This document compares the predicted performance of the Gemini Near-infrared Integral Field Spectrograph (NIFS) to the actual performance of similar near-infrared spectrographs in operation or planned for large telescopes.

2 Applicable Documents

Document ID	Source	Title
SDN0004.01	RSAA	NIFS Performance Model

3 Introduction

The performance of the Gemini Near-infrared Integral Field Spectrograph (NIFS) has been extensively modeled (NIFS Performance Model, SDN0004.01). It was noted at the NIFS Conceptual Design Review that these performance predictions were significantly worse than some claims for the performance of NIRSPEC on Keck. This prompted the NIFS review committee to recommend that more detailed comparisons be made between the predicted performance of NIFS and the actual performance of other near-infrared spectrographs on 8-10 m class telescopes. The results of these comparisons are reported below.

4 Comparison with NIRSPEC

The NIFS performance predictions were originally compared to NIRSPEC performance figures obtained from the NIRSPEC User’s Manual. These limiting magnitude per resolution element estimates for the NIRSPEC low resolution mode (R = 2200 through 0.4² slit) in 3600 s for a signal-to-noise ratio of 10 are listed in Table 1.

Table 1: NIRSPEC User’s Manual Limiting Magnitudes (10:1 in 1 hr)

Wavelength (mm)	R=2000(OH)	R=2000
1.25	20.2	22.2
1.65	19.4	21.5
1.8	19.2	21.3
2.0	19.4	21.1
2.2	19.1	20.1
2.4	18.6	19.9

NIRSPEC performance predictions are also available on the Gemini web pages. These limiting magnitudes and line fluxes for the NIRSPEC low resolution mode on point sources with 0.5² seeing and a 3´3 pixel (0.57²´0.57²) slit for a signal-to-noise ratio of 3 in 30 min are listed in Table 2.

Table 2: Gemini NIRSPEC Limiting Magnitudes (3:1 in 0.5 hr)

Wavelength (mm)	Magnitude	Line Flux (W m^-2)
1.00	19.0	7.5´10^-20
1.25	19.5	3.3´10^-20
1.65	18.5	2.4´10^-20
2.2	18.5	1.8´10^-20

A description of NIRSPEC by Ian McLean (19 August 1999) is available on the UCLA NIRSPEC web page. This quotes a signal-to-noise ratio of 10 in 600 s at H = 16.5 mag for the low resolution mode which is read noise limited. If a 0.5 hr total integration time is comprised on three 600 s exposures, NIRSPEC should achieve a signal-to-noise ratio of 3 in 30 min at H = 18.4 mag based on this estimate. This agrees well with the Gemini value (Table 2). However, a limiting magnitude of H = 17.5 mag (10:1 in 1 hr) is suggested, if the 1 hr total exposure is comprised of six 600 s exposures. This is ~ 2 mag worse than claimed in the NIRSPEC User’s Manual. We therefore adopt the Gemini values as representative of the current performance of NIRSPEC.

Comparison with NIFS requires scaling for the different primary mirror diameters, the different spectral resolution, and the different spatial resolution. Using NIRSPEC on Gemini would degrade signal-to-noise by a factor of (10/8)² = 1.56 because NIRSPEC is read noise limited. Similarly, operating at R = 5300 would degrade signal-to-noise ratio by a further factor of 5300/2000 = 2.65. The better image quality achieved with ALTAIR allows NIFS to use narrower slitlets than NIRSPEC. However, this has no effect on signal-to-noise ratio because NIRSPEC remains read noise limited. We therefore degrade the quoted NIRSPEC signal-to-noise ratios by a factor of 1.56*2.65 = 4.14 to permit a direct comparison with NIFS.

The direct performance comparison of NIFS and NIRSPEC is made in Table 3. This lists point source limiting magnitudes for a signal-to-noise ratio of 3:1 in 0.5 hr on Gemini with R = 5300. The NIFS values are derived from those quoted at CoDR (NIFS Performance Model, SDN0004.01). Based on this comparison, it is clear that the predicted performance of NIFS is much better than the actual performance of NIRSPEC if used on Gemini with R = 5300. This is as expected since the NIRSPEC detector is significantly inferior to the assumed parameters for the NIFS detector and NIFS is matched to the excellent point source images to be delivered by ALTAIR.

