Introduction

In determining the optimal exposure times for near-IR observations, observers must weigh the trade-offs between exposure cadence and the contribution of detector noise. The desire to accurately subtract the rapidly varying OH emission drives one toward shorter individual exposures; we (the MOSFIRE team) have found from experiment that the characteristic timescale for variation of OH lines by more than 0.1% is ~30s, meaning that to optimize for removing OH, the background needs to be sampled on < 30 s timescales, usually through short exposures with small offsets along the slit (or slits). However, such observations are not optimal for obtaining the highest S/N in spectral regions between OH emission lines where the background is much lower. To take advantage of the low background (see the minimum count rates in the table), the noise from counting statistics in the background must be larger than the noise associated with reading the detector.

As listed in the table on the MOSFIRE detector page, the effective read noise upi can achieve depends on the number of Fowler pairs (non-destructive reads) selected in MCDS readout mode; for example, 8 reads yields 7.7e- rms noise, 16 reads gives 5.8e-, and 32 reads allows 4.2e-. Because the clock time per non-destructive read is 1.455 seconds, larger numbers of reads translate into additional overhead equivalent to N_pairs×1.455s, or 11.6 s for 8 reads, 23.3 s for 16 reads, 46.6 s for 32, etc.

The bottom line is that if accurate subtraction of OH emission from the night sky is important to your observing program (e.g., if your targets are bright and detector noise is not an issue) then dwell times at each nod position should be short, <=30 s. In general, though, one would like to achieve accurate background subtraction with the lowest possible noise both near OH lines and between them, where the sky is quite dark. This requires a compromise.

Recommended On-sky Exposure Times

Spectroscopic mode

Accounting for the brightness of the OH lines in each atmospheric band, we recommend the following readout modes and integration times for spectroscopy of faint targets. These are examples that have been tested on-sky with MOSFIRE and demonstrated to yield clean OH subtraction while simultaneously being nearly background limited in the darkest portions of the band between OH lines.

Please note that the background numbers all assume the default 0.7” slits; please scale accordingly if you are using non-standard slit width.
Filter Optimal Exposure Time
[s]
Sampmode Overheads
[s]
Observing
Efficiency
Noise in darkest regions
Sky noise
[e-]
Total noise
[e-]
Y 180 MCDS16 23.3 89% 7.8 9.7
J 120 MCDS16 23.3 84% 6.0 8.3
H 120 MCDS16 23.3 84% 8.3 10.1
K 180 MCDS16 23.3 89% 8.4 10.2

Thus, for example, one might request a “mask nod” sequence of 10 repeats of a 2-position dither pattern in K (meaning a total of 20×180 s exposures, comprised of 10 exposures at position A and 10 at position B). This sequence would require 20×(180+23.3)=4066 s for MCDS16 mode, not including the time for offsetting the telescope (roughly 3-4 seconds per dither). The clock time would then be approximately 4150 s for 3600 s of integration on source.

Imaging Mode

Use these exposure times for on-sky observations of targets fainter than the night sky.

Filter Optimal Exposure Time
[s]
Sky Brightness
[ADUs/pix]
@ OET
Sky Brightness Sky Brightness
(Vega)
[mag/arcsec2]
Sky Brightness
(AB)
[mag/arcsec2]
[ADUs/pix/s] [e-/pix/s]
Y 30 5,600 340-380 730-820 17.44 18.09
J 10.2 11,700 950-1,350 2,040-2,900 15.78 16.58
H 2.9 13,600 4160-5570 8940-11980 13.74 15.14
Ks 5.9 10,200 2,300-2,700 16,000 13.68 15.54
J2 25 11,700 470 1010
J3 30 11,160 372 800
H1 4.4 11,352 2640 5680
H2 4.3 11,910 2770 5960

Recommended Dome Flat Exposure Times

Spectral mode

Use the exposure times listed below to obtain well-exposed images with the MOSFIRE spectral lamp on Keck I. The goal is to achieve a peak illimination of about 13,000 DN/px, which remains well within the regime in which the detector response is linear to within 1%. Note that the exposure times listed below apply to 0.7 arcsec slits; to derive appropriate values for other slit widths, please scale the listed exposure times inversely by slit width.

Filter Requested
Exposure Time1
[s]
Actual
Exposure Time2
[s]
Flat lamp power level3 Observed Count Rate
[DN/pix/s]
Peak Counts for full exposure
[DN/pix]
Y 17 16.004 4 800 12,000-13,000
J 11 10.184 9 1,200 12,000-13,000
H 11 10.184 13.5 1,200 12,000-13,000
K 11 10.184 14 1,200 12,000-13,000

  1. Value you should enter into the MOSFIRE exposure control GUI.
  2. Resulting “true” MOSFIRE integration time (quantized by 1.455 s).
  3. The MOSFIRE Calibration Tool opened through MAGMA will manage spectral flats including dome lamp control and power level settings. Other means to turn on the flat lamps include:
  4. Via keyword, type: domelamps mos K (or appropriate filter) and this will set the power level automatically.
  5. Or from the pulldown menu, Telescope Status Menu-->NEW Dome Lamp control GUI
  6. Power level applied to the MOSFIRE dome lamp (range 0-20) by the calibration script to achieve the desired signal level.

