- Data distribution
- Using the Archive
- Gemini Home
- Telescopes and Sites
- Science Visitors at Gemini
- Observing With Gemini
- Retired Instruments
- Status and Availability
- Observing Modes (Meta-Modes)
- Data Reduction and Calibrations
- Early Science
- Instrument Performance
- Observation Preparation
- Public Data
- Target Duplication Policy
- Campaign Science
- GPI Phase II Checks
- GPI FAQ
- Visiting Instrument Policy
- Visiting Instrument Telescope Interfaces
- DSSI Speckle Camera
- TEXES (North)
- Integration Time Calculators
- Magnitudes and Fluxes
- Near-IR Resources
- Mid-IR Resources
- Observing Condition Constraints
- Performance Monitoring
- SV/Demo Science
- Future Instrumentation & Current Development
- Queue and Schedules
- Data and Results
- Calibration Data Retrieval
- Past/Future Development
Change page style:
performance will be updated as soon as new data will be processed and analyzed.
Please subscribe to the GPI News Email Exploder to receive the last updates.
These contrast performance were determined using First Light run data.
Contrast versus target star V magnitude
The figure below shows the single slice 60s H-band raw contrast for a V-mag 3.86 star (Beta Pic) acquired at 1.08 AM compared to a V-mag 7.36 at 1.35 AM (HD95086, both observations were acquired the same night). The GPI contrast seems fairly constant as a function of magnitude and air mass under similar seeing conditions. More data acquired under similar seeing conditions with various magnitudes and air masses are needed to better characterize GPI performances at higher air masses and for dimmer targets. The night was fairly good with the DIMM in the range of 0.5 arcsec.
The exposure time for HD95086 is 120s.
Images have been unsharp mask with a 13 pixel box kernel and convolved with a 5 pixel diameter aperture. Contrast is obtained by averaging the peak flux of the 4 spots. Contrast in integrated H-band would be approximately up to sqrt(number of independent wavelength channels)=sqrt(8) better, especially at large separations.
Contrast versus IFS spectral band
figure below shows the GPI
contrast as a function of bandpass for a bright V=3.83 star (Beta Pic). The
solid line represents the single slice 60s raw contrast at H-band, while the
dashed, dotted and dot-dashed lines are the J, K1 and K2 bands respectively.
Exposure times were 90s for J, and 60s for K1 and K2 bands. The sequences were
acquired on different nights, so the comparison might not be 100% accurate due
to varying observing conditions and instrument optimizations. Overall, to 1
magnitude, GPI performance seems fairly uniform as a function of wavelength.
Note that the K2 band has sharply declining throughput beyond 2.2 microns. On dimmer stars or at large angles where photons and sky background noise dominate, K2 contrast will be significantly worse
Contrast : Different PSF Subtraction Algorithms Compared
The Beta Pic (November 18, 2013, H-band) sequence, with various PSF subtraction algorithms. The solid line is the 60s raw contrast, while the thin dashed, dot-dashed and dotted lines are the ADI, SSDI and TLOCI (a combination of ADI and SSDI image subtraction using an estimated spectrum for the planet) PSF subtraction residuals respectively (note that SSDI here is using a spectral template, so it is highly similar to TLOCI, except without the ADI information to suppress the PSF; the DUSTY spectrum is used here for SSDI and TLOCI). The three thick dashed, dot-dashed and dotted lines are the 30 minutes combined curves (ADI, SSDI, and TLOCI). While some gain can be achieved with ADI and a least-squares, a better contrast, up to a magnitude, can be achieved with SSDI/TLOCI. The higher speckle correlation seen as a function of wavelength is the reason why a better contrast is obtained when using the spectral information to suppress the PSF. TLOCI, that combines both SSDI and ADI into a single algorithm, achieves the best contrast overall, being slightly better than SSDI alone. The SSDI/TLOCI gain would be even higher for methane-like spectra (strong spectral features).
We used 60s image with unsharp mask (13 pixel box kernel), aperture convolution (5 pixel median box) and spatial magnified to align speckles as a function of wavelength).