--> Sensitivity and Overheads | Gemini Observatory

Change page style: 

Sensitivity and Overheads

You are here


 Coronagraphic Mode Contrast Curves

NICI is optimized for detection of faint, sub-stellar companions of stars by utilizing the simultaneous Angular Spectral Differential Imaging (ASDI) technique.  In this technique, the Cass Rotator is held fixed so that the sky rotates at least several degrees during the entire observation, which may be an hour or more, and companions rotate against the fixed speckle pattern of the primary star.  In NICI, two wavelength channels may be observed simultaneously so that spectral differencing improves suppression of a primary star's light.  The limiting contrast of such a complex dataset is intimately tied to the data reduction and analysis algorithms.   In summary, a typical reduction pipeline would follow these basic steps:  (1) normal image reductions (flat fielded, dark subtraction, and bad pixel removal), (2) high-pass filter the image (for example removing azimuthal average/median profile), (3) generating a 'quasi-static' PSF for removal taking advantage of the ADI observing, (4) registration and subtracting the two filters to produce the SDI images, and (5) finally rotating all of the ASDI images to a common field position angle and combining to create a final science image frame.

In this section we present contrast and sensitivity information derived from NICI commissioning data processed through such an analysis pipeline.

Figs. 1 and 2 show Strehl ratio and contrast curves in ASDI mode (Cassegrain rotator off) under clear skies (CC50) and median (0.7”, or IQ50) to better than median seeing conditions.  The integration times were 30 minutes on stars with brightnesses between V=8 and V=11.   Fig. 1 shows Strehl ratios measured from NICI science images, which include all non-common path errors.  The measured data points in Fig. 2 illustrate the contrasts achieved from the occulting mask edge (0.32” radius) to 1” from the star, in the region where contrast is limited by speckles.    Note that TWA-7 is considerably fainter than the other two targets and, in addition, the number of non-destructive reads used was less than that for the other two fields.

Also shown in the figures are curves derived from simulations.   The simulations account for the static Strehl of 0.8 at 1.6 microns as measured on NICI's internal calibration source. The seeing is estimated from the AO telemetry data, and the WFS counts are per channel per wavefront sample.  NICI nominally samples the wavefront curvature at 1.3kHz.  For bright guide stars (Vgs = 11 or brighter) the measured Strehl ratios agree well with the simulated performance curves. The performance appears to saturate at SR1.6um ~ 50% under very good seeing conditions, but note that the curvature wavefront sensor has been used at only one optical gain (extra-focal distance = 0.4m) during on-sky commissioning. At these seeing values the optimal extra-focal distance determined from the simulations is smaller and from the simulations this difference would account for most of the saturation.

Outside 1 arcsecond, the sensitivity should be dominated by photon noise for long exposures with a high number of NDR. This highlights ADI's strong requirement on minimizing detector noise.  The right panel of Fig. 2 shows the contrast curves scaled to a 2-hour integration. Commissioning data  confirm that the contrast increases with the square root of the exposure time. However, this has only been examined with a small number of datasets, none longer than 45 minutes of integration.

As reference the contrast curves specified in the Gemini Science Campaign Request for Proposals (RfP) and contrast curves obtained during the Gemini Deep Planet Survey (Lafrenière et al. 2007) are shown.

Exposure times can be estimated using the contrast plot in Fig. 2, and the table of zeropoint values.  NICI does not have an ITC because of the difficulty of computing contrast limits in a variety of conditions.  

Please note two important aspects of these plots:

  1. The contrast curves shown represent the detection limits for point sources in ADI/ASDI observations processed with the LOCI algorithm; they do not represent the contrast achievable with non-ADI/ASDI modes. In the ADI/ASDI modes any structure that is circular symmetric over the arc that is covered during the sky rotation would be removed by the post processing. Therefore, structures such as circular shells and bipolar outflows that are symmetric for the particular range of sky rotation would be suppressed in the final image.
  • The plots represent good observing conditions: clear skies (CC50), median seeing (IQ50), and low airmass. In poorer seeing (IQ70-85) and thin clouds (CC70), the AO performance is degraded significantly, with Strehls dropping by factors of 2-3 or more.  Strehl ratio also drops as seeing worsens toward high airmass.  The lower performance may still be satisfactory for many applications, eg. searches for relatively bright companions, but not for deep coronagraphic imaging of faint companions. 
  • Figure 1: Bright star performance measured for median seeing conditions and better. Figure from Chun et al., 2008, SPIE.


    Figure 2: Preliminary contrast curve derived from commissioning data, from Z. Wahhaj, M. Liu, and the NICI Campaign Team. The contrast curves on the left show the achieved contrasts for the actual integration times (approximately 30 minutes) as well as the simulation/RfP contrast curve scaled to 30 minutes. The V-band brightnesses of the stars are listed in the legend. Note that TWA-7 is considerably fainter than the other two targets and the number of non-destructive reads was less than that for the other two targets. The figure on the right shows the contrast curve obtained on the two brighter targets scaled by the square root of the integration time to 2 hours. (Scaling to such long exposures has not yet been fully demonstrated during the commissioning phase.) For reference, the NICI-RfP curve and two contrast curves from the GDPS are also shown scaled to the same total integration time. Figure from Chun et al.,2008, SPIE proceedings.


    Sensitivity and Saturation Limits

    The following table contains information to estimate single-frame exposure times for several typical NICI configurations.  It also indicates the maximum possible target magnitude m_max before saturation of the infrared detectors will occur.

