def flux_intercalibration(ob_name):
rdir=os.getenv('XDIR')+'/spectra/'+ob_name
mdir=os.getenv('XDIR')+'/molecfit/'+ob_name
flux_uvb=rdir+"/FLUX_UVB_spec.fits"
flux_vis=mdir+"/FLUX_VIS_spec_TAC.fits"
# flux_nir=mdir+"/FLUX_NIR_spec_TAC.fits"
w_uvb,f_uvb,n_uvb=unmask_spectra(flux_uvb)
w_vis,f_vis,n_vis=unmask_spectra(flux_vis, tac=True)
# w_nir,f_nir,n_nir=unmask_spectra(flux_nir, tac=True)
f_model_interpol_uvb,flux_name=interpol_fluxstd_modelspec(flux_uvb,w_uvb,"UVB")
f_model_interpol_vis,flux_name=interpol_fluxstd_modelspec(flux_vis,w_vis,"VIS")
# f_model_interpol_nir,flux_name=interpol_fluxstd_modelspec(flux_nir,w_nir,"NIR")
flux_corr_factor_uvb = f_model_interpol_uvb/f_uvb
flux_corr_factor_vis = f_model_interpol_vis/f_vis
plt.plot(w_uvb,flux_corr_factor_uvb)
plt.plot(w_vis,flux_corr_factor_vis)
plt.xlim([520,620])
m=np.median(flux_corr_factor_uvb[w_uvb > 520])
s=np.std(flux_corr_factor_uvb[w_uvb > 520])
plt.ylim([0,m+5*s])
plt.savefig(ob_name+"_UVB_VIS_intercalibration.png")
plt.clf()
def write_spec_starlight(calspec_file,outfile,z,wrange):
wave,flux,noise = unmask_spectra(calspec_file)
wave*=10 ## in Angstrom
##
## de-redshift spectrum
wave/=(1+z)
##
## check that input wavelength range is larger than output wavelength range
if (wave[0] > wrange[0]) or (wave[-1] < wrange[-1]):
raise ValueError("input wavelength range must be larger than output wavelength range")
##
## normalize flux and noise
m=np.median(flux)
flux/=m
noise/=m
##
## define output wavelength vector
wsampling=1 ## wavelength sampling in Angstrom
lamgrid=wrange[0] + wsampling * np.arange(1+np.floor((wrange[1]-wrange[0])/wsampling))
##
## convolve to resolution of spectral base, i.e. BC03 res = 3 \AA
##
## XSHOO resolution is 12600 (VIS) independent of lambda, i.e. choose kernel
## appropriate for central wavelength
xres=12600
FWHM_gal=np.average(wrange)/xres
FWHM_tem=3
FWHM_dif = np.sqrt(FWHM_tem**2 - FWHM_gal**2)
sigma = FWHM_dif/2.355/(wave[2]-wave[1]) # Sigma difference in pixels
flux=gaussian_filter1d(flux,sigma)
##
## interpolation is flux-conserving when accounting for different spectral pixel size
## dl is 0.2 in UVB and VIS, 0.6 in NIR
##
f=interp1d(wave,flux)
iflux=f(lamgrid)
##
## interpolate inverse variance (since this is the additive quantity)
##
## interpolation is not enough; need to also reduce the noise!
v=interp1d(wave,1/(noise**2))
ivariance=v(lamgrid)
inoise=1/np.sqrt(ivariance)
##
## reduce noise by 1/sqrt(N) where N is number of bins averaged over
# pdb.set_trace()
N = wsampling/np.diff(wave)
N2=np.hstack([N[0],N]) ## np.diff reduces size of array by 1
Ni = interp1d(wave,N2)
Ninterp = Ni(lamgrid)
fac=np.sqrt(Ninterp)
inoise/=fac
## masks for transition regions of XSHOOTER spectral arms
imask=np.zeros(iflux.shape)
# add_mask_telluric(lamgrid, imask, [5530,5600], z)
# add_mask_telluric(lamgrid, imask, [10100,10200], z)
## Quality Check plot
##
plt.plot(wave,flux,label="intrinsic resolution")
plt.plot(lamgrid,iflux,label="downsampled resolution")
##
## pick region around CaT if within region
CaT1=8400
CaT2=8800
if (8000 > wrange[0]) & (9000 < wrange[1]):
plt.xlim([8000,9000])
plt.ylim([0.8,1.2])
else:
plt.xlim(wrange)
plt.ylim([0,2])
plt.legend(loc=2)
outfile_qc=outfile.split('.txt')[0]+'_interpolate_QC.pdf'
plt.savefig(outfile_qc)
print("Saved QC plot to ", outfile_qc)
np.savetxt(outfile,np.transpose((lamgrid,iflux,inoise,imask)),fmt=('%5.0f. %8.3f %8.3f %3.0f'))
print("Wrote ", outfile)