Table 3: Comparison of NIFS and NIRSPEC Limiting Magnitudes (3:1 in 0.5 hr)

Wavelength (mm)	NIFS Magnitude	NIRSPEC Magnitude
1.00	20.4	17.5
1.25	19.9	18.0
1.65	20.4	17.0
2.2	19.1	17.0

5 Comparison with ISAAC

The performance of ISAAC on the VLT has been determined using the web-based ISAAC Exposure Time Calculator. The medium resolution grating with a 0.6² slit produces resolving powers of R ~ 5000. The same size of the VLT and Gemini primary mirrors and the similar spectral resolutions allow a direct performance comparison (ignoring ALTAIR). This comparison is presented in Table 4. The predicted performance of NIFS is better than the actual performance of ISAAC. This is also to be expected because ISAAC uses its gratings in high orders which are likely to be less efficient than the first order gratings proposed for NIFS, and ISAAC should suffer from additional background noise in this comparison which ignores the different slit widths of NIFS and ISAAC. The ISAAC model also assumes a telescope emissivity of 25% for the VLT which exceeds the combined assumed emissivities of Gemini and ALTAIR. This accounts for the poorer actual performance at K. In fact, ISAAC and NIRSPEC appear to have similar performance at K.

Table 4: Comparison of NIFS and ISAAC Limiting Magnitudes (3:1 in 0.5 hr)

Wavelength (mm)	NIFS Magnitude	ISAAC Magnitude
1.00	20.4	18.4
1.25	19.9	19.6
1.65	20.4	19.3
2.2	19.1	17.0

6 Comparison with CIRPASS

CIRPASS uses an integral field unit in the J and H bands that can be configured to have a lenslet diameter of 0.12², similar to the NIFS slitlet width of 0.10². The spectral resolving power is R = 2840 at 1.25 mm and R = 3750 at 1.65 mm. The web-based CIRPASS Exposure Time Calculator has been used to quantify the CIRPASS performance. A seeing FWHM of 0.1² was assumed to approximate the effect of ALTAIR. The performance comparison with CIRPASS is presented in Table 5. NIFS (with R = 5280) and CIRPASS (with R = 3750) have almost identical predicted performance at H. The origin of the poorer predicted performance for NIFS at J is unclear. It may be related to the larger spectral resolution difference at J; R = 6040 for NIFS and R = 2840 for CIRPASS. If CIRPASS is detector noise limited at J, doubling the spectral resolving power would brighten the limiting magnitude by 0.75 mag changing the limit only to J ~ 21.0 mag.

Table 5: Comparison of NIFS and CIRPASS Limiting Magnitudes (3:1 in 0.5 hr)

Wavelength

(mm)

NIFS

Magnitude

CIRPASS

Magnitude

1.25

19.9

21.7

1.65

20.4

20.5

Comparison with CGS4

CGS4 sensitivities have been taken from the UKIRT CGS4 web pages. The data are based on the 150 l/mm grating, the long camera, and are appropriate for point sources. The spectral resolving power with this arrangement is comparable to that achieved by NIFS. A correction of a factor of 4 (1.5 mag), with large uncertainty, has been applied to the published CGS4 sensitivities to allow for the larger Gemini collecting area; source signal will be larger by a factor of ~ 4 on Gemini, and the background might be unaffected if the CGS4 slit width is notionally reduced to match that of NIFS. After this correction, the intrinsic predicted sensitivity of NIFS is within ~ 1 mag of the measured CGS4 sensitivity.

Table 6: Comparison of NIFS and CGS4 Limiting Magnitudes (3:1 in 0.5 hr)

Wavelength (mm)	NIFS Magnitude	CGS4 Magnitude
1.25	19.9	19.9
1.65	20.4	19.1
2.2	19.1	18.1

7 Summary

Within the considerable uncertainties associated with predicting the performance of a complex instrument from first principles, it appear that the performance of NIFS calculated by NIFSSIM (NIFS Performance Model, SDN0004.01) is consistent with the measured or predicted performance of other near-infrared spectrographs on large telescopes.