Пример #1
0
def test_load_extinction():
    # Load
    extinct = flux_calib.load_extinction_data(121.6428, 37.3413889)
    np.testing.assert_allclose(extinct['wave'][0], 3200.)
    assert extinct['wave'].unit == units.AA
    np.testing.assert_allclose(extinct['mag_ext'][0], 1.084)
    # Fail
    extinct = flux_calib.load_extinction_data(0., 37.3413889)
    assert extinct is None
Пример #2
0
def test_extinction_correction():
    # Load
    extinct = flux_calib.load_extinction_data(121.6428, 37.3413889)
    # Correction
    wave = np.arange(3000., 10000.) * units.AA
    AM = 1.5
    flux_corr = flux_calib.extinction_correction(wave, AM, extinct)
    # Test
    np.testing.assert_allclose(flux_corr[0], 4.47095192)
Пример #3
0
    def __init__(self, spectrograph, par, sens_file=None, debug=False):

        # Init
        self.spectrograph = spectrograph
        self.par = par

        # Get the extinction data
        self.extinction_data = flux_calib.load_extinction_data(
            self.spectrograph.telescope['longitude'],
            self.spectrograph.telescope['latitude'])

        # Parameters
        if sens_file is None:
            self.sens_file = par['sensfunc']
        else:
            self.sens_file = sens_file
        self.multi_det = par['multi_det']

        # Set telluric option
        self.telluric = par['telluric']

        # Main outputs
        self.sens_dict = None if self.sens_file is None \
                            else self.load_sens_dict(self.sens_file)
        # Attributes
        self.steps = []

        # Key Internals
        self.std = None  # Standard star spectrum (SpecObj object)
        self.std_idx = None  # Nested indices for the std_specobjs list that corresponds
        # to the star!
        # standard/telluric star information
        self.star_type = par['star_type']
        self.star_mag = par['star_mag']
        # telluric mask keywords
        self.BALM_MASK_WID = par['balm_mask_wid']
        # sensfunc fitting parameters
        self.poly_norder = par['poly_norder']
        self.polycorrect = par['polycorrect']
        self.debug = debug
Пример #4
0
    def apply_flux_calib(self,
                         wave_sens,
                         sensfunc,
                         exptime,
                         telluric=None,
                         extinct_correct=False,
                         airmass=None,
                         longitude=None,
                         latitude=None):
        """
        Apply a sensitivity function to our spectrum

        FLAM, FLAM_SIG, and FLAM_IVAR are generated

        Args:
            sens_dict (dict):
                Sens Function dict
            exptime (float):
            telluric_correct:
            extinct_correct:
            airmass (float, optional):
            longitude (float, optional):
                longitude in degree for observatory
            latitude:
                latitude in degree for observatory
                Used for extinction correction

        """
        # Loop on extraction modes
        for attr in ['BOX', 'OPT']:
            if attr + '_WAVE' not in self._data.keys():
                continue
            msgs.info("Fluxing {:s} extraction for:".format(attr) +
                      msgs.newline() + "{}".format(self))

            wave = self[attr + '_WAVE']
            # Interpolate the sensitivity function onto the wavelength grid of the data

            # TODO Telluric corrections via this method are deprecated
            # Did the user request a telluric correction?
            if telluric is not None:
                # This assumes there is a separate telluric key in this dict.
                msgs.info('Applying telluric correction')
                sensfunc = sensfunc * (telluric > 1e-10) / (telluric +
                                                            (telluric < 1e-10))

            sensfunc_obs = np.zeros_like(wave)
            wave_mask = wave > 1.0  # filter out masked regions or bad wavelengths
            try:
                sensfunc_obs[wave_mask] = interpolate.interp1d(
                    wave_sens, sensfunc, bounds_error=True)(wave[wave_mask])
            except ValueError:
                msgs.error(
                    "Your data extends beyond the bounds of your sensfunc. " +
                    msgs.newline() +
                    "Adjust the par['sensfunc']['extrap_blu'] and/or par['sensfunc']['extrap_red'] to extrapolate "
                    "further and recreate your sensfunc.")

            if extinct_correct:
                if longitude is None or latitude is None:
                    msgs.error(
                        'You must specify longitude and latitude if we are extinction correcting'
                    )
                # Apply Extinction if optical bands
                msgs.info("Applying extinction correction")
                msgs.warn(
                    "Extinction correction applyed only if the spectra covers <10000Ang."
                )
                extinct = flux_calib.load_extinction_data(longitude, latitude)
                ext_corr = flux_calib.extinction_correction(
                    wave * units.AA, airmass, extinct)
                senstot = sensfunc_obs * ext_corr
            else:
                senstot = sensfunc_obs.copy()

            flam = self[attr + '_COUNTS'] * senstot / exptime
            flam_sig = (senstot / exptime) / (np.sqrt(
                self[attr + '_COUNTS_IVAR']))
            flam_var = self[attr + '_COUNTS_IVAR'] / (senstot / exptime)**2

