WHT time lost to bad weather

The plot below shows for each month the fraction of night between dusk and dawn astronomical twilights when the planned observations could not be carried out due to bad weather. The plot symbols are the last two digits of the year. Engineering/commissioning nights, service/discretionary nights and nights when observers were using their own instrument, are excluded from the data. The raw data (more up-to-date) from which the plot was constructed are here.

On average, ~ 25% of observing time is lost to bad weather. The fraction varies between 13 and 35% from year to year. E.g. recent winters (2011/12 - 2014/15) have been mild, with less observing time lost.

Clear nights are almost guaranteed in June and July, while in December, the chance of clear sky is only 60%. Saharan dust ('calima') can significantly raises the extinction (by up to 1 mag) on some nights, particularly in July to September.

At the WHT, the dome is closed (i.e. time is lost to 'bad weather') if any of the following apply:

• The sky is clouded over
• The relative humidity exceeds 90% (after which the domes are not re-opened until the humidity has stayed below 80% for ~ 15 mins)
• Ice prevents safe opening or rotation of the dome
• The wind exceeds 80 km/h (gust)
• There is enough dust in the air that the beam from a torch is clearly visible

A comparison of bad-weather statistics for the WHT, INT and JKT can be found on the ING site-quality pages.

At other good observing sites, typically 20 - 35% of observing time is lost to bad weather.

Average usable hours per night
There are on average 6.8 clear (i.e. dome open) hours per night (between astronomical twilights) at the WHT. This number varies only a little with season (6.2 to 7.2) because the varying length of the night happens to be almost compensated by the varying fraction of night lost to poor weather.

For planning surveys requiring a large amount of telescope time (e.g. WEAVE), other considerations must be taken into account:

• Some time will be lost to technical problems. This currently averages ~ 1 - 2 %. For observations with a new instrument, it is likely to be higher, for an initial period after commissioning of the instrument.
• The seeing will be poor on some nights. The seeing recorded by the DIMM seeing monitor outside the WHT building is < 1.5 arcsec on 80% of nights. The seeing measured at the WHT is usually consistent with that measured by the DIMM, except on nights when the primary mirror is more than 3 degC warmer than the air in the dome (~ 10 - 20% of nights), when there may be significant mirror seeing.
• On some nights, extinction by dust or light cloud may be significant, while not requiring the dome to be closed.
• Additional usable time is available during twilight. For typical observing programmes, this amounts to an extra 0.5 - 1 hour per night.
• Commissioning, instrument changes, maintenance and quality control eat into the total amount of time available for observing at the WHT.

Note that the number of clear hours required for the science observations may change with time, e.g. it changes as sky brightness varies during the solar cycle.

How long does a spell of bad-weather last?
The plots below shows the approximate number of hours lost to bad weather each night during 2014 and 2015 (data taken from WHT observing logbook).

The plots show the seasonal variation in bad weather (and in length of night); the strongly bimodal distribution of number of hours lost (usually zero, or nearly all); and bad-weather spells typically lasting several nights. Similar plots for 2012 and 2013 can be found here.

The table below shows, for each month m, the fraction F(m) of bad-weather nights (defined as nights on which more than 2 hours are lost), averaged over 2006-15:

     Jan   Feb   Mar   Apr   May   Jun   Jul   Aug   Sep   Oct   Nov   Dec

     0.33  0.38  0.36  0.30  0.14  0.06  0.02  0.10  0.25  0.37  0.43  0.39

The probability that a bad-weather night is followed N nights later by another is:

  N  Jan   Feb   Mar   Apr   May   Jun   Jul   Aug   Sep   Oct   Nov   Dec
 
  1  0.54  0.66  0.55  0.54  0.33  0.21  0.17  0.29  0.46  0.59  0.62  0.61
  2  0.50  0.50  0.46  0.45  0.31  0.11  0.00  0.19  0.39  0.50  0.53  0.54
  3  0.50  0.50  0.45  0.43  0.33  0.11  0.00  0.19  0.35  0.46  0.47  0.46
  4  0.49  0.49  0.42  0.42  0.29  0.05  0.00  0.10  0.32  0.47  0.42  0.45
  5  0.43  0.45  0.32  0.34  0.21  0.05  0.00  0.19  0.31  0.46  0.41  0.40

  6  0.42  0.45  0.34  0.36  0.17  0.05  0.00  0.13  0.28  0.47  0.46  0.39
  7  0.35  0.47  0.32  0.31  0.14  0.05  0.00  0.13  0.34  0.46  0.43  0.41
  8  0.32  0.44  0.26  0.25  0.14  0.00  0.00  0.16  0.31  0.41  0.39  0.43
  9  0.34  0.43  0.28  0.24  0.12  0.00  0.00  0.06  0.32  0.42  0.41  0.47
 10  0.36  0.43  0.28  0.33  0.05  0.05  0.00  0.10  0.27  0.40  0.48  0.37

In other words, the a priori probability of a random night in month m being a bad-weather night is F(m), but the probability is higher than F(m) for a few nights after any given bad-weather night. For the night immediately following a bad-weather night (i.e. N = 1), the probability is ~ 1.8 x F(m). The probability drops to ~ 1.3 x F(m) after typically N ~ 3 - 4 nights.