Atmospheric Opacity Statistics

Farmers and astronomers are more at the mercy of the weather than probably any other sector of society. The summit of Mauna Kea benefits from having its very own weather forecast (http://mkwc.ifa.hawaii.edu/forecast/mko/index.cgi), provided twice daily by the Mauna Kea Weather Center, administered by the UH Institute for Astronomy in cooperation with the Department of Meteorology. This forecast, along with a plethora of other information, is compiled on the JAC Weather page at http://www.jach.hawaii.edu/weather. There is an excellent summary of the mission and operations of the Mauna Kea Weather Center, in the Autumn 2006 UKIRT Newsletter (available at http://www.jach.hawaii.edu/UKIRT/publications/Newsletter/issue19/special.html)

All of these facilities are extremely useful in learning of the weather and atmospheric conditions now, or over the next few days, but what if one needs an estimate of the probability of excellent/good/poor weather several months hence? For at least one aspect of the atmosphere of great concern to sub-millimetre observers, i.e. the atmospheric opacity, we can check the average conditions over the past several years. Since 1989, the CSO tau-meter has been collecting opacity data at the wavelength of 225GHz. Originally built by NRAO, the instrument has long been operated and maintained by the Caltech Submm Observatory.

Fraction of nights (% of total) over the period 1986-2006, where the τ_225GHz was equal to or less than the value specified, as a function of the month.

In figure 1, we display the relative number of nights (as a percent fraction of the total number) that the atmospheric opacity at 225 GHz (τ_225GHz) is at or below a certain value, for each month of the year. These values are the divisions between the weather bands used by the JCMT, i.e. τ_225GHz < 0.05 is grade 1 (i.e. extremely dry) weather, 0.05 < τ_225GHz < 0.08 is grade 2 (dry), 0.08 < τ_225GHz < 0.12 is grade 3 (medium), 0.12 < τ_225GHz < 0.2 is grade 4 (wet), and τ_225GHz > 0.2 is grade 5 (extremely wet). Figure 2 shows the same, but including only the most recent six years. It?s immediately apparent, as any frequent user would know, that the summer is considerably wetter than the winter. However, the faction of extremely wet (τ_225GHz > 0.2) weather does not change appreciably through the year. It?s also quite apparent that the period 2000-2006 has a much smaller faction of time during the summer with extremely dry (τ_225GHz < 0.05) weather compared to the average over 1989-2006.

Fraction of nights (% of total) over the period 2000-2006, where the τ_225GHz was equal to or less than the value specified, as a function of the month.

Using these figures, one can estimate the probability of, say, band 2 weather in July. From figure 2, the fraction of time with τ_225GHz < 0.08 is about 25%, and τ_225GHz < 0.05 is about 3%. Thus about 22% of the time in July should be band 2, on average.

Total weather losses (added 20100204)

The 10-30% of time shown in the above plots that is ascribed to conditions with τ_225GHz > 0.2, i.e. band-5, contains some proportion of time that is lost altogether. Analysis of recent semesters provides the following

          Semester  Hours Lost / Hours Available   =   % loss
          --------  ----------------------------       -------
            08a         222    /  2000            =     11
            08b         352    /  2287            =     15
            09a         482    /  2233            =     21
            09b         232    /  2175            =     11
                       ----------------               -------
          Totals       1288    /  8695            =     15