            # Mask bad pixels
            msgs.info(" Masking bad pixels")
            msk = np.zeros_like(senstot).astype(bool)
            msk[senstot <= 0.] = True
            msk[self[attr + '_COUNTS_IVAR'] <= 0.] = True
            flam[msk] = 0.
            flam_sig[msk] = 0.
            flam_var[msk] = 0.
            # TODO JFH We need to update the mask here. I think we need a mask for the counts and a mask for the flam,
            # since they can in principle be different. We are masking bad sensfunc locations.

            # Finish
            self[attr + '_FLAM'] = flam
            self[attr + '_FLAM_SIG'] = flam_sig
            self[attr + '_FLAM_IVAR'] = flam_var
Пример #5
0
def coadd_cube(files, parset, overwrite=False):
    """ Main routine to coadd spec2D files into a 3D datacube

    Args:
        files (list):
            List of all spec2D files
        parset (:class:`pypeit.par.core.PypeItPar`):
            An instance of the parameter set.
        overwrite (bool):
            Overwrite the output file, if it exists?
    """
    # Get the detector number
    det = 1 if parset is None else parset['rdx']['detnum']

    # Load the spectrograph
    spec2DObj = spec2dobj.Spec2DObj.from_file(files[0], det)
    specname = spec2DObj.head0['PYP_SPEC']
    spec = load_spectrograph(specname)

    # Grab the parset, if not provided
    if parset is None: parset = spec.default_pypeit_par()
    cubepar = parset['reduce']['cube']

    # Check the output file
    outfile = cubepar['output_filename'] if ".fits" in cubepar[
        'output_filename'] else cubepar['output_filename'] + ".fits"
    out_whitelight = outfile.replace(".fits", "_whitelight.fits")
    if os.path.exists(outfile) and not overwrite:
        msgs.error("Output filename already exists:" + msgs.newline() +
                   outfile)
    elif os.path.exists(
            out_whitelight) and cubepar['save_whitelight'] and not overwrite:
        msgs.error("Output filename already exists:" + msgs.newline() +
                   out_whitelight)
    # Check the reference cube and image exist, if requested
    ref_scale = None  # This will be used to correct relative scaling among the various input frames
    if cubepar['standard_cube'] is not None:
        if not os.path.exists(cubepar['standard_cube']):
            msgs.error("Standard cube does not exist:" + msgs.newline() +
                       cubepar['reference_cube'])
        cube = fits.open(cubepar['standard_cube'])
        ref_scale = cube['REFSCALE'].data
    if cubepar['reference_image'] is not None:
        if not os.path.exists(cubepar['reference_image']):
            msgs.error("Reference cube does not exist:" + msgs.newline() +
                       cubepar['reference_image'])
    if cubepar['flux_calibrate']:
        msgs.error("Flux calibration is not currently implemented" +
                   msgs.newline() + "Please set 'flux_calibrate = False'")

    # prep
    numfiles = len(files)
    combine = cubepar['combine']

    all_ra, all_dec, all_wave = np.array([]), np.array([]), np.array([])
    all_sci, all_ivar, all_idx, all_wghts = np.array([]), np.array(
        []), np.array([]), np.array([])
    all_wcs = []
    dspat = None if cubepar['spatial_delta'] is None else cubepar[
        'spatial_delta'] / 3600.0  # binning size on the sky (/3600 to convert to degrees)
    dwv = cubepar[
        'wave_delta']  # binning size in wavelength direction (in Angstroms)
    wave_ref = None
    whitelight_img = None  # This is the whitelight image based on all input spec2d frames
    weights = np.ones(numfiles)  # Weights to use when combining cubes
    for ff, fil in enumerate(files):
        # Load it up
        spec2DObj = spec2dobj.Spec2DObj.from_file(fil, det)
        detector = spec2DObj.detector

        # Setup for PypeIt imports
        msgs.reset(verbosity=2)

        if ref_scale is None:
            ref_scale = spec2DObj.scaleimg.copy()
        # Extract the information
        sciimg = (spec2DObj.sciimg - spec2DObj.skymodel) * (
            ref_scale / spec2DObj.scaleimg
        )  # Subtract sky and apply relative sky
        ivar = spec2DObj.ivarraw / (ref_scale / spec2DObj.scaleimg)**2
        waveimg = spec2DObj.waveimg
        bpmmask = spec2DObj.bpmmask

