.pop(i)
eflam.pop(i)
filters.pop(i)
pivot.pop(i)
lc = np.array(pivot) # [pivot[ib] for ib in range(len(bands))]
if templates is not None:
eazy_templates = [
Template(arrays=(templates[k].wave, templates[k].flux), name=k)
for k in templates
]
zgrid = log_zgrid(zr=[0.01, 3.4], dz=0.005)
tempfilt = TemplateGrid(
zgrid,
eazy_templates,
filters=filters,
add_igm=True,
galactic_ebv=MW_EBV,
Eb=0,
n_proc=0,
verbose=False,
)
else:
tempfilt = None
phot = OrderedDict(
[
("flam", np.array(flam)),
("eflam", np.array(eflam)),
("filters", filters),
("tempfilt", tempfilt),
("lc", np.array(lc)),
("source", "Vizier " + viz_tables),
]
)
return phot
[docs]def generate_tempfilt(templates, filters, zgrid=None, MW_EBV=0):
"""
Generate a template grid for photometric redshift fitting.
Parameters
----------
templates : dict
Dictionary of templates. Each template should be
an instance of `eazy.templates.Template`.
filters : list
List of filters to be used for the photometric redshift fitting.
zgrid : array-like, optional
Redshift grid. If not provided, a default grid will be used.
MW_EBV : float, optional
Milky Way E(B-V) reddening value. Default is 0.
Returns
-------
tempfilt : `eazy.photoz.TemplateGrid`
Template grid for photometric redshift fitting.
"""
from eazy.templates import Template
from eazy.photoz import TemplateGrid
eazy_templates = [
Template(arrays=(templates[k].wave, templates[k].flux), name=k)
for k in templates
]
if zgrid is None:
zgrid = log_zgrid(zr=[0.01, 3.4], dz=0.005)
tempfilt = TemplateGrid(
zgrid,
eazy_templates,
filters=filters,
add_igm=True,
galactic_ebv=MW_EBV,
Eb=0,
n_proc=0,
verbose=False,
)
return tempfilt
[docs]def combine_phot_dict(phots, templates=None, MW_EBV=0):
"""
Combine photmetry dictionaries
Parameters
----------
phots : list
List of photometry dictionaries to combine.
templates : list, optional
List of templates to use for generating `tempfilt`.
MW_EBV : float, optional
Milky Way E(B-V) reddening value.
Returns
-------
dict
Combined photometry dictionary.
"""
phot = {}
phot["flam"] = []
phot["eflam"] = []
phot["filters"] = []
for p in phots:
phot["flam"] = np.append(phot["flam"], p["flam"])
phot["eflam"] = np.append(phot["eflam"], p["eflam"])
phot["filters"].extend(p["filters"])
if templates is not None:
phot["tempfilt"] = generate_tempfilt(templates, phot["filters"], MW_EBV=MW_EBV)
return phot
[docs]def get_spectrum_AB_mags(spectrum, bandpasses=[]):
"""
Integrate a `~pysynphot` spectrum through filter bandpasses
Parameters
----------
spectrum : type
bandpasses : list
List of `pysynphot` bandpass objects, e.g.,
>>> import pysynphot as S
>>> bandpasses = [S.ObsBandpass('wfc3,ir,f140w')]
Returns
-------
ab_mags : dict
Dictionary with keys from `bandpasses` and the integrated magnitudes
"""
import pysynphot as S
flat = S.FlatSpectrum(0, fluxunits="ABMag")
ab_mags = OrderedDict()
for bp in bandpasses:
flat_obs = S.Observation(flat, bp)
spec_obs = S.Observation(spectrum, bp)
ab_mags[bp.name] = -2.5 * np.log10(spec_obs.countrate() / flat_obs.countrate())
return ab_mags
[docs]def log_zgrid(zr=[0.7, 3.4], dz=0.01):
"""
Make a logarithmically spaced redshift grid
Parameters
----------
zr : [float, float]
Minimum and maximum of the desired grid
dz : float
Step size, dz/(1+z)
Returns
-------
zgrid : array-like
Redshift grid
"""
zgrid = np.exp(np.arange(np.log(1 + zr[0]), np.log(1 + zr[1]), dz)) - 1
return zgrid
[docs]def trapz_dx(x):
"""
Return trapezoid rule coefficients, useful for numerical integration
using a dot product
Parameters
----------
x : array-like
Independent variable
Returns
-------
dx : array_like
Coefficients for trapezoidal rule integration.
"""
dx = np.zeros_like(x)
diff = np.diff(x) / 2.0
dx[:-1] += diff
dx[1:] += diff
return dx
[docs]def get_wcs_pscale(wcs, set_attribute=True):
"""
Get correct pscale from a `~astropy.wcs.WCS` object
Parameters
----------
wcs : `~astropy.wcs.WCS` or `~astropy.io.fits.Header`
set_attribute : bool
Set the `pscale` attribute on `wcs`, along with returning the value.
Returns
-------
pscale : float
Pixel scale from `wcs.cd`
"""
from numpy import linalg
if isinstance(wcs, pyfits.Header):
wcs = pywcs.WCS(wcs, relax=True)
if hasattr(wcs.wcs, "cd"):
det = linalg.det(wcs.wcs.cd)
else:
det = linalg.det(wcs.wcs.pc)
pscale = np.sqrt(np.abs(det)) * 3600.0
with warnings.catch_warnings():
warnings.filterwarnings(
"ignore", "cdelt will be ignored since cd is present", RuntimeWarning
)
if hasattr(wcs.wcs, "cdelt"):
pscale *= wcs.wcs.cdelt[0]
wcs.pscale = pscale
return pscale
[docs]def sip_rot90(input, rot, reverse=False, verbose=False, compare=False):
"""
Rotate a SIP WCS by increments of 90 degrees using direct transformations
between x / y coordinates
Parameters
----------
input : `~astropy.io.fits.Header` or `~astropy.wcs.WCS`
Header or WCS
rot : int
Number of times to rotate the WCS 90 degrees *clockwise*, analogous
to `numpy.rot90`
reverse : bool, optional
If `input` is a header and includes a keyword ``ROT90``, then undo
the rotation and remove the keyword from the output header
verbose : bool, optional
If True, print the root-mean-square difference between the original
and rotated coordinates
compare : bool, optional
If True, plot the difference between the original and rotated
coordinates as a function of x and y
Returns
-------
header : `~astropy.io.fits.Header`
Rotated WCS header
wcs : `~astropy.wcs.WCS`
Rotated WCS
desc : str
Description of the transform associated with ``rot``, e.g,
``x=nx-x, y=ny-y`` for ``rot=±2``.
