15
LI-200SA Spectral Response
The relative spectral response of the silicon photodiode does not extend
uniformly over the full solar radiation range. A typical response curve is
presented in Figure 5. The response is very low at 0.4
µ
m and increases
nearly linear to a maximum at about 0.95
µ
m and then decreases nearly
linear to a cutoff near 1.2
µ
m. Changes in the spectral distribution of the
incident light, coupled with the non-uniform spectral response, can cause
errors in the photodiode output. Hull
3
shows that in the 0.4 to 0.7
µ
m
range, the spectral distribution of sun plus sky radiation on a horizontal
surface is remarkably constant even when clear and overcast days are
compared. However, Gates
2
indicates that the major change in spectral
distribution of solar radiation occurs in the near infrared where water vapor
absorption takes place on cloudy days. Data collected at low solar
elevations can show significant error because of altered spectral distribution
which changes in atmospheric transmission. This is a small part of the daily
total so the possible observed error usually has an insignificant effect on
daily integrations.
The area under the spectral irradiance curve of the source is directly
proportional to the energy received by a horizontal surface. Under specific
but typical conditions, energy received on a completely overcast day has
been estimated to be 11.3% of that received on a clear day. When both
spectral distributions are weighted according to a typical response curve of a
silicon photodiode, the response on this cloudy day is 12.6%. Therefore,
errors incurred under different sky conditions, due to the spectral response
of the photodiode, will be small. The field tests of Federer and Tanner
1
and
Kerr, Thurtell and Tanner
4
confirm this conclusion.
Calibration
The LI-200SA Pyranometer has been calibrated against an Eppley Precision
Spectral Pyranometer (PSP) of which the calibration is periodically
confirmed. The calibration was performed under daylight conditions by a
computer sampling of instantaneous readings from the Eppley and LI-COR
pyranometers. Instantaneous readings were taken continuously for 10
minutes and then averaged. Sequential ten minute averaging periods were
run from sunup to sundown for 3-4 days. These ten minute averages were
then evaluated and used to compute an average calibration constant. The
uncertainty of calibration is
±
5%.