Kipp&Zonen BSRN Scientific Solar Monitoring System - Manual (Page 66)

54
AOD values obtained from the archive would continue to be based solely on the subm itted transm ission
data.
7.2 Instrument and Wavelength Specifications
7.2.1 Instrum ent Specifications
Historically, sunphotom eters have been designed as either single detector instrum ents that use a rotating
wheel to place a filter between the opening aperture and the detector, or instrum ents that m easure
each bandwidth by m atching a detector and an interference filter in its own optical fram e. Newer
instrum ents are now being developed that use prism s or gratings to separate the solar spectrum and
detector arrays are being used to measure the spectral intensity. Each of these m ethods have advantages
and disadvantages that should be studied before a particular type of instrum ent is purchased or built.
As with the radiation sensors, the type of spectral radiom eter used at individual BSRN sites for the
m easurem ent of aerosol optical depth is the responsibility of the principal scientist. The choice should
be based upon achieving the lowest m easurem ent uncertainty possible, but never greater than the
BSRN guidelines. Several com m ercial instrum ents are presently available that can be configured to
m eet the present specifications (see Annex B5), but noncom m ercial research instrum ents that m eet
these m inim um requirem ents are fully acceptable. To encourage instrum ent developm ent, radiom eter
specifications have been lim ited to the following:
(1) The instrum ent Field-of-View (FOV) m ust be less than 2° for radiom eters that use a lens to
focus the solar im age or have an opening angle of less than 2.5° and a slope angle of no
greater than 1° for instrum ents without optics
(2) The instrum ent, or instrum ent in com bination with a solar pointing device, m ust track the solar
disk to better than 0.1°.
(3) The stability of the detector m ust be better than 0.5% per annum at a nom inal tem perature
of 25° C. This is separate from changes in the spectral bandpass (see below),
(4) The detector should be tem perature stabilized with a tem perature-related drift of less than
of less than 0.1% over the com plete range of tem peratures at which the instrum ent is to be
operated. If the tem perature cannot be stabilized, then the characteristics of the detector
m ust be known and corrected. The efficiency of typical silicon photodiode detectors can change
by m ore than -0.5% °C at wavelengths less than 350 nm and by m ore than 0.5% °C at
-1 -1
wavelengths greater than 950 nm . The typical tem perature response for wavelengths between
these is -0.2% °C .
-1
(5) As in (4) the tem perature of the filters m ust either be stabilised or m onitored to ensure that
tem perature related shifts in the wavelength do not exceed the specifications set out below.
It is recom m ended that the filters be tem perature stabilised to reduce cycling of the passband
with tem perature. Typically the change in the peak transm ittance of an interference filter is
~0.01% °C and the central wavelength increases by 0.01 - 0.03 nm °C . Daily tem perature
-1 -1
variations could thus alter the central wavelength in the order of 0.6 nm and seasonal
tem perature variations in continental clim ates could shift the central wavelength by m ore
than 2 nm between cold and warm season extrem es.
7.2.2 W avelength Specifications
The following specifications have been adopted by the BSRN and have followed, to a large extent,
the recom m endations of the W orld Optical Radiation Calibration Centre (W ORCC). The m ajor exception
2 2
is the selection by the W ORCC of passbands centred on H O and NO absorption bands. The selection
criteria of the four prim ary wavelengths for the BSRN are that they are in relatively flat areas of the
solar spectrum and away from water vapour or trace gas absorption. The m ajor exception to the selection
criteria is the choice of the 500 nm waveband. This was selected because for historical reasons.
The actual central wavelength m ust be determ ined to better than 0.3° and the out-of-band rejection
should be at least 10 for a m inim um of ±40 nm from the central wavelength. In all cases, but especially
-4
for filters in the UV, the transm ittance of the filter should be tested over the entire spectral range of
the detector. In cases where out-of-band rejection is less than 10 beyond 40 nm of the central wavelength,
-4
longpass or shortpass blocking filters should be used as appropriate.