        # Grab the slit edges
        slits = spec2DObj.slits

        wave0 = waveimg[waveimg != 0.0].min()
        diff = waveimg[1:, :] - waveimg[:-1, :]
        dwv = float(np.median(diff[diff != 0.0]))
        msgs.info(
            "Using wavelength solution: wave0={0:.3f}, dispersion={1:.3f} Angstrom/pixel"
            .format(wave0, dwv))

        msgs.info("Constructing slit image")
        slitid_img_init = slits.slit_img(pad=0,
                                         initial=True,
                                         flexure=spec2DObj.sci_spat_flexure)
        onslit_gpm = (slitid_img_init > 0) & (bpmmask == 0)

        # Grab the WCS of this frame
        wcs = spec.get_wcs(spec2DObj.head0, slits, detector.platescale, wave0,
                           dwv)
        all_wcs.append(copy.deepcopy(wcs))

        # Find the largest spatial scale of all images being combined
        # TODO :: probably need to put this in the DetectorContainer
        pxscl = detector.platescale * parse.parse_binning(
            detector.binning)[1] / 3600.0  # This should be degrees/pixel
        slscl = spec.get_meta_value([spec2DObj.head0], 'slitwid')
        if dspat is None:
            dspat = max(pxscl, slscl)
        elif max(pxscl, slscl) > dspat:
            dspat = max(pxscl, slscl)

        # Generate an RA/DEC image
        msgs.info("Generating RA/DEC image")
        raimg, decimg, minmax = slits.get_radec_image(
            wcs, initial=True, flexure=spec2DObj.sci_spat_flexure)

        # Perform the DAR correction
        if wave_ref is None:
            wave_ref = 0.5 * (np.min(waveimg[onslit_gpm]) +
                              np.max(waveimg[onslit_gpm]))
        # Get DAR parameters
        raval = spec.get_meta_value([spec2DObj.head0], 'ra')
        decval = spec.get_meta_value([spec2DObj.head0], 'dec')
        obstime = spec.get_meta_value([spec2DObj.head0], 'obstime')
        pressure = spec.get_meta_value([spec2DObj.head0], 'pressure')
        temperature = spec.get_meta_value([spec2DObj.head0], 'temperature')
        rel_humidity = spec.get_meta_value([spec2DObj.head0], 'humidity')
        coord = SkyCoord(raval, decval, unit=(units.deg, units.deg))
        location = spec.location  # TODO :: spec.location should probably end up in the TelescopePar (spec.telescope.location)
        ra_corr, dec_corr = dc_utils.dar_correction(waveimg[onslit_gpm],
                                                    coord,
                                                    obstime,
                                                    location,
                                                    pressure,
                                                    temperature,
                                                    rel_humidity,
                                                    wave_ref=wave_ref)
        raimg[onslit_gpm] += ra_corr
        decimg[onslit_gpm] += dec_corr

        # Get copies of arrays to be saved
        wave_ext = waveimg[onslit_gpm].copy()
        flux_ext = sciimg[onslit_gpm].copy()
        ivar_ext = ivar[onslit_gpm].copy()

        # Perform extinction correction
        msgs.info("Applying extinction correction")
        longitude = spec.telescope['longitude']
        latitude = spec.telescope['latitude']
        airmass = spec2DObj.head0[spec.meta['airmass']['card']]
        extinct = load_extinction_data(longitude, latitude)
        # extinction_correction requires the wavelength is sorted
        wvsrt = np.argsort(wave_ext)
        ext_corr = extinction_correction(wave_ext[wvsrt] * units.AA, airmass,
                                         extinct)
        # Correct for extinction
        flux_sav = flux_ext[wvsrt] * ext_corr
        ivar_sav = ivar_ext[wvsrt] / ext_corr**2
        # sort back to the original ordering
        resrt = np.argsort(wvsrt)

        # Calculate the weights relative to the zeroth cube
        if ff != 0:
            weights[ff] = np.median(flux_sav[resrt] *
                                    np.sqrt(ivar_sav[resrt]))**2

        # Store the information
        numpix = raimg[onslit_gpm].size
        all_ra = np.append(all_ra, raimg[onslit_gpm].copy())
        all_dec = np.append(all_dec, decimg[onslit_gpm].copy())
        all_wave = np.append(all_wave, wave_ext.copy())
        all_sci = np.append(all_sci, flux_sav[resrt].copy())
        all_ivar = np.append(all_ivar, ivar_sav[resrt].copy())
        all_idx = np.append(all_idx, ff * np.ones(numpix))
        all_wghts = np.append(all_wghts, weights[ff] * np.ones(numpix))