"""
import copy
import astropy.io.fits
import astropy.wcs
import matplotlib.pyplot as plt
if isinstance(input, astropy.io.fits.Header):
orig = copy.deepcopy(input)
new = copy.deepcopy(input)
if "ROT90" in input:
if reverse:
rot = -orig["ROT90"]
new.remove("ROT90")
else:
new["ROT90"] = orig["ROT90"] + rot
else:
new["ROT90"] = rot
else:
orig = to_header(input)
new = to_header(input)
orig_wcs = pywcs.WCS(orig, relax=True)
### CD = [[dra/dx, dra/dy], [dde/dx, dde/dy]]
### x = a_i_j * u**i * v**j
### y = b_i_j * u**i * v**j
ix = 1
if compare:
xarr = np.arange(0, 2048, 64)
xp, yp = np.meshgrid(xarr, xarr)
rd = orig_wcs.all_pix2world(xp, yp, ix)
if rot % 4 == 1:
# CW 90 deg : x = y, y = (nx - x), u=v, v=-u
desc = "x=y, y=nx-x"
new["CRPIX1"] = orig["CRPIX2"]
new["CRPIX2"] = orig["NAXIS1"] - orig["CRPIX1"] + 1
new["CD1_1"] = orig["CD1_2"]
new["CD1_2"] = -orig["CD1_1"]
new["CD2_1"] = orig["CD2_2"]
new["CD2_2"] = -orig["CD2_1"]
for i in range(new["A_ORDER"] + 1):
for j in range(new["B_ORDER"] + 1):
Aij = f"A_{i}_{j}"
if Aij not in new:
continue
new[f"A_{i}_{j}"] = orig[f"B_{j}_{i}"] * (-1) ** j
new[f"B_{i}_{j}"] = orig[f"A_{j}_{i}"] * (-1) ** j * -1
new_wcs = astropy.wcs.WCS(new, relax=True)
if compare:
xr, yr = new_wcs.all_world2pix(*rd, ix)
xo = yp
yo = orig["NAXIS1"] - xp
elif rot % 4 == 3:
# CW 270 deg : y = x, x = (ny - u), u=-v, v=u
desc = "x=ny-y, y=x"
new["CRPIX1"] = orig["NAXIS2"] - orig["CRPIX2"] + 1
new["CRPIX2"] = orig["CRPIX1"]
new["CD1_1"] = -orig["CD1_2"]
new["CD1_2"] = orig["CD1_1"]
new["CD2_1"] = -orig["CD2_2"]
new["CD2_2"] = orig["CD2_1"]
for i in range(new["A_ORDER"] + 1):
for j in range(new["B_ORDER"] + 1):
Aij = f"A_{i}_{j}"
if Aij not in new:
continue
new[f"A_{i}_{j}"] = orig[f"B_{j}_{i}"] * (-1) ** i * -1
new[f"B_{i}_{j}"] = orig[f"A_{j}_{i}"] * (-1) ** i
new_wcs = astropy.wcs.WCS(new, relax=True)
if compare:
xr, yr = new_wcs.all_world2pix(*rd, ix)
xo = orig["NAXIS2"] - yp
yo = xp
elif rot % 4 == 2:
# CW 180 deg : x=nx-x, y=ny-y, u=-u, v=-v
desc = "x=nx-x, y=ny-y"
new["CRPIX1"] = orig["NAXIS1"] - orig["CRPIX1"] + 1
new["CRPIX2"] = orig["NAXIS2"] - orig["CRPIX2"] + 1
new["CD1_1"] = -orig["CD1_1"]
new["CD1_2"] = -orig["CD1_2"]
new["CD2_1"] = -orig["CD2_1"]
new["CD2_2"] = -orig["CD2_2"]
for i in range(new["A_ORDER"] + 1):
for j in range(new["B_ORDER"] + 1):
Aij = f"A_{i}_{j}"
if Aij not in new:
continue
new[f"A_{i}_{j}"] = orig[f"A_{i}_{j}"] * (-1) ** j * (-1) ** i * -1
new[f"B_{i}_{j}"] = orig[f"B_{i}_{j}"] * (-1) ** j * (-1) ** i * -1
new_wcs = astropy.wcs.WCS(new, relax=True)
if compare:
xr, yr = new_wcs.all_world2pix(*rd, ix)
xo = orig["NAXIS1"] - xp
yo = orig["NAXIS2"] - yp
else:
# rot=0, do nothing
desc = "x=x, y=y"
new_wcs = orig_wcs
if compare:
xo = xp
yo = yp
xr, yr = new_wcs.all_world2pix(*rd, ix)
if verbose:
if compare:
xrms = nmad(xr - xo)
yrms = nmad(yr - yo)
print(f"Rot90: {rot} rms={xrms:.2e} {yrms:.2e}")
if compare:
fig, axes = plt.subplots(1, 2, figsize=(10, 5), sharex=True, sharey=True)
axes[0].scatter(xp, xr - xo)
axes[0].set_xlabel("dx")
axes[1].scatter(yp, yr - yo)
axes[1].set_xlabel("dy")
for ax in axes:
ax.grid()
fig.tight_layout(pad=0.5)
return new, new_wcs, desc
[docs]def get_fits_slices(file1, file2):
"""
Get overlapping slices of FITS files
Parameters
----------
file1 : str, `~astropy.io.fits.Header`, or `~astropy.io.fits.HDUList`
First file, header or HDU
file2 : str, `~astropy.io.fits.Header`, or `~astropy.io.fits.HDUList`
Second file, header or HDU
Returns
-------
nx, ny : int
Size of the overlapping region
valid : bool
True if there is some overlap
sl1 : (slice, slice)
y and x slices of the overlap in ``file1``
sl2 : (slice, slice)
y and x slices of the overlap in ``file2``
"""
# First image
if isinstance(file1, pyfits.HDUList):
h1 = file1[0].header
elif isinstance(file1, str):
im1 = pyfits.open(file1)
h1 = im1[0].header
else:
h1 = file1
# Second image
if isinstance(file2, pyfits.HDUList):
h2 = file2[0].header
elif isinstance(file2, str):
im2 = pyfits.open(file2)
h2 = im2[0].header
else:
h2 = file2
# origin and shape tuples
o1 = np.array([-h1["CRPIX2"], -h1["CRPIX1"]]).astype(int)
sh1 = np.array([h1["NAXIS2"], h1["NAXIS1"]])
o2 = np.array([-h2["CRPIX2"], -h2["CRPIX1"]]).astype(int)
sh2 = np.array([h2["NAXIS2"], h2["NAXIS1"]])
# slices
sl1, sl2 = get_common_slices(o1, sh1, o2, sh2)
nx = sl1[1].stop - sl1[1].start
ny = sl1[0].stop - sl1[0].start
valid = (nx > 0) & (ny > 0)
return nx, ny, valid, sl1, sl2
[docs]def get_common_slices(a_origin, a_shape, b_origin, b_shape):
"""
Get slices of overlaps between two rectangular grids
Parameters
----------
a_origin : tuple
The origin coordinates of grid A.
a_shape : tuple
The shape of grid A.
b_origin : tuple
The origin coordinates of grid B.
b_shape : tuple
The shape of grid B.
Returns
-------
a_slice : tuple
The slices of grid A that overlap with grid B.
b_slice : tuple
The slices of grid B that overlap with grid A.
"""
ll = np.min([a_origin, b_origin], axis=0)
ur = np.max([a_origin + a_shape, b_origin + b_shape], axis=0)
# other in self
lls = np.minimum(b_origin - ll, a_shape)
urs = np.clip(b_origin + b_shape - a_origin, [0, 0], a_shape)
# self in other
llo = np.minimum(a_origin - ll, b_shape)
uro = np.clip(a_origin + a_shape - b_origin, [0, 0], b_shape)
a_slice = (slice(lls[0], urs[0]), slice(lls[1], urs[1]))
b_slice = (slice(llo[0], uro[0]), slice(llo[1], uro[1]))
return a_slice, b_slice
[docs]def reproject_faster(input_hdu, output, pad=10, **kwargs):
"""
Speed up `reproject` module with array slices of the input image
Parameters
----------
input_hdu : `~astropy.io.fits.ImageHDU`
Input image HDU to reproject.
output : `~astropy.wcs.WCS` or `~astropy.io.fits.Header`
Output frame definition.
pad : int
Pixel padding on slices cut from the `input_hdu`.
kwargs : dict
Arguments passed through to `~reproject.reproject_interp`. For
example, `order='nearest-neighbor'`.
Returns
-------
reprojected : `~numpy.ndarray`
Reprojected data from `input_hdu`.
footprint : `~numpy.ndarray`
Footprint of the input array in the output frame.
Notes
-----
`reproject' is an astropy-compatible module that can be installed with
`pip`. See https://reproject.readthedocs.io.
