Notes for Observing with the UVOT Visible Grism

Last updated: 27-July-2016: 2015 calibration & software

The UVOT Visible grism can provide 2850 - 6500 Å spectra of modest S/N and spectral resolution R ~ 75 for stars in the magnitude range 13-17. Note that the V grism spectrum longward of 5200 Å can be confused by overlap from second-order light. Note that the V grism is more sensitive than the UV grism for the longest wavelength UV light (2900-3200 Å).

The following decisions need to be made when observing with the V grism:

  • Slew in place?

The pointing accuracy after an initial slew may be only 1 or 2 arc minutes! The use of a second slew after the arrival at the target (a “slew in place”) can reduce the pointing uncertainty to ~15 arc seconds. Unlike the UV grism, the sensitivity and wavelength calibration of the V grism do not have a large dependence on the detector position though the 2015 calibration Kuin et al. (2015) corrects for that. A large offset may place part of the spectrum off of the detector, or (for a clocked observation) in a vignetted region. Note that generally the Swift UVOT planners try to avoid observing with the grism during the first few and last few minutes of an orbit anyway (because the sky background is large) so the use of a slew in place does not affect the prime grism science time.

  • Clocked or Nominal Mode?

Zeroth-order spectra only appear in the area covered by the grism, whereas first-order spectra can be dispersed off the edge of the grism image. Thus, the use of clocked mode can minimise the contamination of first-order spectra by unrelated zeroth-order images. Clocked mode also suppresses the total sky flux, and for that reason must be used in crowded fields (e.g. low Galactic latitudes). One disadvantage of clocked mode is that it introduces vignetting and so spectra are retrieved over a smaller field, which may be important for extended sources or if a slew in place is not used. For sparse fields, the choice between clocked and nominal mode is often decided by which one gives a better roll angle for a particular observing date.

  • Roll Angle

The contamination of grism spectra by zero and first order of unrelated sources depends on the roll angle of the observation. An IDL grism simulator is available at which displays a DSS image with simulated dispersed spectra for a given roll angle. (This software can be run under the IDL virtual machine, without an IDL license.) For a given observing time, the planners have very limited flexibility in choosing the roll angle. However, the simulator can be used to select a range of good observing dates or to choose between clocked and nominal model.

  • Exposure time

The effective area (no coincidence loss) for the V grism is given in the figure below, see Kuin et al. (2015). Note that this effective area is for computing the count rate integrated perpendicular to the dispersion, and that the dispersion varies from about 5.0 Å/pixel near 2800 Å to about 6.5 Å/pixel near 5300 Å. The “Cts/s” column gives the predicted counts/s (per pixel integrated perpendicular to the dispersion) for a source with a flat spectrum of 10^(-14) erg cm(-2) s(-1) Å(-1).

Effective Area

The effective area for all four grism modes is plotted below. Note that although the UV grism can record a visible spectrum, the sensitivity is less than half that of the V grism; can suffer significant contamination from second-order UV light; and at long wavelengths has a very low resolution due to the extended PSF.


Figure: The effective areas of the grisms at the default position.


Some considerations to be aware of when using the V(isible) grism.

1 A single grism spectrum shows a fixed pattern noise that is only partially suppressed by applying a mod-8 correction which is due to the coincidence loss that cannot be corrected for that way. It may be useful to combine spectra with shifted zero order positions to further suppress this fixed-pattern noise.

2 Coincidence loss is corrected for when using the 2014 UVOTPY software which implements the 2015 calibration.

3 The wavelength anchor point is defined by the centroid of the zeroth order in the old approach (using uvotimgrism), and at 4200 Å in first order in the UVOTPY software. The UVOTPY software will indirectly use the zeroth orders for determining the anchor point when no lenticular filter observation was taken along with the grism exposure by updating the aspect solution. Note that because the zeroth orders are dispersed the position of the anchor point may vary with spectral shape. In addition, mod-8 noise may shift the position of the centroid. Finally, the zeroth order is often saturated. So there may be a zero point shift of up to 2-4 pixels ( 20 Å).

Example Spectra Four examples of UVOT V grism spectra are shown below. The first shows a UVOT spectrum of NGC 5548 compared with a HST/FOS near-UV spectrum, and optical spectrophotometry from Kennicutt. The second compares the V=13.8 hot white dwarf GD 50 with combined optical spectrophotometry and IUE observations. (Note that the IUE spectrum does not appear to smoothly connect with the optical spectrophotometry at 3200 Å.) The third examples compares a UVOT spectrums of the hot white dwarf BPM 16274 with a model atmosphere. The last spectrum compares UVOT spectra of the V=13.7 solar analog P177D with a STIS spectrum.


figure: examples of spectra taken with the visible grism (using uvotimgrism).