    # Grab cos(dec) for convenience
    cosdec = np.cos(np.mean(all_dec) * np.pi / 180.0)

    # Register spatial offsets between all frames if several frames are being combined
    if combine:

        # Check if a reference whitelight image should be used to register the offsets
        if cubepar["reference_image"] is None:
            # Generate white light images
            whitelight_imgs, _, _ = dc_utils.make_whitelight(
                all_ra, all_dec, all_wave, all_sci, all_wghts, all_idx, dspat)
            # ref_idx will be the index of the cube with the highest S/N
            ref_idx = np.argmax(weights)
            reference_image = whitelight_imgs[:, :, ref_idx].copy()
            msgs.info(
                "Calculating spatial translation of each cube relative to cube #{0:d})"
                .format(ref_idx + 1))
        else:
            ref_idx = -1  # Don't use an index
            # Load reference information
            reference_image, whitelight_imgs, wlwcs = \
                dc_utils.make_whitelight_fromref(all_ra, all_dec, all_wave, all_sci, all_wghts, all_idx, dspat,
                                                 cubepar['reference_image'])
            msgs.info(
                "Calculating the spatial translation of each cube relative to user-defined 'reference_image'"
            )
        # Calculate the image offsets - check the reference is a zero shift
        ra_shift_ref, dec_shift_ref = calculate_image_offset(
            reference_image.copy(), reference_image.copy())
        for ff in range(numfiles):
            # Don't correlate the reference image with itself
            if ff == ref_idx:
                continue
            # Calculate the shift
            ra_shift, dec_shift = calculate_image_offset(
                whitelight_imgs[:, :, ff], reference_image.copy())
            # Convert to reference
            ra_shift -= ra_shift_ref
            dec_shift -= dec_shift_ref
            # Convert pixel shift to degress shift
            ra_shift *= dspat / cosdec
            dec_shift *= dspat
            msgs.info(
                "Spatial shift of cube #{0:d}: RA, DEC (arcsec) = {1:+0.3f}, {2:+0.3f}"
                .format(ff + 1, ra_shift * 3600.0, dec_shift * 3600.0))
            # Apply the shift
            all_ra[all_idx == ff] += ra_shift
            all_dec[all_idx == ff] += dec_shift

        # Generate a white light image of *all* data
        msgs.info("Generating global white light image")
        if cubepar["reference_image"] is None:
            whitelight_img, _, wlwcs = dc_utils.make_whitelight(
                all_ra, all_dec, all_wave, all_sci, all_wghts,
                np.zeros(all_ra.size), dspat)
        else:
            _, whitelight_img, wlwcs = \
                dc_utils.make_whitelight_fromref(all_ra, all_dec, all_wave, all_sci, all_wghts, np.zeros(all_ra.size),
                                                 dspat, cubepar['reference_image'])

        # Calculate the relative spectral weights of all pixels
        all_wghts = dc_utils.compute_weights(
            all_ra,
            all_dec,
            all_wave,
            all_sci,
            all_ivar,
            all_idx,
            whitelight_img[:, :, 0],
            dspat,
            dwv,
            relative_weights=cubepar['relative_weights'])
    # Check if a whitelight image should be saved
    if cubepar['save_whitelight']:
        # Check if the white light image still needs to be generated - if so, generate it now
        if whitelight_img is None:
            msgs.info("Generating global white light image")
            if cubepar["reference_image"] is None:
                whitelight_img, _, wlwcs = dc_utils.make_whitelight(
                    all_ra, all_dec, all_wave, all_sci, all_wghts,
                    np.zeros(all_ra.size), dspat)
            else:
                _, whitelight_img, wlwcs = \
                    dc_utils.make_whitelight_fromref(all_ra, all_dec, all_wave, all_sci, all_wghts,
                                                     np.zeros(all_ra.size),
                                                     dspat, cubepar['reference_image'])
        # Prepare and save the fits file
        msgs.info("Saving white light image as: {0:s}".format(out_whitelight))
        img_hdu = fits.PrimaryHDU(whitelight_img.T, header=wlwcs.to_header())
        img_hdu.writeto(out_whitelight, overwrite=overwrite)

    # Setup the cube ranges
    ra_min = cubepar['ra_min'] if cubepar['ra_min'] is not None else np.min(
        all_ra)
    ra_max = cubepar['ra_max'] if cubepar['ra_max'] is not None else np.max(
        all_ra)
    dec_min = cubepar['dec_min'] if cubepar['dec_min'] is not None else np.min(
        all_dec)
    dec_max = cubepar['dec_max'] if cubepar['dec_max'] is not None else np.max(
        all_dec)
    wav_min = cubepar['wave_min'] if cubepar[
        'wave_min'] is not None else np.min(all_wave)
    wav_max = cubepar['wave_max'] if cubepar[
        'wave_max'] is not None else np.max(all_wave)
    if cubepar['wave_delta'] is not None: dwv = cubepar['wave_delta']
    # Generate a master WCS to register all frames
    coord_min = [ra_min, dec_min, wav_min]
    coord_dlt = [dspat, dspat, dwv]
    masterwcs = dc_utils.generate_masterWCS(coord_min,
                                            coord_dlt,
                                            name=specname)
    msgs.info(msgs.newline() + "-" * 40 + msgs.newline() +
              "Parameters of the WCS:" + msgs.newline() +
              "RA   min, max = {0:f}, {1:f}".format(ra_min, ra_max) +
              msgs.newline() +
              "DEC  min, max = {0:f}, {1:f}".format(dec_min, dec_max) +
              msgs.newline() +
              "WAVE min, max = {0:f}, {1:f}".format(wav_min, wav_max) +
              msgs.newline() + "Spaxel size = {0:f}''".format(3600.0 * dspat) +
              msgs.newline() + "Wavelength step = {0:f} A".format(dwv) +
              msgs.newline() + "-" * 40)