"""
import reproject
# Output WCS
if isinstance(output, pywcs.WCS):
out_wcs = output
else:
out_wcs = pywcs.WCS(output, relax=True)
if "SIP" in out_wcs.wcs.ctype[0]:
print("Warning: `reproject` doesn't appear to support SIP projection")
# Compute pixel coordinates of the output frame corners in the input image
input_wcs = pywcs.WCS(input_hdu.header, relax=True)
out_fp = out_wcs.calc_footprint()
input_xy = input_wcs.all_world2pix(out_fp, 0)
slx = slice(int(input_xy[:, 0].min()) - pad, int(input_xy[:, 0].max()) + pad)
sly = slice(int(input_xy[:, 1].min()) - pad, int(input_xy[:, 1].max()) + pad)
# Make the cutout
sub_data = input_hdu.data[sly, slx]
sub_header = get_wcs_slice_header(input_wcs, slx, sly)
sub_hdu = pyfits.PrimaryHDU(data=sub_data, header=sub_header)
# Get the reprojection
seg_i, fp_i = reproject.reproject_interp(sub_hdu, output, **kwargs)
return seg_i.astype(sub_data.dtype), fp_i.astype(np.uint8)
DRIZZLE_KEYS = [
"GEOM",
"DATA",
"DEXP",
"OUDA",
"OUWE",
"OUCO",
"MASK",
"WTSC",
"KERN",
"PIXF",
"COEF",
"OUUN",
"FVAL",
"WKEY",
"SCAL",
"ISCL",
]
[docs]def make_maximal_wcs(
files,
pixel_scale=None,
get_hdu=True,
pad=90,
verbose=True,
theta=0,
poly_buffer=1.0 / 3600,
nsci_extensions=4,
):
"""
Compute an ImageHDU with a footprint that covers all of ``files``
Parameters
----------
files : list
List of HST FITS files (e.g., FLT.) or WCS objects.
pixel_scale : float, optional
Pixel scale of output WCS, in `~astropy.units.arcsec`. If `None`,
get pixel scale of first file in `files`.
get_hdu : bool, optional
If True, return an `~astropy.io.fits.ImageHDU` object. If False, return
a tuple of `~astropy.io.fits.Header` and `~astropy.wcs.WCS`.
pad : float, optional
Padding to add to the total image size, in `~astropy.units.arcsec`.
theta : float, optional
Position angle, in degrees.
poly_buffer : float, optional
Buffer size to apply to the footprint polygon, in degrees.
nsci_extensions : int, optional
Number of 'SCI' extensions to try in the exposure files.
Returns
-------
hdu : `~astropy.io.fits.ImageHDU`
If `get_hdu` is True.
-or-
header, wcs : `~astropy.io.fits.Header`, `~astropy.wcs.WCS`
If `get_hdu` is False.
"""
import numpy as np
from shapely.geometry import Polygon
import astropy.io.fits as pyfits
import astropy.wcs as pywcs
group_poly = None
if hasattr(files, "buffer"):
# Input is a shapely object
group_poly = files
elif isinstance(files[0], pywcs.WCS):
# Already wcs_list
wcs_list = [(wcs, "WCS", -1) for wcs in files]
else:
wcs_list = []
for i, file in enumerate(files):
if not os.path.exists(file):
continue
with pyfits.open(file) as im:
for ext in range(nsci_extensions):
if ("SCI", ext + 1) not in im:
continue
wcs = pywcs.WCS(im["SCI", ext + 1].header, fobj=im)
wcs_list.append((wcs, file, ext))
if pixel_scale is None:
pixel_scale = get_wcs_pscale(wcs_list[0][0])
if group_poly is None:
for i, (wcs, file, chip) in enumerate(wcs_list):
p_i = Polygon(wcs.calc_footprint())
if group_poly is None:
if poly_buffer > 0:
group_poly = p_i.buffer(1.0 / 3600)
else:
group_poly = p_i
else:
if poly_buffer > 0:
group_poly = group_poly.union(p_i.buffer(1.0 / 3600))
else:
group_poly = group_poly.union(p_i)
x0, y0 = np.asarray(group_poly.centroid.xy,dtype=float)[:, 0]
if verbose:
msg = "{0:>3d}/{1:>3d}: {2}[SCI,{3}] {4:>6.2f}"
print(
msg.format(
i,
len(files),
file,
chip + 1,
group_poly.area * 3600 * np.cos(y0 / 180 * np.pi),
)
)
px = np.asarray(group_poly.convex_hull.boundary.xy,dtype=float).T
#x0, y0 = np.asarray(group_poly.centroid.xy,dtype=float)[:,0]
x0 = (px.max(axis=0) + px.min(axis=0)) / 2.0
cosd = np.array([np.cos(x0[1] / 180 * np.pi), 1])
_mat = np.array([[np.cos(theta), -np.sin(theta)], [np.sin(theta), np.cos(theta)]])
# Rotated
pr = ((px - x0) * cosd).dot(_mat) / cosd + x0
size_arcsec = (pr.max(axis=0) - pr.min(axis=0)) * cosd * 3600
sx, sy = size_arcsec
# sx = (px.max()-px.min())*cosd*3600 # arcsec
# sy = (py.max()-py.min())*3600 # arcsec
size = np.maximum(sx + pad, sy + pad)
if verbose:
msg = "\n Mosaic WCS: ({0:.5f},{1:.5f}) "
msg += "{2:.1f}'x{3:.1f}' {4:.3f}\"/pix\n"
print(
msg.format(x0[0], x0[1], (sx + pad) / 60.0, (sy + pad) / 60.0, pixel_scale)
)
out = make_wcsheader(
ra=x0[0],
dec=x0[1],
size=(sx + pad * 2, sy + pad * 2),
pixscale=pixel_scale,
get_hdu=get_hdu,
theta=theta / np.pi * 180,
)
return out
[docs]def half_pixel_scale(wcs):
"""
Create a new WCS with half the pixel scale of another that can be
block-averaged 2x2
Parameters
----------
wcs : `~astropy.wcs.WCS`
Input WCS
Returns
-------
half_wcs : `~astropy.wcs.WCS`
New WCS with smaller pixels
"""
h = to_header(wcs)
for k in ["NAXIS1", "NAXIS2"]: # , 'CRPIX1', 'CRPIX2']:
h[k] *= 2
for k in ["CRPIX1", "CRPIX2"]:
h[k] = h[k] * 2 - 0.5
for k in ["CD1_1", "CD1_2", "CD2_1", "CD2_2"]:
if k in h:
h[k] /= 2
if 0:
# Test
new = pywcs.WCS(h)
sh = new.pixel_shape
wcorner = wcs.all_world2pix(
new.all_pix2world([[-0.5, -0.5], [sh[0] - 0.5, sh[1] - 0.5]], 0), 0
)
print("small > large")
print(", ".join([f"{w:.2f}" for w in wcorner[0]]))
print(", ".join([f"{w:.2f}" for w in wcorner[1]]), wcs.pixel_shape)
sh = wcs.pixel_shape
wcorner = new.all_world2pix(
wcs.all_pix2world([[-0.5, -0.5], [sh[0] - 0.5, sh[1] - 0.5]], 0), 0
)
print("large > small")
print(", ".join([f"{w:.2f}" for w in wcorner[0]]))
print(", ".join([f"{w:.2f}" for w in wcorner[1]]), new.pixel_shape)
new_wcs = pywcs.WCS(h, relax=True)
return new_wcs
[docs]def parse_s3_url(url="s3://bucket/path/to/file.txt"):
"""
Parse s3 path string
Parameters
----------
url : str
Full S3 path, e.g., ``[s3://]{bucket_name}/{s3_object}``
Returns
-------
bucket_name : str
Bucket name
s3_object : str
Full path of the S3 file object
filename : str
File name of the object, e.g. ``os.path.basename(s3_object)``
"""
surl = url.strip("s3://")
spl = surl.split("/")
if len(spl) < 2:
print(f"bucket / path not found in {url}")
return None, None, None
bucket_name = spl[0]
s3_object = "/".join(spl[1:])
filename = os.path.basename(s3_object)
return bucket_name, s3_object, filename
[docs]def fetch_s3_url(
url="s3://bucket/path/to/file.txt",
file_func=lambda x: os.path.join("./", x),
skip_existing=True,
verbose=True,
):
"""
Fetch file from an S3 bucket
Parameters
----------
url : str
S3 url of a file to download
file_func : function
Function applied to the file name extracted from `url`, e.g., to
set output directory, rename files, set a prefix, etc.
skip_existing : bool, optional
If True, skip downloading if the local file already exists.
Default is True.
verbose : bool, optional
If True, print download progress and status messages.
Default is True.
