Modeling Solar Radiation at the Earth’s Surface

Instrumentation: Solar Radiometer

A pyrheliometer measures B, the direct beam radiation. Pyrheliometers have a nar­row aperture (generally between 5° and 6° total solid angle), admitting only beam radiation with some inadvertent circumsolar contribution from the Sun’s aureole within the field of view of the instrument, but still excluding all diffuse radiation from the rest of the sky (WMO 1983). Pyrheliometers must be pointed at, and track the Sun throughout the day. Their sensor is always normal to the direct beam, so that B is often called “direct normal irradiance” (DNI).

A pyranometer measures G, the global total hemispherical, or D, the diffuse sky hemispherical radiation. Pyranometers have a 180° (2n steradian) field of view. The measurement of D is accomplished by blocking out the beam radiation with a disk or ball placed over the instrument and in the path of the direct beam that subtends a solid angle matching the pyrheliometer field of view. This requires tracking the Sun with the blocking device through the day. A lower-cost alternative is to use a fixed band or ring of opaque material placed to shadow the pyranometer throughout the day. The shading band/ring approach introduces errors into the measurement of D, since part of the sky radiation is blocked by the shading device. This blocking effect varies with the shading device’s geometry, time, and atmospheric conditions. An attempt to compensate for this is usually done by applying a geometric or empirical correction function to the data (e. g., Drummond 1956; Siren 1987), but this is far from perfect. For this reason, only the tracking-shade method is used at research- class sites.

Figure 1.3 portrays typical instruments used to measure G, B, and D in the field. These instruments all have thermopile detectors, except as noted. The thermopile - based detectors are sensitive to the whole shortwave spectrum, in contrast with

solid-state detectors, discussed further below. Note, however, that nearly all ra­diometers are protected from the elements by a window. This limits the spectral sensitivity of thermopile-based instruments to either 290-2800 nm for glass domes (used in most pyranometers) or 290-4000 nm for quartz plane windows (used in pyrheliometers).

The blackened absorbing surface of a thermopile is heated by the incident so­lar radiation. A number of thermojunctions between dissimilar metals (typically “type T” thermocouple junctions made of copper and constantan) are in contact with the absorbing surface. Thermal flux upon the junctions produces a voltage propor­tional to the difference in temperature between the heated junctions, and a similar set of “cold junctions” in series with the hot junctions. The output of thermocouples is slightly nonlinear, resulting in some curvature in the relationship between signal and temperature (NBS 1974).

Typically, a thermopile is made of approximately 20 to 40 junctions, and temper­ature differences between hot and cold junctions are 5° C for a 1000Wm~2 optical input, resulting in a 4 mV to 8 mV signal. In addition to the nonlinearities in the thermal response of the thermocouples described above, the absorbing surfaces of the detectors are not perfect isotropic (or “Lambertian”) surfaces; and finally there are exchanges of infrared radiation between the radiometers/detectors and the (usu­ally much colder) sky, all of which contribute to the uncertainty in calibrations and measurements using these detectors (Haeffelin et al. 2001).

Solid-state silicon photodiodes, mounted beneath diffusers, respond to inci­dent radiation by generating a photocurrent, which is proportional to the incident flux. However, these devices have narrow spectral response ranges (e. g., about 350-1000 nm for crystalline silicon) and do not produce a signal proportional to the entire optical radiation spectrum. Since the path length of solar radiation through the atmosphere varies, and the atmosphere is not stable in composition throughout the day, changing infrared spectral content of the solar radiation is not captured by photodiode radiometers. As a result, solid-state detectors are less accurate than most thermopile radiometers discussed above.

“Burning” sunshine recorders were first developed by John Francis Campbell in 1853 and later modified in 1879 by Sir George Gabriel Stokes. The original instrument was based upon glass spheres filled with water, and later solid glass spheres. The latter device, which is known as the Campbell-Stokes (CS) recorder (Fig. 1.4), is still manufactured and used today, and constitutes the oldest solar radi­ation instrument still in service.

Modern instrumentation may be used to determine percent sunshine as well, by comparing the amount of time a pyrheliometer signal is above the bright sunshine threshold of 120Wm~2 with the day length (WMO 1996). Specially designed elec­tronic sunshine recorders (using photodiodes) detect when the beam is above the

threshold (Fig. 1.4). These modern, automated devices have a much finer time reso­lution, a far more precise threshold, and eliminate the daily burden of replacing the special card used by CS instruments and of manually analyzing the burnt trace to estimate the daily hours of sunshine. These advantages considerably improve the re­liability, value, and accuracy of this measurement. Side-by-side experimental com­parisons, however, have demonstrated that there are significant and non-systematic differences between the crude CS sunshine data and the more refined electronic sunshine data. This prevents the replacement of older instrumentation at many sites with long records, due to the unwanted discontinuity in climatological sunshine trends that such a change produces. Considering the limited value of sunshine data compared to irradiance data, the former type of measurement is now considered ob­solete and has essentially lost its role in atmospheric research. Consequently, some countries (such as the USA) have already stopped its routine measurement.