    # Generate the output binning
    if combine:
        numra = int((ra_max - ra_min) * cosdec / dspat)
        numdec = int((dec_max - dec_min) / dspat)
        numwav = int((wav_max - wav_min) / dwv)
        xbins = np.arange(1 + numra) - 0.5
        ybins = np.arange(1 + numdec) - 0.5
        spec_bins = np.arange(1 + numwav) - 0.5
    else:
        slitlength = int(
            np.round(
                np.median(slits.get_slitlengths(initial=True, median=True))))
        numwav = int((np.max(waveimg) - wave0) / dwv)
        xbins, ybins, spec_bins = spec.get_datacube_bins(
            slitlength, minmax, numwav)

    # Make the cube
    msgs.info("Generating pixel coordinates")
    if combine:
        pix_coord = masterwcs.wcs_world2pix(all_ra, all_dec,
                                            all_wave * 1.0E-10, 0)
        hdr = masterwcs.to_header()
    else:
        pix_coord = wcs.wcs_world2pix(
            np.vstack((all_ra, all_dec, all_wave * 1.0E-10)).T, 0)
        hdr = wcs.to_header()

    # Find the NGP coordinates for all input pixels
    msgs.info("Generating data cube")
    bins = (xbins, ybins, spec_bins)
    datacube, edges = np.histogramdd(pix_coord,
                                     bins=bins,
                                     weights=all_sci * all_wghts)
    norm, edges = np.histogramdd(pix_coord, bins=bins, weights=all_wghts)
    norm_cube = (norm > 0) / (norm + (norm == 0))
    datacube *= norm_cube
    # Create the variance cube, including weights
    msgs.info("Generating variance cube")
    all_var = (all_ivar > 0) / (all_ivar + (all_ivar == 0))
    var_cube, edges = np.histogramdd(pix_coord,
                                     bins=bins,
                                     weights=all_var * all_wghts**2)
    var_cube *= norm_cube**2

    # Save the datacube
    debug = False
    if debug:
        datacube_resid, edges = np.histogramdd(pix_coord,
                                               bins=(xbins, ybins, spec_bins),
                                               weights=all_sci *
                                               np.sqrt(all_ivar))
        norm, edges = np.histogramdd(pix_coord, bins=(xbins, ybins, spec_bins))
        norm_cube = (norm > 0) / (norm + (norm == 0))
        outfile = "datacube_resid.fits"
        msgs.info("Saving datacube as: {0:s}".format(outfile))
        hdu = fits.PrimaryHDU((datacube_resid * norm_cube).T,
                              header=masterwcs.to_header())
        hdu.writeto(outfile, overwrite=overwrite)

    msgs.info("Saving datacube as: {0:s}".format(outfile))
    final_cube = dc_utils.DataCube(datacube.T,
                                   var_cube.T,
                                   specname,
                                   refscale=ref_scale,
                                   fluxed=cubepar['flux_calibrate'])
    final_cube.to_file(outfile, hdr=hdr, overwrite=overwrite)
Пример #6
0
    def apply_flux_calib(self,
                         sens_dict,
                         exptime,
                         telluric_correct=False,
                         extinct_correct=False,
                         airmass=None,
                         longitude=None,
                         latitude=None):
        """
        Apply a sensitivity function to our spectrum

        FLAM, FLAM_SIG, and FLAM_IVAR are generated

        Args:
            sens_dict (dict):
                Sens Function dict
            exptime (float):
            telluric_correct:
            extinct_correct:
            airmass (float, optional):
            longitude (float, optional):
                longitude in degree for observatory
            latitude:
                latitude in degree for observatory
                Used for extinction correction

        """
        # Loop on extraction modes
        for attr in ['BOX', 'OPT']:
            if attr + '_WAVE' not in self._data.keys():
                continue
            msgs.info("Fluxing {:s} extraction for:".format(attr) +
                      msgs.newline() + "{}".format(self))
            #
            #try:
            #    wave = np.copy(np.array(extract['WAVE_GRID']))
            #except KeyError:
            wave = self[attr + '_WAVE']
            wave_sens = sens_dict['wave']
            sensfunc = sens_dict['sensfunc'].copy()