Returns
-------
local_file : str
Name of local file or `None` if failed to parse `url`
status : int
Bit flag of results: **1** == file found, **2** = download successful
"""
import traceback
import boto3
import botocore.exceptions
s3 = boto3.resource("s3")
bucket_name, s3_object, filename = parse_s3_url(url=url)
if bucket_name is None:
return url, os.path.exists(url)
bkt = s3.Bucket(bucket_name)
local_file = file_func(filename)
status = os.path.exists(local_file) * 1
if (status > 0) & skip_existing:
print(f"{local_file} exists, skipping.")
else:
try:
bkt.download_file(
s3_object, local_file, ExtraArgs={"RequestPayer": "requester"}
)
status += 2
if verbose:
print(f"{url} > {local_file}")
except botocore.exceptions.ClientError:
trace = traceback.format_exc(limit=2)
msg = trace.split("\n")[-2].split("ClientError: ")[1]
if verbose:
print(f"Failed {url}: {msg}")
# Download failed due to a ClientError
# Forbidden probably means insufficient bucket access privileges
pass
return local_file, status
[docs]def niriss_ghost_mask(
im,
init_thresh=0.05,
init_sigma=3,
final_thresh=0.01,
final_sigma=3,
erosions=0,
dilations=9,
verbose=True,
**kwargs,
):
"""
Make a mask for NIRISS imaging ghosts
See also Martel. JWST-STScI-004877 and
https://github.com/spacetelescope/niriss_ghost
Parameters
----------
im : `~astropy.io.fits.HDUList`
Input image HDUList.
init_thresh : float, optional
Initial threshold for detecting ghost pixels. Default is 0.05.
init_sigma : float, optional
Initial sigma threshold for detecting ghost pixels. Default is 3.
final_thresh : float, optional
Final threshold for detecting ghost pixels. Default is 0.01.
final_sigma : float, optional
Final sigma threshold for detecting ghost pixels. Default is 3.
erosions : int, optional
Number of binary erosions to apply to the ghost mask. Default is 0.
dilations : int, optional
Number of binary dilations to apply to the ghost mask. Default is 9.
verbose : bool, optional
If True, print diagnostic messages. Default is True.
Returns
-------
ghost_mask : `~numpy.ndarray`
Boolean array indicating the positions of the ghost pixels.
"""
import scipy.ndimage as nd
if im[0].header["PUPIL"] not in ["F115W", "F150W", "F200W"]:
return False
if im[0].header["PUPIL"] == "F115W":
xgap, ygap = 1156, 927
elif im[0].header["PUPIL"] == "F115W":
xgap, ygap = 1162, 938
else:
xgap, ygap = 1156, 944 - 2
yp, xp = np.indices((2048, 2048))
yg = 2 * (ygap - 1) - yp
xg = 2 * (xgap - 1) - xp
dx = xp - xgap
dy = yp - ygap
in_img = (xg >= 0) & (xg < 2048)
in_img &= (yg >= 0) & (yg < 2048)
in_img &= np.abs(dx) < 400
in_img &= np.abs(dy) < 400
if "MDRIZSKY" in im["SCI"].header:
bkg = im["SCI"].header["MDRIZSKY"]
else:
bkg = np.nanmedian(im["SCI"].data[im["DQ"].data == 0])
thresh = (im["SCI"].data - bkg) * init_thresh > init_sigma * im["ERR"].data
thresh &= in_img
_reflected = np.zeros_like(im["SCI"].data)
for xpi, ypi, xgi, ygi in zip(xp[thresh], yp[thresh], xg[thresh], yg[thresh]):
_reflected[ygi, xgi] = im["SCI"].data[ypi, xpi] - bkg
ghost_mask = _reflected * final_thresh > final_sigma * im["ERR"].data
if erosions > 0:
ghost_mask = nd.binary_erosion(ghost_mask, iterations=erosions)
ghost_mask = nd.binary_dilation(ghost_mask, iterations=dilations)
im[0].header["GHOSTMSK"] = True, "NIRISS ghost mask applied"
im[0].header["GHOSTNPX"] = ghost_mask.sum(), "Pixels in NIRISS ghost mask"
msg = "NIRISS ghost mask {0} Npix: {1}\n".format(
im[0].header["PUPIL"], ghost_mask.sum()
)
log_comment(LOGFILE, msg, verbose=verbose)
return ghost_mask
[docs]def get_photom_scale(header, verbose=True):
"""
Get tabulated scale factor
Parameters
----------
header : `~astropy.io.fits.Header`
Image header
verbose : bool, optional
Whether to display verbose output. Default is True.
Returns
-------
key : str
Detector + filter key
scale : float
Scale value. If `key` not found in the ``data/photom_correction.yml``
table or if the CTX is newer than that indicated in the correction
table then return 1.0.
"""
import yaml
if "TELESCOP" in header:
if header["TELESCOP"] not in ["JWST"]:
msg = f"get_photom_scale: TELESCOP={header['TELESCOP']} is not 'JWST'"
log_comment(LOGFILE, msg, verbose=verbose)
return header["TELESCOP"], 1.0
else:
return None, 1.0
corr_file = os.path.join(os.path.dirname(__file__), "data/photom_correction.yml")
if not os.path.exists(corr_file):
msg = f"{corr_file} not found."
log_comment(LOGFILE, msg, verbose=verbose)
return None, 1
with open(corr_file) as fp:
corr = yaml.load(fp, Loader=yaml.SafeLoader)
if "CRDS_CTX" in header:
if header["CRDS_CTX"] > corr["CRDS_CTX_MAX"]:
msg = f"get_photom_scale {corr_file}: {header['CRDS_CTX']} > {corr['CRDS_CTX_MAX']}"
log_comment(LOGFILE, msg, verbose=verbose)
return header["CRDS_CTX"], 1.0
key = "{0}-{1}".format(header["DETECTOR"], header["FILTER"])
if "PUPIL" in header:
key += "-{0}".format(header["PUPIL"])
if key not in corr:
msg = f"get_photom_scale {corr_file}: {key} not found"
log_comment(LOGFILE, msg, verbose=verbose)
return key, 1.0
else:
msg = f"get_photom_scale {corr_file}: Scale {key} by {1./corr[key]:.3f}"
log_comment(LOGFILE, msg, verbose=verbose)
return key, 1.0 / corr[key]
[docs]def jwst_crds_photom_scale(hdul, context="jwst_1130.pmap", update=True, verbose=False):
"""
Scale factors between different JWST CRDS_CONTEXT
Parameters
----------
hdul : `astropy.io.fits.HDUList`
Exposure file HDUList, which has header keywords like CRDS_CTX, PHOTMJSR, etc.
