Further information

Questions regarding this page should be directed to Kentaro Aoki ( ).

Measured system throughput


There are a few restrictions applid in the operation of HR. Applicants/observers should go to this page and keep the notes there in mind about the use of HR.

Fig. 1 shows the measured system throughput as a function of wavelength in the High Resolution (HR) mode (red line) and Low Resolution (LR) more (blue line) (Kimura, M., et al. 2010, PASJ, 62, 1135). Note that due to the effects of the OH masks, there are a number of wavelengths with low throughput looking like absorption lines. In fact, there is a fiber-to-fiber variation in the throughput (about ~3 %), and the median values among the fibers are plotted over the spectral coverage in this figure.

This throughput was measured by illuminating the Echidna fibers on the prime focal plane through the wide-field corrector lenses with a calibrated black-body radiation source and taking spectra with IRS1. The result was then multiplied by 0.85 to consider the reflectivity of the telescope main mirror. (The throughput of IRS2 has also been recently measured by observing stars and galaxies with J & H magnitudes known, suggesting that it is consistent with IRS1.)

For the features numbered in the plot, brief explanations are given as below:

  • (1): The peak of diffraction efficiency of the primary grating in the spectrographs (see details here for the optics layout) is around 1.35um. Hence the throughput decreases rather rapidly at shorter wavelengths, instead of a higher throughput kept over the H band.
  • (2): This decrease corresponds to the wing of the strong absorption feature of fused silica at 1.4um which are used for the correctors in the spectrographs.
  • (3): The edge of the passband of the thermal blocking filter in the camera dewar is visible here.
  • (4): When the high resolution spectra are formed on the mask mirror, the 1.35um-1.40um part has to be blocked by the fiber slit and is therefore not available on the detector.
  • (5)It is emphasized that the throughput is lower in LR than in HR: A VPH grating is added in the LR mode to anti-disperse the OH-suppressed high resolution spectra and allow the entire spectral coverage (0.9 μm to 1.8 μm) to be covered by the detector at once. Therefore the LR system throughput is approximately (HR throughput) x (VPH throughput).

Fig. 1 - Measured system throughput in the HR mode (red line) and LR mode (blue line). From Kimura, M., et al. (2010).

ATTENTION: Fig. 1 was updated in Aug 2010 and the throughput value plotted there is lower in both LR and HR than in the same plot having been indicated until then. This is purely because there was a bug-fix in the calculation of the throughput (a wrong temperature, 1000 K instead of 1090 K, was used before as that of the black body radiation source), NOT because the actual instrument throughput became lower than the previous semester. Consequently, there has been no change in the guideline of sensitivity.
Go to the top

Guideline of sensitivity

Guidelines of continuum level and emission line flux to achieve S/N=5 detection with 1 hour on-source integration time are presented below. Please pay attention to the following notes that are valid both to LR and HR in applying this information to your observation:

  • For continuum, S/N was measured after applying four pixel binning to spectra (cf. the FWHM of a single emission line (i.e. ~one spectral resolution element) is 4-5 pixel). Most of the data used to derive the sensitivity are actually from faint galaxies, and the magnitudes referred here are their total magnitudes (mostly MAG_AUTO from SExtractor).
  • For emission line, S/N was measured per emission line: It was calculated by integrating the line profile and comparing it with the noise level expected in the corresponding spectral element. This sensitivity estimation was performed using the spectra of galaxies, not AGN/QSO. Observers therefore need to be careful when applying this number to such objects having broad lines.
  • The emission line flux in the tables below were derived assuming that an object is a point source, i.e., most (~80%) of the energy falls on the an Echidna fiber on the focal plane.
  • The numbers in the following tables are the averages in the corresponding bands. Observers need to be careful when applying them to features near the edge of each band, especially 0.9-1.0 μm, 1.40-1.45 μm, and 1.75-1.80 μm, where the instrument throughput is very low (see Fig. 1).
  • These sensitivity information is based on the data that were taken by employing the cross-beam switching (CBS) mode. We will investigate how much difference is given to the final data quality between normal beam switching (NBS) and CBS with the same on-source integration time.
We have still been in the process of collecting systematic on-sky data for better performance verification (especially in HR). We should therefore mention that the sensitivity information below are still preliminary and is subject to change in future.

For observers to investigate more detailed feasibilities of their science targets as a function of wevelength, FMOS spectrum simulator is also available.

  1. Low Resolution mode (LR)
  2. The sensitivity varies as a function of wavelength, mainly according to the variation of instrument throughput. Hence we divide the spectral coverage (0.9-1.8 μm) into four bands like in the High Resolution mode and list the limiting magnitude (flux) for continuum (emission line) within each band, respectively.

    Wavelength Resolution Continuum level Emission line flux
    [μm] [R=λ/Δ λ] [AB mag] [erg cm-2 s-1]
    0.92-1.12 470 18.1 6.2 x 10-16
    1.09-1.35 550 20.1 1.0 x 10-16
    1.40-1.66 690 19.8 1.0 x 10-16
    1.54-1.80 760 19.1 1.9 x 10-16

    Tab. 1 - Resolving power and sensivity to contiuum and emission line in the Low Resolution (LR) mode.