    The exposure time is texp = t0 * 10m / 2.5, where t0 is indicated in the table and m is the target magnitude in the chosen filter.  This time will result in the core and halo of the PSF being properly exposed at approximately 50% full well in median seeing (IQ50).  The core signal value varies strongly with seeing, and can be 2 to 4 times higher in the best conditions, so the table is intentionally conservative to avoid saturation.

    Note that sensitivities are substantially lower in thin clouds (CC70) and poorer seeing (IQ70-85) due to the lower Strehl ratio of the AO correction.  The Strehl ratio also drops steadily towards higher airmass.

    The mmax indicates the brightest magnitude observable before the infrared detectors (not the AO WFS) will saturate at the minimum 0.38 sec single-frame exposure time.  Note that there is a 0.38 sec overhead for the readout, so at texp = 0.38 sec the efficiency drops to 50%.  Longer exposure times are preferred whenever possible to maintain reasonable efficiency.


    Table 1: Recommended exposure times and maximum magnitudes.

     Mask  Dichroic  Filter  Well   t_0 (sec)
      Clear  50/50  1%  Med  9.34E-04  6.52
     Clear  50/50  4%  Med  2.34E-04  8.03
     Clear  50/50  J  Med  4.76E-05  9.75
     Clear  50/50  H  Med  2.71E-05  10.37
     Clear  50/50  K  Med  4.86E-05  9.73
     Clear  50/50  Ks  Med  8.22E-05  9.16
     Clear  50/50  Kp  Med  7.29E-05  9.29
     Clear  50/50  Lp  Med  1.6E-03  6.0
     Clear  Mirror/Open  1%  Med  4.67E-04  7.28
     Clear  Mirror/Open  4%  Med  1.17E-04  8.78
     Clear  Mirror/Open  J  Med  2.38E-05  10.51
     Clear  Mirror/Open  H  Med  1.35E-05  11.12
     Clear  Mirror/Open  K  Med  2.43E-05  10.49
     Clear  Mirror/Open  Ks  Med  4.11E-05  9.91
     Clear  Mirror/Open  Kp  Med  3.64E-05  10.05
     0.32  50/50  1%  Med  4.04E-01   -0.07
     0.32  50/50  4%  Med  1.01E-01  1.44
     0.32  50/50  J  Med  2.06E-02  3.16
     0.32  50/50  H  Med  1.17E-02  3.78
     0.32  50/50  K  Med  2.10E-02  3.14
     0.32  50/50  Ks  Med  3.56E-02  2.57
     0.32  50/50 Kp
     Med  3.15E-02  2.70
     0.32  Mirror/Open  1%  Med  2.02E-01  0.69
     0.32  Mirror/Open  4%  Med  5.05E-02  2.19
     0.32  Mirror/Open  J  Med  1.03E-02  3.92
     0.32  Mirror/Open  H  Med  5.86E-03  4.53
     0.32  Mirror/Open  K Med
     1.05E-02  3.90
     0.32  Mirror/Open  Ks  Med  1.78E-02  3.32
     0.32  Mirror/Open  Kp  Med  1.58E-02  3.46
     0.32  50/50  4%  Deep  3.5  -2.41
     0.32  50/50  1%  Deep  14.0  -3.92



    Measuring NICI's sensitivity directly is difficult because stars which are sufficiently bright for the AO system to lock on with high Strehl ratio saturate the infrared detectors in just a few seconds in the narrowest available filters.  Therefore, typical sensitivities have been compiled from short integrations on stars in the 1% filters, which are read-noise limited, and extrapolations of the signal and noise levels to the broader filters and longer integration times.

    The listed sensitivity values represent the nearly background-limited case, which for the narrow-band filters is achieved with single-frame exposure times of about 30 sec or longer.  Shorter single-frame exposures tend to be more read-noise limited, with noise up to 3 times higher. The point source signal was measured in a 12 pixel diameter aperture.

    Table 2: NICI point-source sensitivity.

    Filter Beamsplitter Mag S/N=1 in 1s Mag S/N=5 in 1 hr
    CH4 1% L yes 16.1 18.8
    CH4 1% S yes 15.9 18.6
    CH4 4% yes 16.9 19.6
    CH4 4% no 17.3 20
    L' yes 13.2 15.9



    Acquisition overheads associated with setting up on each new science target include time for slewing the telescope, acquiring the guide star on the peripheral WFS (P2) and the AO guide star on the NICI on-instrument wavefront sensor (OIWFS).  The nominal NICI acquisition time is 10 minutes.  This time applies to both coronographic and non-coronographic imaging modes when the target is a bright (V <~11), single star.  Fainter targets, or targets in very crowded fields requiring the use of finder charts, may require longer acquisitions.  Please note two important requirements for efficient acquisitions:

    - Due to NICI's small field of view and subarcsecond occulting spots, accurate target coordinates and proper motions are essential.

    - The correct visible (V) magnitude in the OIWFS Mag field is required for proper setting of the wavefront sensor neutral density filter. 

    Long imaging observations (> 1h) should be split to gain flexibility for changing conditions and ease queue scheduling.

    In imaging mode automated dither patterns have an estimated on-source efficiency of ~75%, where 25% of the elapsed time is used for telescope offsetting, NICI OI re-acquisition, detector readout etc; the exact overheads vary with the size of the offset and the exposure time.

    Gemini Observatory Participants