context : str
Target CRDS context version
update : bool
Scale photometry header keywords by the ratio of NEW_PHOTMJSR / OLD_PHOTMJSR
verbose: bool
Messaging
Returns
-------
scale : float
Relative photometry scaling factor NEW_PHOTMJSR / OLD_PHOTMJSR. Defaults to
1.0 if not a JWST instrument or if certain necessary header keywords not found
"""
try:
from .jwst_utils import get_crds_zeropoint, get_jwst_filter_info
except ImportError:
print(
"jwst_crds_photom_scale: failed to import grizli.jwst_utils.get_crds_zeropoint"
)
return 1.0
if "INSTRUME" not in hdul[0].header:
return 1.0
instrument = hdul[0].header["INSTRUME"]
if instrument not in ["NIRCAM", "MIRI", "NIRISS"]:
return 1.0
mode = {"context": context, "verbose": False, "instrument": instrument}
for k in ["FILTER", "PUPIL", "DETECTOR"]:
if k in hdul[0].header:
mode[k.lower()] = hdul[0].header[k]
ref_ctx, r_photom, ref_photmjsr, ref_pixar_sr = get_crds_zeropoint(**mode)
if ref_photmjsr is None:
return 1.0
if "PHOTMJSR" not in hdul["SCI"].header:
return 1.0
old_photmjsr = hdul["SCI"].header["PHOTMJSR"]
scale = ref_photmjsr / old_photmjsr
msg = f"jwst_crds_photom_scale: {context} photmjsr old, new = "
msg += f"{old_photmjsr:.3f} {ref_photmjsr:.3f} scale = {scale:.3f}"
if update:
# Check image units
if "OBUNIT" in hdul["SCI"].header:
unit_key = "OBUNIT"
else:
unit_key = "BUNIT"
if hdul["SCI"].header[unit_key].upper() == "MJy/sr".upper():
# Image was already scaled (cal), so use scale factor
to_mjysr = scale
else:
# Image wasn't scaled, so just use mjsr
to_mjysr = ref_photmjsr
# Recalculate PHOTFNU, PHOTFLAM from PIXAR_SR, which could also change
if ref_pixar_sr is None:
ref_pixar_sr = hdul["SCI"].header["PIXAR_SR"]
photfnu = to_mjysr * ref_pixar_sr * 1.0e6
filter_info = get_jwst_filter_info(hdul[0].header)
if filter_info is not None:
plam = filter_info["pivot"] * 1.0e4
else:
plam = 5.0e4
photflam = photfnu * 2.99e-5 / plam ** 2
for e in [0, "SCI"]:
hdul[e].header["PIXAR_SR"] = ref_pixar_sr
hdul[e].header["PHOTFLAM"] = photflam
hdul[e].header["PHOTFNU"] = photfnu
for k in ["TO_MJYSR", "PHOTMJSR"]:
if k in hdul[e].header:
hdul[e].header[k] *= scale
if "ZP" in hdul[e].header:
hdul[e].header["ZP"] -= 2.5 * np.log10(scale)
hdul[e].header["CRDS_CTX"] = context
hdul[e].header["R_PHOTOM"] = os.path.basename(r_photom)
log_comment(LOGFILE, msg, verbose=verbose)
return scale
DEFAULT_SNOWBLIND_KWARGS = dict(
require_prefix="jw",
max_fraction=0.3,
new_jump_flag=1024,
min_radius=4,
growth_factor=1.5,
unset_first=True,
)
[docs]def jwst_snowblind_mask(
rate_file,
require_prefix="jw",
max_fraction=0.3,
new_jump_flag=1024,
min_radius=4,
growth_factor=1.5,
unset_first=True,
verbose=True,
skip_after_cal_version='1.14',
**kwargs,
):
"""
Update JWST DQ mask with `snowblind`. See
https://github.com/mpi-astronomy/snowblind.
Requires ``snowblind > 0.1.2``, which currently is just in the fork at
https://github.com/gbrammer/snowblind.
Parameters
----------
rate_file : str
Filename of a ``rate.fits`` exposure
require_prefix : str
Only run if ``rate_file.startswith(require_prefix)``
max_fraction : float
Maximum allowed fraction of flagged pixels relative to the total
new_jump_flag : int
Integer DQ flag of identified snowballs
min_radius : int
Minimum radius of ``JUMP_DET`` flagged groups of pixels
growth_factor : float
Scale factor of the DQ mask
unset_first : bool
Unset the `new_jump_flag` bit of the DQ array before processing
verbose : bool
Whether to print verbose output
kwargs : dict
Additional keyword arguments to be passed to `snowblind.SnowblindStep.call()`
Returns
-------
dq : array-like
Image array with values ``new_jump_flag`` with identified snowballs
mask_frac : float
Fraction of masked pixels
"""
import jwst.datamodels
from packaging.version import Version
from . import jwst_utils
if not os.path.basename(rate_file).startswith(require_prefix):
return None, None
try:
from snowblind import snowblind
from snowblind import __version__ as snowblind_version
except ImportError:
return None, None
if snowblind_version <= "0.1.2":
msg = (
"ImportError: snowblind > 0.1.2 required, get it from the fork at "
"https://github.com/gbrammer/snowblind if not yet available on the "
"main repository at https://github.com/mpi-astronomy/snowblind"
)
log_comment(LOGFILE, msg, verbose=True)
return None, None
step = snowblind.SnowblindStep
# Do we need to reset header keywords?
reset_header = False
with pyfits.open(rate_file) as im:
reset_header = "OINSTRUM" in im[0].header
reset_header &= im[0].header["INSTRUME"] == "WFC3"
if "CAL_VER" in im[0].header:
im_cal_ver = im[0].header["CAL_VER"]
if Version(im_cal_ver) >= Version(skip_after_cal_version):
msg = f"mask_snowballs: {rate_file} "
msg += f"{im_cal_ver} > {skip_after_cal_version}, skip"
log_comment(LOGFILE, msg, verbose=True)
return np.zeros(im["SCI"].data.shape, dtype=int), 0.0
if reset_header:
_ = jwst_utils.set_jwst_to_hst_keywords(rate_file, reset=True, verbose=False)
with jwst.datamodels.open(rate_file) as dm:
if unset_first:
dm.dq -= dm.dq & new_jump_flag
res = step.call(
dm,
save_results=False,
new_jump_flag=new_jump_flag,
min_radius=min_radius,
growth_factor=growth_factor,
**kwargs,
)
if reset_header:
_ = jwst_utils.set_jwst_to_hst_keywords(rate_file, reset=False, verbose=False)
_mask_frac = ((res.dq & new_jump_flag) > 0).sum() / res.dq.size
if _mask_frac > max_fraction:
msg = f"grizli.utils.jwst_snowblind_mask: {rate_file} problem "
msg += f" fraction {_mask_frac*100:.2f} > {max_fraction*100:.2f}"
msg += " turning off..."
res.dq &= 0
else:
msg = f"grizli.utils.jwst_snowblind_mask: {rate_file} {_mask_frac*100:.2f}"
msg += f" masked with DQ={new_jump_flag}"
log_comment(LOGFILE, msg, verbose=verbose)
return (res.dq & new_jump_flag), _mask_frac
[docs]def drizzle_from_visit(
visit,
output=None,
pixfrac=1.0,
kernel="point",
clean=True,
include_saturated=True,
keep_bits=None,
dryrun=False,
skip=None,
extra_wfc3ir_badpix=True,
verbose=True,
scale_photom=True,
context="jwst_1130.pmap",
weight_type="jwst_var",
rnoise_percentile=99,
calc_wcsmap=False,
niriss_ghost_kwargs={},
snowblind_kwargs=None,
jwst_dq_flags=JWST_DQ_FLAGS,
nircam_hot_pixel_kwargs={},
niriss_hot_pixel_kwargs=None, # {'hot_threshold': 7, 'plus_sn_min': 3},
get_dbmask=True,
saturated_lookback=1e4,
write_sat_file=False,
sat_kwargs={},
query_persistence_pixels=True,
**kwargs,
):
"""
Make drizzle mosaic from exposures in a visit dictionary
Parameters
----------
visit : dict
Visit dictionary with 'product' and 'files' keys
output : `~astropy.wcs.WCS`, `~astropy.io.fits.Header`, `~astropy.io.ImageHDU`
Output frame definition. Can be a WCS object, header, or FITS HDU. If
None, then generates a WCS with `grizli.utils.make_maximal_wcs`
pixfrac : float
Drizzle `pixfrac`
kernel : str
Drizzle `kernel` (e.g., 'point', 'square')
clean : bool
Remove exposure files after adding to the mosaic
include_saturated : bool
Include pixels with saturated DQ flag
keep_bits : int, None
Extra DQ bits to keep as valid
dryrun : bool
If True, don't actually produce the output
skip : int
Slice skip to drizzle a subset of exposures
extra_wfc3ir_badpix : bool
Apply extra WFC3/IR bad pix to DQ
verbose : bool
Some verbose message printing
scale_photom : bool
For JWST, apply photometry scale corrections from the
`grizli/data/photom_correction.yml` table
context : str
JWST calibration context to use for photometric scaling
weight_type : 'err', 'median_err', 'time', 'jwst', 'jwst_var', 'median_variance'
Exposure weighting strategy.
- The default 'err' strategy uses the full uncertainty array defined in the
`ERR` image extensions. The alternative
- The 'median_err' strategy uses the median of the `ERR` extension
- The 'time' strategy weights 'median_err' by the `TIME` extension, if
available
- For the 'jwst' strategy, if 'VAR_POISSON' and 'VAR_RNOISE' extensions found,
weight by VAR_RNOISE + median(VAR_POISSON). Fall back to 'median_err'
otherwise.