            # Did the user request a telluric correction from the same file?
            if telluric_correct and 'telluric' in sens_dict.keys():
                # This assumes there is a separate telluric key in this dict.
                telluric = sens_dict['telluric']
                msgs.info('Applying telluric correction')
                sensfunc = sensfunc * (telluric > 1e-10) / (telluric +
                                                            (telluric < 1e-10))

            sensfunc_obs = interpolate.interp1d(wave_sens,
                                                sensfunc,
                                                bounds_error=False,
                                                fill_value='extrapolate')(wave)
            if extinct_correct:
                if longitude is None or latitude is None:
                    msgs.error(
                        'You must specify longitude and latitude if we are extinction correcting'
                    )
                # Apply Extinction if optical bands
                msgs.info("Applying extinction correction")
                msgs.warn(
                    "Extinction correction applyed only if the spectra covers <10000Ang."
                )
                extinct = flux_calib.load_extinction_data(longitude, latitude)
                ext_corr = flux_calib.extinction_correction(
                    wave * units.AA, airmass, extinct)
                senstot = sensfunc_obs * ext_corr
            else:
                senstot = sensfunc_obs.copy()

            flam = self[attr + '_COUNTS'] * senstot / exptime
            flam_sig = (senstot / exptime) / (np.sqrt(
                self[attr + '_COUNTS_IVAR']))
            flam_var = self[attr + '_COUNTS_IVAR'] / (senstot / exptime)**2

            # Mask bad pixels
            msgs.info(" Masking bad pixels")
            msk = np.zeros_like(senstot).astype(bool)
            msk[senstot <= 0.] = True
            msk[self[attr + '_COUNTS_IVAR'] <= 0.] = True
            flam[msk] = 0.
            flam_sig[msk] = 0.
            flam_var[msk] = 0.

            # Finish
            self[attr + '_FLAM'] = flam
            self[attr + '_FLAM_SIG'] = flam_sig
            self[attr + '_FLAM_IVAR'] = flam_var
Пример #7
0
def coadd_cube(files, det=1, overwrite=False):
    """ Main routine to coadd spec2D files

    Args:
        files (list):
            List of all spec2D files
        det (int):
            detector
        overwrite (bool):
            Overwrite the output file, if it exists?
    """
    outfile = "datacube.fits"
    if os.path.exists(outfile) and not overwrite:
        msgs.error("Output filename already exists:" + msgs.newline() +
                   outfile)
    # prep
    numfiles = len(files)
    combine = True if numfiles > 1 else False

    all_ra, all_dec, all_wave = np.array([]), np.array([]), np.array([])
    all_sci, all_ivar, all_idx, all_wghts = np.array([]), np.array(
        []), np.array([]), np.array([])
    all_wcs = []
    dspat = None  # binning size on the sky (in arcsec)
    ref_scale = None  # This will be used to correct relative scaling among the various input frames
    wave_ref = None
    weights = np.ones(numfiles)  # Weights to use when combining cubes
    for ff, fil in enumerate(files):
        # Load it up
        spec2DObj = spec2dobj.Spec2DObj.from_file(fil, det)

        # Load the spectrograph
        specname = spec2DObj.head0['SPECTROG']
        spec = load_spectrograph(specname)
        detector = spec2DObj.detector

        # Setup for PypeIt imports
        msgs.reset(verbosity=2)

        if ref_scale is None:
            ref_scale = spec2DObj.scaleimg.copy()
        # Extract the information
        sciimg = (spec2DObj.sciimg - spec2DObj.skymodel) * (
            ref_scale / spec2DObj.scaleimg
        )  # Subtract sky and apply relative sky
        ivar = spec2DObj.ivarraw * (ref_scale / spec2DObj.scaleimg)**2
        waveimg = spec2DObj.waveimg
        bpmmask = spec2DObj.bpmmask

        # Grab the slit edges
        slits = spec2DObj.slits

        wave0 = waveimg[waveimg != 0.0].min()
        diff = waveimg[1:, :] - waveimg[:-1, :]
        dwv = float(np.median(diff[diff != 0.0]))
        msgs.info(
            "Using wavelength solution: wave0={0:.3f}, dispersion={1:.3f} Angstrom/pixel"
            .format(wave0, dwv))

        msgs.info("Constructing slit image")
        slitid_img_init = slits.slit_img(pad=0,
                                         initial=True,
                                         flexure=spec2DObj.sci_spat_flexure)
        onslit_gpm = (slitid_img_init > 0) & (bpmmask == 0)