Spectrophotometry mode

Use the exposure times listed below to obtain well-exposed data with the MOSFIRE spectral lamp on Keck I when using wide slits. This is a manual procedure. The goal is to achieve a peak illimination of about 13,000 DN/px, which remains well within the regime in which the detector response is linear to within 1%.
Slit Width Filter Requested
Exposure Time1
[s]
Actual
Exposure Time2
[s]
Flat lamp power level3 Observed Count Rate
[DN/pix/s]
Peak Counts for full exposure
[DN/pix]
15 K 2 1.45 17.5 6,900 10,000-11,000
    Manual Procedure:
  1. Insert Dark filter
  2. Setup CSU
  3. Open the hatch
  4. Turn on flat lamps: flatlamp on 17.5
  5. Acquire flats
  6. Turn off flat lamps: flatlamp off
  7. Insert Dark filter

Imaging mode

Use these exposure times to obtain well-exposed images with the Dome Lamps (new), the same lamps used for spectroscopy. The goal is to achieve a peak illimination of less than 18,000 DN/px, which remains well within the regime in which the detector response is linear to within 1%.

Filter Dome lamp
Power Level
Requested
Exposure Time1
[s]
Actual
Exposure Time2
[s]
Observed Count Rate
[DN/pix/s]
Observed Counts
[DN/pix]
Y 20.0 6 5.8 2400 14,000
J 3 19.5 15 14.3 900 15,000
H 3 20 6 5.82 2,300 13,500
Ks lamps off 2 1.4 5,500 8,000
H1 ???? 4 2.92 4,000 11,700
H2 ???? 10 8.73 2,130 15,500
J2 ???? 10 8.73 1,650 14,400
H2 ???? 8 7.27 2,060 15,000

  1. Value you should enter into the MOSFIRE exposure control GUI.
  2. Resulting “true” MOSFIRE integration time (quantized by 1.455 s).
  3. The dome flat lamp brightness varies and the J- and H-band imaging flats may saturate. If this happens, we recommend taking twilight flats instead. For twilight flats, we can use the box9 command in a mosfire server terminal to take flat images with 9-point dithering.

A0V Calibrators

Spectral type A0V stars are typically preferred for calibrating IR spectra because they have a minimum of instrinsic lines in their spectra. A handy list of such stars suitable for observing with MOSFIRE appears below. The recommended exposure times were scaled from the values determined for 20s exposures acquired during commissioning.

Object Filter V
[mag]
Recommended Counts
[DN/px] for
20s exposure
Exopsure Time
[s]
Count Rate
[Dn/pix]
Continuum OH peak
HD 22859Y7.80 1055001150011500
HD 22859J7.80 1055001100011500
HD 22859H7.80 2400031000¹35000¹
HD 199006H9.10 104500850014000
HD 22859K7.80 1070001400015000

  1. Non-linear regime of the detector

Notes

A0V throughput list

The list below is used by MOSFIRE instrument scientists to monitor thoughput of the instrument. These stars are AO stars and were selected to span a range of RA and have declinations near +19 deg. Typical exposure times for these standards are 6-10s for all bands. This list of stars is stored in the starlist file /kroot/starlists/mosfire/standards.

HIP13917           02 59 16.76 +01 14 40.4  2000.0 vmag=8.64  rotmode=pa rotdest=0
HIP17971           03 50 32.03 +29 44 41.7  2000.0 vmag=8.80  rotmode=pa rotdest=0
HIP23226           04 59 53.07 +27 19 13.5  2000.0 vmag=9.11  rotmode=pa rotdest=0
HIP24508           05 15 25.79 +15 37 48.4  2000.0 vmag=9.12  rotmode=pa rotdest=0
HIP30155           06 20 43.29 +15 41 39.1  2000.0 vmag=8.98  rotmode=pa rotdest=0
HIP43018           08 45 59.26 +13 15 48.6  2000.0 vmag=8.70  rotmode=pa rotdest=0
HIP55627           11 23 46.47 +34 14 33.7  2000.0 vmag=9.26  rotmode=pa rotdest=0
HIP56736           11 37 57.81 +15 46 36.8  2000.0 vmag=8.80  rotmode=pa rotdest=0
HIP61138           12 31 41.27 +22 07 24.4  2000.0 vmag=8.85  rotmode=pa rotdest=0
HIP64248           13 10 04.72 +12 19 13.2  2000.0 vmag=9.53  rotmode=pa rotdest=0
HIP68767           14 04 30.38 +21 23 17.3  2000.0 vmag=8.53  rotmode=pa rotdest=0
HIP87643           17 54 07.86 +16 56 37.8  2000.0 vmag=8.91  rotmode=pa rotdest=0
HIP98400           19 59 35.11 +11 53 21.7  2000.0 vmag=9.02  rotmode=pa rotdest=0
HIP98640           20 02 04.94 +12 19 19.5  2000.0 vmag=9.38  rotmode=pa rotdest=0

Acknowledgements

Many thanks to the MOSFIRE team, especially Dr. C. Steidel, for providing these recomendations for our user community.