Kipp&Zonen BSRN Scientific Solar Monitoring System - Manual (Page 85)

73
considered in these cases. The first is the norm al range of the instrum ent, for exam ple a pyranom eter
range m ay be -0.1 to 12 m V, while the second is the absolute range such as 0 - 5 V for a pressure
transducer. In setting bounds checks on the form er, one can sim ply be observing an unusual phenom enon,
while on the latter, if the lim it is exceeded, an instrum ent problem has occurred.
9.3.1.4 Conversion to solar tim e
W hile visually inspecting the data while it is being archived in local standard or UTC is useful, converting
the data, either in real tim e or post-processing, provides an excellent m eans of determ ining accuracy
of the system tim e. For system s recording data with a frequency of one-m inute or greater, the sym m etry
around solar noon (clear or partly clear days) can provide a m eans of independently checking the system
tim e. Corrections to the tim e can be m ade by adjusting the tim e stam p on the data to restore the curve’s
sym m etry. Care m ust be taken in correcting data in this m anner and a tim ing flag should be set to
allow future users to know a tim e shift has occurred. Once noted, correction of the problem should
be undertaken as rapidly as possible.
9.3.1.5 Scanning Minim um , Maxim um , Standard Deviations
W hile difficult to accom plish in real tim e, post-processing scans or plots of the m in, m ax, and standard
deviations of the signals should be done both on individual channels and on m ultiple com m on channels
to check for any short term uncertainties that have not been noticed using only the m ean values. Minim um
values dipping below zero or m axim um s exceeding reasonable values, particularly in com parison with
other sim ilar signals, provide a rapid way of focussing on potential problem s. A sim ple exam ple would
be the cleaning of the dom e of an instrum ent during cloudy bright conditions. Although the m ean value
of one m inute m ay not be significantly altered, the m inim um value and the standard deviation could
be altered profoundly. In this m anner, single data points could be flagged for tim es of increased uncertainty.
During any period, if several peculiar events occur, the frequency and periodicity of the events should
be tested. Such periodic problem s could indicate potential electronic failures, buffering problem s in
the transfer of data or difficulties associated with the DAS.
9.3.2 Procedures for specific fluxes
9.3.2.1 Direct, diffuse and global
Testing the direct and diffuse against the global radiation is a sim ple and straightforward test, with
the exception of tim e near sunrise and sunset, and to a lesser extent during tim es of rapidly changing
irradiance levels (because of different instrum ent response tim es). This test should be done on all
irradiance data before subm itting the data to the archive. Sim ply,
GLOBAL = DIFFUSE + DIRECT cos(2)
where the zenith angle (2) m ust be calculated according to the station location, date and tim e
(Annex I provides an algorithm )
During clear sky or stable conditions, the difference between the global and the sum m ation should
be within the uncertainty levels given to the instrum ents. In the case of a cavity radiom eter and two
well-calibrated pyranom eters the differences should be less than 2% or 15 W m , whichever is less,
-2
at solar elevations greater than 10°. At lower solar elevations and during changing conditions, the
differences should be less then 3.5% or 20 W m , whichever is less. For larger differences further
-2
tests should be done to determ ine the cause of the discrepancy.
NOTE: If the direct and shaded radiom eters are on the sam e platform care m ust be taken when using
this procedure. During tim es when the solar tracking is slightly m isaligned, the errors in the direct beam
and the diffuse can be offsetting within the range of the uncertainty values given above.