- For 'jwst_var', use the *weight* as in ``weight_type='jwst'`` but also
make a full variance map propagated from the ``ERR`` noise model.
rnoise_percentile : float
Percentile defining the upper limit of valid `VAR_RNOISE` values, if that
extension is found in the exposure files(e.g., for JWST)
calc_wcsmap : bool
Calculate and return the WCS map
get_dbmask : bool
Get the bad pixel mask from the database
niriss_ghost_kwargs : dict
Keyword arguments for `~grizli.utils.niriss_ghost_mask`
snowblind_kwargs : dict
Arguments to pass to `~grizli.utils.jwst_snowblind_mask` if `snowblind` hasn't
already been run on JWST exposures
jwst_dq_flags : list
List of JWST flag names to include in the bad pixel mask. To ignore, set to
``None``
nircam_hot_pixel_kwargs : dict
Keyword arguments for `grizli.jwst_utils.flag_nircam_hot_pixels`. Set to
``None`` to disable and use the static bad pixel tables.
niriss_hot_pixel_kwargs : dict
Keyword arguments for `grizli.jwst_utils.flag_nircam_hot_pixels` when running
on NIRISS exposures. Set to ``None`` to disable and use the static bad pixel
tables.
saturated_lookback : float
Time, in seconds, to look for saturated pixels in previous exposures
that can cause persistence. Skip if ``saturated_lookback <= 0``.
write_sat_file : bool
Write persistence saturation tables
sat_kwargs : dict
keyword arguments to `~grizli.jwst_utils.get_saturated_pixels`
query_persistence_pixels : bool
Also try to query the full saturated pixel history from the DB with
``saturated_lookback``
Returns
-------
outsci : array-like
SCI array
outwht : array-like
Inverse variance WHT array
outvar : array-like
Optional variance array, if the input weights are not explicitly inverse
variance
header : `~astropy.io.fits.Header`
Image header
flist : list
List of files that were drizzled to the mosaic
wcs_tab : `~astropy.table.Table`
Table of WCS parameters of individual exposures
"""
from shapely.geometry import Polygon
import scipy.ndimage as nd
from astropy.io.fits import PrimaryHDU, ImageHDU
from .prep import apply_region_mask_from_db
from .version import __version__ as grizli__version
from .jwst_utils import get_jwst_dq_bit, flag_nircam_hot_pixels
from .jwst_utils import get_saturated_pixel_table, query_persistence
bucket_name = None
try:
import boto3
from botocore.exceptions import ClientError
s3 = boto3.resource("s3")
s3_client = boto3.client("s3")
except ImportError:
s3 = None
ClientError = None
s3_client = None
_valid_weight_type = [
"err", "median_err", "time", "jwst", "jwst_var", "median_variance",
]
if weight_type not in _valid_weight_type:
print(f"WARNING: weight_type '{weight_type}' must be 'err', 'median_err', ")
print(f" 'jwst', 'median_variance', or 'time'; falling back to 'err'.")
weight_type = "err"
if isinstance(output, pywcs.WCS):
outputwcs = output
elif isinstance(output, pyfits.Header):
outputwcs = pywcs.WCS(output)
elif isinstance(output, PrimaryHDU) | isinstance(output, ImageHDU):
outputwcs = pywcs.WCS(output.header)
elif output is None:
_hdu = make_maximal_wcs(
files=visit["files"], pixel_scale=None, get_hdu=True, verbose=False, pad=4
)
outputwcs = pywcs.WCS(_hdu.header)
else:
return None
if not hasattr(outputwcs, "_naxis1"):
outputwcs._naxis1, outputwcs._naxis2 = outputwcs._naxis
outputwcs.pscale = get_wcs_pscale(outputwcs)
output_poly = Polygon(outputwcs.calc_footprint())
count = 0
ref_photflam = None
indices = []
for i in range(len(visit["files"])):
if "footprints" in visit:
if hasattr(visit["footprints"][i], "intersection"):
olap = visit["footprints"][i].intersection(output_poly)
if olap.area > 0:
indices.append(i)
elif hasattr(visit["footprints"][i], "__len__"):
for _fp in visit["footprints"][i]:
olap = _fp.intersection(output_poly)
if olap.area > 0:
indices.append(i)
break
else:
indices.append(i)
else:
indices.append(i)
if skip is not None:
indices = indices[::skip]
NTOTAL = len(indices)
wcs_rows = []
wcs_colnames = None
wcs_keys = {}
bpdata = 0
saturated_tables = {}
for i in indices:
file = visit["files"][i]
msg = "\n({0:4d}/{1:4d}) Add exposure {2} "
msg += "(weight_type='{3}', rnoise_percentile={4})\n"
msg = msg.format(count + 1, NTOTAL, file, weight_type, rnoise_percentile)
log_comment(LOGFILE, msg, verbose=verbose)
if dryrun:
continue
if (not os.path.exists(file)) & (s3 is not None):
bucket_i = visit["awspath"][i].split("/")[0]
if bucket_name != bucket_i:
bucket_name = bucket_i
bkt = s3.Bucket(bucket_name)
s3_path = "/".join(visit["awspath"][i].split("/")[1:])
remote_file = os.path.join(s3_path, file)
print(" (fetch from s3://{0}/{1})".format(bucket_i, remote_file))
try:
bkt.download_file(
remote_file, file, ExtraArgs={"RequestPayer": "requester"}
)
except ClientError:
print(" (failed s3://{0}/{1})".format(bucket_i, remote_file))
continue
try:
flt = pyfits.open(file)
except OSError:
print(f"open({file}) failed!")
continue
sci_list, wht_list, wcs_list = [], [], []
if weight_type == "jwst_var":
var_list = []
else:
var_list = None
keys = OrderedDict()
for k in ['EXPTIME', 'TELESCOP', 'FILTER','FILTER1', 'FILTER2',
'PUPIL', 'DETECTOR', 'INSTRUME', 'PHOTFLAM', 'PHOTPLAM',
'PHOTFNU', 'PHOTZPT', 'PHOTBW', 'PHOTMODE', 'EXPSTART',
'EXPEND', 'DATE-OBS', 'TIME-OBS',
'UPDA_CTX', 'CRDS_CTX', 'R_DISTOR', 'R_PHOTOM', 'R_FLAT']:
if k in flt[0].header:
keys[k] = flt[0].header[k]
bpdata = None
_nsat = None
if flt[0].header["TELESCOP"] in ["JWST"]:
bits = 4
include_saturated = False
# bpdata = 0
_inst = flt[0].header["INSTRUME"]
_det = flt[0].header['DETECTOR']
if (extra_wfc3ir_badpix) & (_inst in ['NIRCAM','NIRISS']):
bpfiles = [os.path.join(os.path.dirname(__file__),
f'data/nrc_badpix_240627_{_det}.fits.gz')]
bpfiles += [os.path.join(os.path.dirname(__file__),
f'data/nrc_badpix_240112_{_det}.fits.gz')]
bpfiles += [os.path.join(os.path.dirname(__file__),
f'data/nrc_badpix_231206_{_det}.fits.gz')]
bpfiles += [os.path.join(os.path.dirname(__file__),
f'data/jwst_nircam_newhot_{_det}_extra20231129.fits.gz')]
bpfiles += [os.path.join(os.path.dirname(__file__),
f'data/nrc_badpix_20230710_{_det}.fits.gz')]
bpfiles += [os.path.join(os.path.dirname(__file__),
f'data/{_det.lower()}_badpix_20241001.fits.gz')] # NIRISS
bpfiles += [os.path.join(os.path.dirname(__file__),
f'data/{_det.lower()}_badpix_20230710.fits.gz')] # NIRISS
bpfiles += [os.path.join(os.path.dirname(__file__),
f'data/nrc_badpix_230701_{_det}.fits.gz')]
bpfiles += [os.path.join(os.path.dirname(__file__),
f'data/nrc_badpix_230120_{_det}.fits.gz')]
bpfiles += [os.path.join(os.path.dirname(__file__),
f'data/nrc_lowpix_0916_{_det}.fits.gz')]
for bpfile in bpfiles:
if os.path.exists(bpfile):
bpdata = pyfits.open(bpfile)[0].data
if True:
bpdata = nd.binary_dilation(bpdata > 0) * 1024
else:
bpdata = (bpdata > 0) * 1024
msg = f"Use extra badpix in {bpfile}"
log_comment(LOGFILE, msg, verbose=verbose)
break
if bpdata is None:
bpdata = np.zeros(flt["SCI"].data.shape, dtype=int)
# Directly flag hot pixels rather than use mask
if (_inst in ["NIRCAM"]) & (nircam_hot_pixel_kwargs is not None):
_sn, dq_flag, _count = flag_nircam_hot_pixels(
flt, **nircam_hot_pixel_kwargs
)
if (_count > 0) & (_count < 8192):
bpdata |= ((dq_flag > 0) * 1024).astype(bpdata.dtype)
# extra_wfc3ir_badpix = False
else:
msg = f" flag_nircam_hot_pixels: {_count} out of range"
log_comment(LOGFILE, msg, verbose=verbose)
elif (_inst in ["NIRISS"]) & (niriss_hot_pixel_kwargs is not None):
_sn, dq_flag, _count = flag_nircam_hot_pixels(
flt, **niriss_hot_pixel_kwargs
)
if (_count > 0) & (_count < 8192):
bpdata |= ((dq_flag > 0) * 1024).astype(bpdata.dtype)
# extra_wfc3ir_badpix = False
else:
msg = f" flag_nircam_hot_pixels: {_count} out of range (NIRISS)"
log_comment(LOGFILE, msg, verbose=verbose)
if (snowblind_kwargs is not None) & (_inst in ["NIRCAM", "NIRISS"]):
