We want to know how much time we could save if we had continuous focusing.

From metrics we can estimate the length of time we now spend focusing on each instrument. (This is probably best done on an instrument-by-instrument basis.) Unfortunately, some Mira runs do not get recorded. We can ignore those nights with zero recorded focus time.

For those nights in 2003 and 2004 (as of 1/27/04) that have nonzero focusing numbers in metrics, we calculate the mean and median number of seconds spent focusing each night. Note that we are being very simplistic here, and ignoring time lost to weather.

Table 1. Mean and median times (in seconds) per night in
2003 and early January 2004 for each instrument.

Instrument

Mean

Median

LRIS

897

737

HIRES

561

504

NIRC

689

433

LWS

649

786

DEIMOS

805

735

ESI

620

644

NIRSPEC

1041

839

NIRC-2

0

0

Also, do we want to ignore the first focus of the night? How about tilts that we calculate with Mira? How stable are they during a run?

How to estimate the total number of nights? In metrics, list all nights for various instruments for 2003-2004. Convert this to a fraction of nights. Then multiply this fraction by the 365.25 nights and by the median number of seconds spent focusing per night shown above. (For LWS-NIRC use 670 sec/night.) Convert to hours.

Table 2. Time spent focusing from Table 1 converted to hours
per year, weighted by the fraction of nights each instrument was
on the telescope.

 

Instrument

Fraction of nights

Total hours focusing

Keck I

LRIS

51.3%

38.4

HIRES

43.0%

22.0

NIRC-LWS

20.0%

13.6

Keck II

DEIMOS

28.9%

21.6

ESI

12.3%

8.1

NIRSPEC

35.5%

30.2

NIRC-2

23.2%

0

So, on Keck I we use roughly 74 hours per 365.25 nights to focusing. On Keck II we use roughly 60 hours per 365.25 nights. Per calendar year, these must be decreased by the fraction of nights actually used for science (as opposed to engineering) and/or the fraction of nights lost to weather.