        # Grab the WCS of this frame
        wcs = spec.get_wcs(spec2DObj.head0, slits, detector.platescale, wave0,
                           dwv)
        all_wcs.append(copy.deepcopy(wcs))

        # Find the largest spatial scale of all images being combined
        # TODO :: probably need to put this in the DetectorContainer
        pxscl = detector.platescale * parse.parse_binning(
            detector.binning)[1] / 3600.0  # This should be degrees/pixel
        slscl = spec.get_meta_value([spec2DObj.head0], 'slitwid')
        if dspat is None:
            dspat = max(pxscl, slscl)
        elif max(pxscl, slscl) > dspat:
            dspat = max(pxscl, slscl)

        # Generate an RA/DEC image
        msgs.info("Generating RA/DEC image")
        raimg, decimg, minmax = slits.get_radec_image(
            wcs, initial=True, flexure=spec2DObj.sci_spat_flexure)

        # Perform the DAR correction
        if wave_ref is None:
            wave_ref = 0.5 * (np.min(waveimg[onslit_gpm]) +
                              np.max(waveimg[onslit_gpm]))
        # Get DAR parameters
        raval = spec.get_meta_value([spec2DObj.head0], 'ra')
        decval = spec.get_meta_value([spec2DObj.head0], 'dec')
        obstime = spec.get_meta_value([spec2DObj.head0], 'obstime')
        pressure = spec.get_meta_value([spec2DObj.head0], 'pressure')
        temperature = spec.get_meta_value([spec2DObj.head0], 'temperature')
        rel_humidity = spec.get_meta_value([spec2DObj.head0], 'humidity')
        coord = SkyCoord(raval, decval, unit=(units.deg, units.deg))
        location = spec.location  # TODO :: spec.location should probably end up in the TelescopePar (spec.telescope.location)
        ra_corr, dec_corr = dc_utils.dar_correction(waveimg[onslit_gpm],
                                                    coord,
                                                    obstime,
                                                    location,
                                                    pressure,
                                                    temperature,
                                                    rel_humidity,
                                                    wave_ref=wave_ref)
        raimg[onslit_gpm] += ra_corr
        decimg[onslit_gpm] += dec_corr

        # Get copies of arrays to be saved
        wave_ext = waveimg[onslit_gpm].copy()
        flux_ext = sciimg[onslit_gpm].copy()
        ivar_ext = ivar[onslit_gpm].copy()

        # Perform extinction correction
        msgs.info("Applying extinction correction")
        longitude = spec.telescope['longitude']
        latitude = spec.telescope['latitude']
        airmass = spec2DObj.head0[spec.meta['airmass']['card']]
        extinct = load_extinction_data(longitude, latitude)
        # extinction_correction requires the wavelength is sorted
        wvsrt = np.argsort(wave_ext)
        ext_corr = extinction_correction(wave_ext[wvsrt] * units.AA, airmass,
                                         extinct)
        # Correct for extinction
        flux_sav = flux_ext[wvsrt] * ext_corr
        ivar_sav = ivar_ext[wvsrt] / ext_corr**2
        # sort back to the original ordering
        resrt = np.argsort(wvsrt)

        # Calculate the weights relative to the zeroth cube
        if ff != 0:
            weights[ff] = np.median(flux_sav[resrt] *
                                    np.sqrt(ivar_sav[resrt]))**2

        # Store the information
        numpix = raimg[onslit_gpm].size
        all_ra = np.append(all_ra, raimg[onslit_gpm].copy())
        all_dec = np.append(all_dec, decimg[onslit_gpm].copy())
        all_wave = np.append(all_wave, wave_ext.copy())
        all_sci = np.append(all_sci, flux_sav[resrt].copy())
        all_ivar = np.append(all_ivar, ivar_sav[resrt].copy())
        all_idx = np.append(all_idx, ff * np.ones(numpix))
        all_wghts = np.append(all_wghts, weights[ff] * np.ones(numpix))

    # Grab cos(dec) for convenience
    cosdec = np.cos(np.mean(all_dec) * np.pi / 180.0)

    # Register spatial offsets between all frames if several frames are being combined
    if combine:
        # Generate white light images
        whitelight_imgs, _ = dc_utils.make_whitelight(all_ra,
                                                      all_dec,
                                                      all_wave,
                                                      all_sci,
                                                      all_wghts,
                                                      all_idx,
                                                      dspat,
                                                      numfiles=numfiles)

