Sunshine Duration (SD) Measurement
using a pyranometer.
1 . Preliminary remarks :
1.1. A pyranometer is used to measure the Global solar radiation received on an horizontal
surface at ground level. This is the main application of the pyranometer. If desired this
instrument can also be used as a SD sensor. This second use is a plus, that possibly allows us
to save up the high cost of a true SD sensor (every penny counts !).
Nonetheless, the fact remains that SD measurement results are important climatologic data
for characterizing the climate of a region for the general public.
We'll consider below, only pyranometers equipped with a thermopile. We exclude
photovoltaic (silicon) pyranometers. We have no experience with this kind of radiometers
used to measure SD.
1.2. The "best" SD instrument is an hypothetical Pyrheliometer the characteristics of which
are still to be specified. Is this WMO recommendation always of topical interest ? For the
time being we have no choice but to use instruments available from our manufacturers
(K & Z, Eppley, etc.). But, as the pyranometer, these instruments are never used as SD
sensors (too expensive with their equatorial mounts). They are only used to constitute the
"Reference" for SD measurements. Sunny periods are counted – during a period of one hour,
one day, one month, etc., – only if the Direct solar irradiances measured are superior to
120 Wm-2. According to WMO, this threshold is conventional and can vary between 96 Wm-2
and 144 Wm-2 (20%) !
Such variations in the SD threshold cause less than 2% in the 1991/2000 decade SD
"Reference" at Carpentras; and probably less than 3% at Trappes. Carpentras is located in
the Rhone valley about 80 km from the Mediterranean Cost. Trappes is close to Paris, 27 km
in the Southwest. Carpentras' area is sunnier than Trappes' area.
It seems that the variations of SD reference for shorter periods than one decade (one year,
one month, or even one 10-day period) are almost equivalent (about 3%) for 20% variations
of the conventional threshold. Therefore the expanded uncertainty U95 equal to 1.9% for
a CH1 pyrheliometer (with an approximate 95% confidence level for a coverage factor of
2.0 – Carpentras' estimation), is of no consequence in practice on the "accuracy" of the SD
Reference. This uncertainty is about 10 times lower than the variations (20%) tolerated by
WMO for the SD threshold. As a result, we can consider that commercial pyrheliometers give
a satisfactory answer to our needs. We'll consider only the conventional threshold of
120 Wm-2 afterwards.
2. SD measurements using a pyranometer and the Carpentras' Algorithm :
2.1. Equations : We consider that the Sun shines at a given minute if the Global irradiance
measured Gmeas (Wm-2) is superior to a certain threshold Gthres (Wm-2):
Gmeas > Gthres
with : Gthres = Fc Gmod equal to the product of the Factor Fc by the model Gmod, that
represents a possible variation of the Global irradiance under clear skies and a
mean value of the air turbidity :
Fc = A + B cos [(2 365) d]
d is the day number in the annual sequence,
A and B are two parameters the values of which depend on the site (latitude
and meteorological conditions). Some values are given afterwards.
Gmod = 1080 (sin h)1.25
h is the Sun altitude : h + SZA = /2 radian (or 90 degrees). SZA is the Solar
Zenith Angle.
Sin h = sin sin + cos cos cosAH
is the latitude of the site
is the geocentric declination of the Sun
AH is horary angle of the Sun calculated from twelve (TST or LAT)
Note : the h calculation do not take into account the atmospheric
refraction.
At the end of each round minute, if Gmeas > Gthres and h > 0.05236 radian or
h > 3 degrees, then the SD meter must be incremented by 1 unit. This data processing may
be done in real-time, or at some point.
2.2. The lowest angle h (solar elevation) for which sunshine hours are calculated is 3 degrees,
although some experimenters tested lower values of h successfully. We think that
conventionally h > 3 degrees in a good choice.
2.3. The values of A and B Coefficients. Some values of A and B are given in the
following Table :
Site A B
Carpentras (BSRN Station – France) 0.73 0.06
Reunion Island (Indian Ocean) 0.66 -0.03
Tahiti Island (Pacific Ocean) 0.63 -0.04
Momote (BSRN Station – New Guinea) 0.68 -0.06
Tateno (BSRN Station – Japan) 0.74 0.06
The mathematical sign of B changes in Southern Hemisphere (because of the
season difference in the two hemispheres). We do not need a great accuracy for the B
Coefficients. The A Coefficients diminish slightly with the latitude of the site, and
probably depend on the mean value of the atmospheric turbidity.
The References of SD measurements in the BSRN stations are pyrhéliomètres CH1
(K & Z) at Carpentras and Tateno, and a NIP (Eppley) at Momote. The pyranometers
are a CM11 at Carpentras, a CM21 at Tateno and a PSP (Eppley) at Momote.
More generally meteorologists have a clear preference for Kipp and Zonen
radiometers ….
2.4. The errors. At sunrise, for example, under clear skies when the atmospheric
turbidity is high, Gmeas < Fc Gmod. Then SD = 0 even if the Direct irradiance is
superior to 120 Wm-2. In such a case the coefficient A is not suitable. The SD
measurement error can reach 1 hour or roughly 10% of the daily SD.
For a site particularly turbid A must be adapted to this specific situation.