# Already processed with snowblind?
if "SNOWBLND" in flt["SCI"].header:
msg = "Already processed with `snowblind`"
log_comment(LOGFILE, msg, verbose=verbose)
else:
sdq, sfrac = jwst_snowblind_mask(file, **snowblind_kwargs)
if sdq is not None:
bpdata |= sdq
if get_dbmask:
dbmask = apply_region_mask_from_db(
os.path.basename(file), in_place=False, verbose=True
)
if dbmask is not None:
bpdata |= dbmask * 1
# NIRISS ghost mask
if (_inst in ["NIRISS"]) & (niriss_ghost_kwargs is not None):
if "verbose" not in niriss_ghost_kwargs:
niriss_ghost_kwargs["verbose"] = verbose
_ghost = niriss_ghost_mask(flt, **niriss_ghost_kwargs)
bpdata |= _ghost * 1024
# Negative
if "MDRIZSKY" in flt["SCI"].header:
_low = (flt["SCI"].data - flt["SCI"].header["MDRIZSKY"]) < -5 * flt[
"ERR"
].data
msg = f"Extra -5 sigma low pixels: N= {_low.sum()} "
msg += f" ( {_low.sum()/_low.size*100:.1} %)"
log_comment(LOGFILE, msg, verbose=verbose)
bpdata |= _low * 1024
# History of saturated pixels for persistence
if saturated_lookback > 0:
if _det not in saturated_tables:
saturated_tables[_det] = {'expstart':[], 'ij':[]}
_start = flt[0].header['EXPSTART']
_df = get_saturated_pixel_table(
file=file,
output="df",
**sat_kwargs,
)
_sat_tab = saturated_tables[_det]
_sat_tab["expstart"].append(_start)
_sat_tab["ij"].append(_df)
if write_sat_file:
sat_file = file.replace(".fits", ".sat.csv.gz")
_df.to_csv(sat_file, index=False)
_sat_history = np.zeros_like(bpdata, dtype=bool)
_sat_count = 0
for _starti, _df in zip(_sat_tab["expstart"], _sat_tab["ij"]):
if (
(_starti < _start)
& ((_start - _starti)*86400 < saturated_lookback)
):
_sat_history[_df.i, _df.j] |= True
_sat_count += 1
_nsat = _sat_history.sum()
bpdata |= _sat_history * 1024
msg = (
f"Found {_nsat} saturated pixels in {_sat_count} "
f" previous {_det} exposures within {saturated_lookback:.0f} sec"
)
log_comment(LOGFILE, msg, verbose=verbose)
if query_persistence_pixels & (saturated_lookback > 0):
try:
_pers = query_persistence(
file,
saturated_lookback=saturated_lookback
)
if len(_pers) > 0:
bpdata[_pers["i"], _pers["j"]] |= 1024
except:
pass
elif flt[0].header["DETECTOR"] == "IR":
bits = 576
if extra_wfc3ir_badpix:
if (i == indices[0]) | (not hasattr(bpdata, "shape")):
bpfile = os.path.join(
os.path.dirname(__file__),
"data/wfc3ir_badpix_spars200_22.03.31.fits.gz",
)
bpdata = pyfits.open(bpfile)[0].data
msg = f"Use extra badpix in {bpfile}"
log_comment(LOGFILE, msg, verbose=verbose)
else:
bits = 64 + 32
bpdata = 0
if include_saturated:
bits |= 256
if keep_bits is not None:
bits |= keep_bits
if scale_photom:
# Scale to a particular JWST context and update header keywords
# like PHOTFLAM, PHOTFNU
_scale_jwst_photom = jwst_crds_photom_scale(
flt, update=True, context=context, verbose=verbose
)
# These might have changed
for k in ["PHOTFLAM", "PHOTFNU", "PHOTMJSR", "ZP", "R_PHOTOM", "CRDS_CTX"]:
if k in flt[0].header:
keys[k] = flt[0].header[k]
# Additional scaling
_key, _scale_photom = get_photom_scale(flt[0].header, verbose=verbose)
else:
_scale_photom = 1.0
if "PHOTFLAM" in keys:
msg = " 0 PHOTFLAM={0:.2e}, scale={1:.3f}"
msg = msg.format(keys["PHOTFLAM"], _scale_photom)
log_comment(LOGFILE, msg, verbose=verbose)
if ref_photflam is None:
ref_photflam = keys["PHOTFLAM"]
median_weight = None
for ext in [1, 2, 3, 4]:
if ("SCI", ext) in flt:
h = flt[("SCI", ext)].header
if "MDRIZSKY" in h:
sky_value = h["MDRIZSKY"]
else:
sky_value = 0
if ("BKG", ext) in flt:
has_bkg = True
sky = flt["BKG", ext].data + sky_value
else:
has_bkg = False
sky = sky_value
if h["BUNIT"] == "ELECTRONS":
to_per_sec = 1.0 / keys["EXPTIME"]
else:
to_per_sec = 1.0
phot_scale = to_per_sec * _scale_photom
if "PHOTFLAM" in h:
if ref_photflam is None:
ref_photflam = h["PHOTFLAM"]
phot_scale = h["PHOTFLAM"] / ref_photflam * _scale_photom
if "PHOTFNU" not in h:
h["PHOTFNU"] = (
photfnu_from_photflam(h["PHOTFLAM"], h["PHOTPLAM"]),
"Inverse sensitivity, Jy/DN",
)
msg = " PHOTFLAM={0:.2e}, scale={1:.3f}"
msg = msg.format(h["PHOTFLAM"], phot_scale)
log_comment(LOGFILE, msg, verbose=verbose)
keys["PHOTFLAM"] = h["PHOTFLAM"]
for k in [
"PHOTFLAM",
"PHOTPLAM",
"PHOTFNU",
"PHOTZPT",
"PHOTBW",
"PHOTMODE",
"PHOTMJSR",
"PIXAR_SR",
]:
if k in h:
keys[k] = h[k]
phot_scale *= to_per_sec
try:
wcs_i = pywcs.WCS(header=flt[("SCI", ext)].header, fobj=flt)
wcs_i.pscale = get_wcs_pscale(wcs_i)
except KeyError:
print(f"Failed to initialize WCS on {file}[SCI,{ext}]")
continue
wcsh = to_header(wcs_i)
row = [file, ext, keys["EXPTIME"]]
if wcs_colnames is None:
wcs_colnames = ["file", "ext", "exptime"]
for k in wcsh:
wcs_colnames.append(k.lower())
wcs_keys[k.lower()] = wcsh[k]
for k in wcs_colnames[3:]:
ku = k.upper()
if ku not in wcsh:
print(f"Keyword {ku} not found in WCS header")
row.append(wcs_keys[k] * 0)
else:
row.append(wcsh[ku])
for k in wcsh:
if k.lower() not in wcs_colnames:
print(f"Extra keyword {ku} found in WCS header")
wcs_rows.append(row)
err_data = flt[("ERR", ext)].data * phot_scale
# JWST: just 1,1024,4096 bits
if flt[0].header["TELESCOP"] in ["JWST"]:
bad_bits = 1 | 1024 | 4096
if jwst_dq_flags is not None:
bad_bits |= get_jwst_dq_bit(jwst_dq_flags, verbose=verbose)
dq = flt[("DQ", ext)].data & bad_bits
dq |= bpdata.astype(dq.dtype)