        # ref_idx will be the index of the cube with the highest S/N
        ref_idx = np.argmax(weights)
        msgs.info(
            "Calculating the relative spatial translation of each cube (reference cube = {0:d})"
            .format(ref_idx + 1))
        # Calculate the image offsets - check the reference is a zero shift
        ra_shift_ref, dec_shift_ref = calculate_image_offset(
            whitelight_imgs[:, :, ref_idx], whitelight_imgs[:, :, ref_idx])
        for ff in range(numfiles):
            # Don't correlate the reference image with itself
            if ff == ref_idx:
                continue
            # Calculate the shift
            ra_shift, dec_shift = calculate_image_offset(
                whitelight_imgs[:, :, ff], whitelight_imgs[:, :, ref_idx])
            # Convert to reference
            ra_shift -= ra_shift_ref
            dec_shift -= dec_shift_ref
            # Convert pixel shift to degress shift
            ra_shift *= dspat / cosdec
            dec_shift *= dspat
            msgs.info(
                "Image shift of cube {0:d}: RA, DEC (arcsec) = {1:+0.3f}, {2:+0.3f}"
                .format(ff + 1, ra_shift * 3600.0, dec_shift * 3600.0))
            # Apply the shift
            all_ra[all_idx == ff] += ra_shift
            all_dec[all_idx == ff] += dec_shift

        # Calculate the relative spectral weights of all pixels
        all_wghts = dc_utils.compute_weights(all_ra,
                                             all_dec,
                                             all_wave,
                                             all_sci,
                                             all_ivar,
                                             all_wghts,
                                             all_idx,
                                             dspat,
                                             dwv,
                                             numfiles=numfiles)

    # Generate a master WCS to register all frames
    coord_min = [np.min(all_ra), np.min(all_dec), np.min(all_wave)]
    coord_dlt = [dspat, dspat, dwv]
    masterwcs = dc_utils.generate_masterWCS(coord_min, coord_dlt)

    # Generate the output binning
    if combine:
        numra = int((np.max(all_ra) - np.min(all_ra)) * cosdec / dspat)
        numdec = int((np.max(all_dec) - np.min(all_dec)) / dspat)
        numwav = int((np.max(all_wave) - np.min(all_wave)) / dwv)
        xbins = np.arange(1 + numra) - 0.5
        ybins = np.arange(1 + numdec) - 0.5
        spec_bins = np.arange(1 + numwav) - 0.5
    else:
        slitlength = int(
            np.round(
                np.median(slits.get_slitlengths(initial=True, median=True))))
        numwav = int((np.max(waveimg) - wave0) / dwv)
        xbins, ybins, spec_bins = spec.get_datacube_bins(
            slitlength, minmax, numwav)

    # Make the cube
    msgs.info("Generating pixel coordinates")
    if combine:
        pix_coord = masterwcs.wcs_world2pix(all_ra, all_dec,
                                            all_wave * 1.0E-10, 0)
        hdr = masterwcs.to_header()
    else:
        pix_coord = wcs.wcs_world2pix(
            np.vstack((all_ra, all_dec, all_wave * 1.0E-10)).T, 0)
        hdr = wcs.to_header()

    # Find the NGP coordinates for all input pixels
    msgs.info("Generating data cube")
    bins = (xbins, ybins, spec_bins)
    datacube, edges = np.histogramdd(pix_coord,
                                     bins=bins,
                                     weights=all_sci * all_wghts)
    norm, edges = np.histogramdd(pix_coord, bins=bins, weights=all_wghts)
    norm_cube = (norm > 0) / (norm + (norm == 0))
    datacube *= norm_cube
    # Create the variance cube, including weights
    msgs.info("Generating variance cube")
    all_var = (all_ivar > 0) / (all_ivar + (all_ivar == 0))
    var_cube, edges = np.histogramdd(pix_coord,
                                     bins=bins,
                                     weights=all_var * all_wghts**2)
    var_cube *= norm_cube**2

    # Save the datacube
    debug = False
    if debug:
        datacube_resid, edges = np.histogramdd(pix_coord,
                                               bins=(xbins, ybins, spec_bins),
                                               weights=all_sci *
                                               np.sqrt(all_ivar))
        norm, edges = np.histogramdd(pix_coord, bins=(xbins, ybins, spec_bins))
        norm_cube = (norm > 0) / (norm + (norm == 0))
        outfile = "datacube_resid.fits"
        msgs.info("Saving datacube as: {0:s}".format(outfile))
        hdu = fits.PrimaryHDU((datacube_resid * norm_cube).T,
                              header=masterwcs.to_header())
        hdu.writeto(outfile, overwrite=overwrite)

    msgs.info("Saving datacube as: {0:s}".format(outfile))
    primary_hdu = fits.PrimaryHDU(header=spec2DObj.head0)
    sci_hdu = fits.ImageHDU(datacube.T, name="scicube", header=hdr)
    var_hdu = fits.ImageHDU(var_cube.T, name="varcube", header=hdr)
    hdulist = fits.HDUList([primary_hdu, sci_hdu, var_hdu])
    hdulist.writeto(outfile, overwrite=overwrite)