Kipp&Zonen BSRN Scientific Solar Monitoring System - Manual (Page 68)

56
original filters and from the sam e m anufacturing lot. In this m anner, the waveband characteristics can
be m aintained over longer periods of tim e.
7.3 Data Acquisition
7.3.1 Sam pling
Observations are to be m ade at a frequency of once per m inute. Unlike other irradiance observations
that obtain one-m inute averages, the m easurem ent of direct spectral irradiance is to be an instantaneous
observation at a given tim e (t). When m ultiple m easurem ents are m ade with a single-detector instrum ent
or as part of a large set of observations, the difference in tim e (Mt) between the first and last observation
m ust also be known. If m ultiple instrum ents are to be used, both t and Mt m ust be known for both
instrum ents, and the observation tim es m ust by synchronized.
The dark (or zero) signal should be m easured for each transducer as part of the observation sequence
when possible. For instrum ents that are not capable of obtaining a dark signal with each observation,
dark signals should be m easured at night, or a sam pling schem e should be designed so that a dark
signal is m easured at least daily. This m ay be as sim ple as blocking the entrance aperture of the instrum ent
during an observation tim e.
Measurem ents should continue throughout the period that the sun is above the local horizon (see Section
3.1.2). The sam pling program m ust not autom atically screen observations because of cloud or other
form s of contam ination.
The use of sam pling frequencies of less than once per m inute will be accepted into the archive until
2004 for autom ated instrum ents that m eet all other BSRN specifications.
Measurem ents obtained from handheld sunphotom eters will not be accepted by the archive.
7.3.2 Data Acquisition
Depending upon the radiom eter, the data acquisition system m ay be part of the instrum ent package.
Other radiom eters require separate instrum entation for data collection. In all cases, the collection system
m ust be capable of providing:
(1) A signal-to-noise ratio no less than 5000:1 for the sm allest output signal of the spectral irradiance
channels m easured when the sun is at its highest elevation for the station. In high turbidity
500
cases (
J
a > 0.750 occurs m ore than 20% of the tim e) the signal-to-noise ration of the data
collection system should be at least 12000:1. For exam ple, if the highest signal output by
the least sensitive channel was 500 m V, the collection system would have to m easure accurately
a signal of 100 :V or less (5000:1) or ~40 :V or less (12000:1). In contrast, the com plete
system signal-to-noise ratio for irradiance m easurem ents is 10000:1.
(2) An ability to m aintain tim ing accuracy to better than one second (see Section 2.3.1).
(3) A tim e stam p so that Mt can be given or calculated.
7.4 Calibration
7.4.1 General
The calibration of spectral radiom eters has yet to be fully resolved. At present there is no hierarchy
such as the W RR for spectral radiation. Until such tim e as a calibration hierarchy becom es available,
the BSRN W orking Group on Aerosol Optical Depth suggests the following calibration m ethods:
(1) An independent detector-based standard. This m ethod uses a fully characterized silicon
photodiode trap detector for the m easurem ent of a stable irradiance source that illum inates
both the detector to be calibrated and the trap detector sim ultaneously. This m ethodology
is very successful for calibrating instrum ents using specific laser wavelengths. The use of
this m ethod for the calibration of wider bandwidths rem ains under developm ent.