2017 update of the Flux calibration of the UVOT grisms

UV Clocked Grism

The flux below about :math:`1900~AA` in the UV clocked grism has been found to show a larger rate of decrease in sensitivity than the 1%/year seen at longer wavelengths.

A preliminary correction has been prepared using observations of the calibration source WD1057+719. The observations used were only available for the years 2005 and 2017, with the spectrum either at the default (source at boresight) position on the detector, or at an offset of about (xdet=0 arcmin,ydet=5.6 arcmin).

The data at the offset position were only for 2017 and were compared to the flux calibration as published in Kuin et al. (2015) .

Analysis Details

Observations used for the analysis

date-obs/time	tstart	exposure	filter	aspcorr	filename+ext	roll
2005-09-29T05:19	149663964	1395.4	UGRISM	GRASPC	sw00055200014ugu_dt.img+1	151.7
2005-09-29T06:55	149669723	1455.0	UGRISM	GRASPC	sw00055200014ugu_dt.img+2	151.7
2005-09-29T08:42	149676143	805.1	UGRISM	GRASPC	sw00055200014ugu_dt.img+3	151.7
2005-09-29T10:11	149681484	1217.3	UGRISM	GRASPC	sw00055200014ugu_dt.img+4	151.7
2017-03-20T09:47	511696043	101.1	UVW1	DIRECT	sw00055200034uw1_sk.img+1	350.0
2017-03-20T09:49	511696154	1010.6	UGRISM	GRASPC	sw00055200034ugu_dt.img+1	350.0
2017-03-20T10:06	511697189	90.0	UVW1	DIRECT	sw00055200034uw1_sk.img+2	350.0
2017-11-22T22:32	533082773	71.6	UVW1	DIRECT	sw00055211004uw1_sk.img+1	105.0
2017-11-22T22:34	533082852	720.2	UGRISM	GRASPC	sw00055211004ugu_dt.img+1	105.0
2017-11-22T22:46	533083591	87.8	UVW1	DIRECT	sw00055211004uw1_sk.img+2	105.0

The 2005 data were summed after extracting the spectra. No wavelength shift was applied. The data are consistent with no sensitivity loss, as expected.

The spectra taken in 2017 are noisy, so an 11-point boxcar smoothing was applied. There is a small offset between the result around 1800 A with the sensitivity loss slightly larger at the offset position. However, there is some concern that weak sources in the background may be present, and there is significant noise. For the preliminary calibration update a smooth curve was created that lies between those two cases. Near :math:`1700~AA` the extra sensitivity loss is near to 80%. At longer wavelength, the sensitivity loss seems to have been only ~0.5%, but this needs confirmation by longer exposures. This update applies an approximate fit below 3000A.

figure 1: ratio of expected to measured flux

Extra sensitivity loss above 1%/year The flux extracted using UVOTPY after applying the coincidence loss and 1%/yr sensitivity loss correction, divided by the STIS spectrum of WD1057+719.

Results

• The 1%/year rate of sensitivity loss holds for wavelengths 2000-2750 A.
• Below 1800 A there is a progressive loss in sensitivity up to ~80%.

Caveats

The analysis described here leads to the following caveats for this preliminary calibration update:

• The calibration data did not include data of intermediate years. It is therefore possible that the sensitivity change happened not linearly in time, as has been assumed for the correction, nor do we know if the sensitivity change was happening at all wavelengths from the starts or if it started at the short wavelengths and then creeped up to longer wavelengths.
• The data used were noisy, and there is a hint that the profile at the offset could be different from that at the centre position. The current solution smooths over the differences and noise.
• The calibration source is only good for wavelengths below 2750 A.

UVOTPY implementation

In the “UVOTPY” the module “UVOTIO” has been modified to include an array with the linear excess rate below 3000A.:

if uvotgetspec.senscorr:
    sens_corr = 1.0/(1.0 - sens_rate*(swifttime-126230400.000)/31556952.0 )
    if wave is not None and (wheelpos == 160):
       fscale = (swifttime-126230400.000) / (12.6*365.26*86400)
       extracorr = np.array(
   [[1.650, 0.19],
    [  1.68810484e+03,   1.93506494e-01],
    [  1.70579637e+03,   3.42857143e-01],
    [  1.72757056e+03,   5.52813853e-01],
    [  1.75206653e+03,   6.93506494e-01],
    [  1.79153226e+03,   7.93073593e-01],
    [  1.82419355e+03,   8.58008658e-01],
    [  1.89223790e+03,   9.05627706e-01],
    [  1.96436492e+03,   9.55411255e-01],
    [  2.04465726e+03,   9.98701299e-01],
    [  2.10317540e+03,   1.00735931e+00],
    [  2.17121976e+03,   1.02900433e+00],
    [  2.25423387e+03,   1.04415584e+00],
    [  2.32227823e+03,   1.05281385e+00],
    [  2.42026210e+03,   1.08961039e+00],
    [  2.52096774e+03,   1.09177489e+00],
    [  2.60398185e+03,   1.09177489e+00],
    [  2.72101815e+03,   1.09610390e+00],
    [  2.77953629e+03,   1.07445887e+00],
    [  2.83805444e+03,   1.07229437e+00],
    [  2.90745968e+03,   1.05714286e+00],
    [  2.96597782e+03,   1.05281385e+00],
    [  2.99863911e+03,   1.02900433e+00],
    [3000,1.],[8000,1.]])
       f_extrasenscorr=interp1d(extracorr[:,0],extracorr[:,1],
       bounds_error=False,fill_value=1.0)
       sens_corr=sens_corr/(f_extrasenscorr(wave) * fscale)
       print ("\nsensitivityCorrection: applied additional changes in UV 1700-3000\n")