The coherence time of an interferometer is the interval during which
the fringe phase remains stable, changing by 
radian. The
coherence time is determined by the stability of the atmosphere along
the line of sight to the source. At lower frequencies (at or below 5
GHz) the ionospheric contribution is dominant. At higher frequencies,
the tropospheric influence is more important. The coherence time found
at each frequency may vary rapidly by over an order of magnitude and
is determined by a number of factors. Close to sunspot maximum the
ionosphere is far more likely to exhibit periods of intense
activity. In addition, the ionosphere is likely to be more active
during the day than at night; and more active during the Winter than
during the Summer. The troposphere, by comparison, is more active
during the Summer rather than the Winter and is strongly influenced by
the detailed weather (especially the water vapour content of the atmosphere).
The coherence time determines the time interval over which data can be averaged for calibration, the cycle time between observations of phase-cal and target and the accuracy of transfer of phase solutions. Experiments where the target is too weak to self-calibrate, or those requiring full polarization or narrow-band spectral line observations using a phase-cal observed in a wide bandwidth, are particularly vulnerable. Under difficult conditions shorter integration times and shorter cycle times between target and phase-cal can be used. However the former can restrict the spectral resolution and the latter leads to loss of data during slewing, so such measures are not undertaken unless essential. The default on-line integration times used for sources of modest angular extent will be set so that coherence time problems are avoided. There are times when the phases on the longest MERLIN spacings could almost be drawn by ruler over many hours; there are also times when decorrelation occurs within the on-line integration time. We attempt to minimise the effects of the latter by selecting suitable weather for 22-GHz observations (ideally dry winter and spring nights) and avoiding low frequency observations close to solar maximum.
Since coherence times are so variable, it is difficult to set simple
guidelines, so we present values based on the upper and lower
quartiles of the distribution. Thus good values represent what the
user should reasonably expect to get for observations under favourable
circumstances; and poor values those expected under adverse
conditions.
Table 4.5 gives typical coherence times on the longest
MERLIN baselines for sources at elevations above 
. Coherence
times at low elevation - and so for sources at low 
-
will be shorter. The values given for 151 MHz relate to observations
made near solar minimum.