# Clipping threshold for BKG extensions, global at top
# BKG_CLIP = [scale, percentile_lo, percentile_hi]
if has_bkg & (BKG_CLIP is not None):
# percentiles
bkg_lo, bkg_hi = np.nanpercentile(
flt["BKG"].data[dq == 0], BKG_CLIP[1:3]
)
# make sure lower (upper) limit is negative (positive)
clip_lo = -np.abs(bkg_lo)
clip_hi = np.abs(bkg_hi)
_bad_bkg = flt["BKG"].data < BKG_CLIP[0] * bkg_lo
_bad_bkg |= flt["BKG"].data > BKG_CLIP[0] * bkg_hi
# OR into dq mask
msg = f"Bad bkg pixels: N= {_bad_bkg.sum()} "
msg += f" ( {_bad_bkg.sum()/_bad_bkg.size*100:.1} %)"
log_comment(LOGFILE, msg, verbose=verbose)
dq |= _bad_bkg * 1024
else:
dq = mod_dq_bits(flt[("DQ", ext)].data, okbits=bits) | bpdata
wht = 1 / err_data ** 2
_msk = (err_data <= 0) | (dq > 0)
wht[_msk] = 0
if weight_type == "jwst_var":
_var = err_data**2
_var[_msk] = 0
var_list.append(_var)
if weight_type.startswith("jwst"):
if (("VAR_RNOISE", ext) in flt) & (rnoise_percentile is not None):
_rn_data = flt[("VAR_RNOISE", ext)].data
rnoise_value = np.nanpercentile(
_rn_data[~_msk], rnoise_percentile
)
_msk |= _rn_data >= rnoise_value
if ("VAR_POISSON", ext) in flt:
# Weight by VAR_RNOISE + median(VAR_POISSON)
if (~_msk).sum() > 0:
_var_data = flt[("VAR_POISSON", ext)].data[~_msk]
med_poisson = np.nanmedian(_var_data)
var = flt["VAR_RNOISE", ext].data + med_poisson
var *= phot_scale ** 2
wht = 1.0 / var
wht[_msk | (var <= 0)] = 0
else:
# Fall back to median_err
median_weight = np.nanmedian(wht[~_msk])
if (not np.isfinite(median_weight)) | ((~_msk).sum() == 0):
median_weight = 0
wht[~_msk] = median_weight
median_weight = np.nanmedian(wht[~_msk])
if (not np.isfinite(median_weight)) | ((~_msk).sum() == 0):
median_weight = 0
msg = f" ext (SCI,{ext}), sky={sky_value:.3f}"
msg += f" has_bkg:{has_bkg}"
msg += f" median_weight:{median_weight:.2e}"
log_comment(LOGFILE, msg, verbose=verbose)
# Use median(ERR) as the full image weight,
# optionally scaling by the TIME array
if weight_type in ["median_err", "time"]:
wht[~_msk] = median_weight
if weight_type == "time":
if ("TIME", ext) in flt:
if flt[("TIME", ext)].data is not None:
_time = flt[("TIME", ext)].data * 1
tmax = np.nanmax(_time[~_msk])
_time /= tmax
wht[~_msk] *= _time[~_msk]
msg = f"scale weight by (TIME,{ext})/{tmax:.1f}"
log_comment(LOGFILE, msg, verbose=verbose)
wht_list.append(wht)
sci_i = (flt[("SCI", ext)].data - sky) * phot_scale
sci_i[wht <= 0] = 0
sci_list.append(sci_i)
if not hasattr(wcs_i, "pixel_shape"):
wcs_i.pixel_shape = wcs_i._naxis1, wcs_i._naxis2
if not hasattr(wcs_i, "_naxis1"):
wcs_i._naxis1, wcs_i._naxis2 = wcs_i._naxis
wcs_list.append(wcs_i)
pscale_ratio = (wcs_i.pscale / outputwcs.pscale)
if count == 0:
res = drizzle_array_groups(
sci_list,
wht_list,
wcs_list,
var_list=var_list,
median_weight=(weight_type == 'median_variance'),
outputwcs=outputwcs,
scale=0.1,
kernel=kernel,
pixfrac=pixfrac,
calc_wcsmap=calc_wcsmap,
verbose=verbose,
data=None,
)
outsci, outwht, outvar, outctx, header, xoutwcs = res
header["EXPTIME"] = flt[0].header["EXPTIME"]
header["NDRIZIM"] = 1
header["PIXFRAC"] = pixfrac
header["KERNEL"] = kernel
header["OKBITS"] = (bits, "FLT bits treated as valid")
header["PHOTSCAL"] = _scale_photom, "Scale factor applied"
header["GRIZLIV"] = grizli__version, "Grizli code version"
header["WHTTYPE"] = weight_type, "Exposure weighting strategy"
header["RNPERC"] = rnoise_percentile, "VAR_RNOISE clip percentile for JWST"
header["PSCALER"] = pscale_ratio, "Ratio of input to output pixel scales"
for k in keys:
header[k] = keys[k]
else:
# outvar = Sum(wht**2 * var) / Sum(wht)**2, so
# need to accumulate updates to Sum(wht * (wht * var)) / Sum(wht)
if outvar is None:
varnum = None
else:
varnum = outvar * outwht
data = outsci, outwht, outctx, varnum
res = drizzle_array_groups(
sci_list,
wht_list,
wcs_list,
median_weight=(weight_type == 'median_variance'),
var_list=var_list,
outputwcs=outputwcs,
scale=0.1,
kernel=kernel,
pixfrac=pixfrac,
calc_wcsmap=calc_wcsmap,
verbose=verbose,
data=data,
)
outsci, outwht, outvar, outctx = res[:4]
header["EXPTIME"] += flt[0].header["EXPTIME"]
header["NDRIZIM"] += 1
count += 1
header["FLT{0:05d}".format(count)] = file
if median_weight is not None:
header["WHT{0:05d}".format(count)] = (
median_weight,
f"Median weight of exposure {count}",
)
if _nsat is not None:
header["SAT{0:05d}".format(count)] = (
_nsat, f"Number of pixels flagged for persistence"
)
flt.close()
# xfiles = glob.glob('*')
# print('Clean: ', clean, xfiles)
if clean:
os.remove(file)
if "awspath" in visit:
awspath = visit["awspath"]
else:
awspath = ["." for f in visit["files"]]
if len(awspath) == 1:
awspath = [awspath[0] for f in visit["files"]]
elif isinstance(awspath, str):
_awspath = [awspath for f in visit["files"]]
awspath = _awspath
flist = ["{0}/{1}".format(awspath, visit["files"][i]) for i in indices]
if dryrun:
return flist
elif count == 0:
return None
else:
wcs_tab = GTable(names=wcs_colnames, rows=wcs_rows)
outwht *= pscale_ratio**4 # (wcs_i.pscale / outputwcs.pscale) ** 4
if outvar is not None:
# Extra factors of the pixel area ratio in variance, which comes from
# outvar = varnum / outwht
outvar *= pscale_ratio**-2
return outsci, outwht, outvar, outctx, header, flist