  3. High Resolution mode (HR)
  4. While the higher spectral resolution in HR tends to allow a smaller number of photons to arrive at each pixel than LR, this is compensated to some extent by its higher throughput (see Fig. 1). If the high resolution is essential e.g. to see a line profile, only ~4-5 pixels should be binned at most and the sensitivity (per pixel and per spectral resolution element) should be lower in HR. Meanwhile, if the main goal is e.g. the detection of emission line(s) and measurement of total line flux(es), a number of pixels could be binned and more benefit would be expected from the higher throughput (at the expense of spectral resolution instead), thanks mainly to the negligible contribution of readout noise when the ramp sampling method is applied.

    Below we list the sensitivity information to achieve S/N=5 with 1 hour on-source integration in HR, after 4 pixel binning for continuum and per emission line for emission line. As of Jan 2011, we have been focusing more on investigating the sensitivity of emission lines, so we are still short of real data to estimate the sensitivity to continuum. Therefore, we estimated it using (1) the noise level measured on the actual HR spectra and (2) an assumption that the sensitivity of the instrument in HR is twice as high as in LR based on the comparison of real data for emission lines (see the next section).

    Band name Coverage Resolution Continuum level Emission line flux
    [μm] [R=λ/Δ λ] [AB mag] [erg cm-2 s-1]
    J-short 0.92-1.12 1600 18.0 1.7 x 10-16
    J-long 1.11-1.35 1900 19.1 0.4 x 10-16
    H-short 1.40-1.60 2400 19.0 0.5 x 10-16
    H-short prime 1.45-1.67 2400 19.1 0.5 x 10-16
    H-long 1.60-1.80 2600 19.1 0.5 x 10-16

    Tab. 2 - Spectral coverage, resolving power, and sensivity to contiuum and emission line in the High Resolution (HR) mode.

  5. Correlation between emission line flux and S/N
  6. We detected a number of emission lines of faint galaxies at various wavelengths during the engineering observations and GTOs. We estimated the S/Ns (per emission line) of those detections and normalized them to the values expected for 1 hr on-source integration. In Fig. 2, these normalized S/Ns per emission line are plotted against emission line fluxes (after the apergure effect at the fiber entrance for a point source is applied): The data from LR mode, J-long in HR, and H-long in HR are indicated by green, blue and red squares, respectively. This suggests, although there is significant scatter especially in the LR data, S/N=5 (S/N=10) can be achieved by 1 hour on-source integration for an emission line with a flux of 1.0 x 10-16 erg cm-2 s-1 in LR (HR), respectively. For the LR data, a number of objects are located below the correlation but they tend to be emission lines that were observed at 0.9-1.1 μm or 1.7-1.8 μm where the instrument throughput is lower, or on OH masks.

    Fig. 2 - For emission lines of faint galaxies on the data from engineering observations and GTOs, S/Ns (for 1 hr integration time, per emission line) are plotted against emission line fluxes (after the apergure effect at the fiber entrance for a point source is applied). The data from LR mode, J-long in HR, and H-long in HR are indicated by green, blue and red squares, respectively.

Go to the top

Spectra from engineering observations

Below the spectra of a few astronomical objects are presented. The data were taken during the engineering observation in Dec 2009, with IRS1 in the LR mode. These spectra have been flux-calibrated taking the energy loss at a fiber aperture into consideration.

A low resolution spectrum of a galaxy with J(AB)~20.1 mag and H(AB)~19.7 mag. The on-source integration time is 1.5 hours in the CBS mode. The red line shows the spectrum after data reduction, calibration, and 4 pix binning. The S/N of the continuum emission (dashed line is shown for reference) is estimated to be about 5 from 1.1 μm to 1.7 μm (i.e. in J band and H band). A 2-D image of the reduced and calibrated spectrum is indicated on top of the plot.
An emission line galaxy at z=1.5. The on-source integration time is 1.5 hours in the NBS mode. 3-pixel binning is applied. The emission lines clearly detected are [OIII]5007 and Hα. A 2-D image at the top of this panel shows the spectrum of this object, where the positions of the emission lines are indicated by circles.
An AGN at z=1.35. The on-source exposure time is 1.5 hours in the CBS mode. 3-pixel binning is applied. In addition to the broad Hα emission, the narrow [OIII]5007 emission is detected with S/N~5, of which flux is estimated to be ~8 x 10-17 erg cm-2 s-1. A 2-D image at the top of this panel shows the positive and negative spectrum of this object from the CBS observation (so the net integaration is 0.5 x 1.5 hours for each) and the positions of the emission lines are indicated by circles.

Go to the top

FMOS spectrum simulator

Go to the top

Last updated: July 23, 2012

Copyrightę 2000-2011 Subaru Telescope, NAOJ. All rights reserved.