Previous sections have described the sources of errors that affect solar radiation measurements. During the last decade, the results from high-quality research in ra – diometry have considerably improved our understanding of these errors and have provided ways to correct them or avoid them altogether. For instance, Eq. (1.6) above provides an effective method to correct a specific pyranometer for its cosine response error (but not necessarily thermal offset errors).
Another way of looking at the cosine error problem has resulted from the consideration that it is mostly caused by direct irradiance, the main component under clear skies. As mentioned above, direct irradiance measurements have a lower uncertainty than global measurements. Hence the development of a method derived from the component-summation technique described in Sect. 5. In such a setup, the global irradiance is calculated as the sum of the measured diffuse and direct irradiances according to Eq. (1.1). An unshaded pyranometer is still useful, first to obtain an independent measurement of global irradiance (which might be needed in case of tracking problems, etc.), and second, for quality assurance (by experimentally verifying that Eq. (1.1) is respected). The gain in accuracy is significant, assuming high-quality instrumentation and maintenance, typically «15Wm~2 (Michalsky et al. 1999).
This technique requires that diffuse irradiance be measured properly. This implies a ventilated instrument to homogenize temperatures and avoid condensation or frost on the dome. Furthermore, the thermal offset must be minimized (less than «2Wm~2), by using a pyranometer with either a black-and-white sensor or an allblack sensor with proper correction. Some correction techniques have been proposed (Bush 2000; Dutton et al. 2001; Haeffelin et al. 2001; Philipona 2002), based on the observed relationship between thermal offset and the nighttime net infrared (IR) radiation balance between the instrument and the sky. Accurate diffuse measurements with an all-black sensor require monitoring upwelling and downwelling IR by a colocated pyrgeometer. The comparison of the performance of various types (all-black and black-and-white) of shaded pyranometers has been the subject of a series of experiments (Michalsky et al. 2003, 2005, 2007; Reda et al. 2005). This research has resulted into a proposed working standard for the measurement of diffuse irradiance.
What is the practical significance of all these recent changes in measurement procedures? Only modest improvements in direct irradiance result from the adoption of windowed ACRs or pyrheliometers with low environmental influences on their signals. Conversely, large improvements are obtained if a pyrheliometer is used in lieu of the common indirect method where DNI is obtained by computation from global and diffuse data, through application of Eq. (1.1). This is particularly the case if diffuse irradiance is measured with a shadow band, and if the pyranometers are not corrected for cosine errors and thermal offset (Gueymard and Myers 2007). The implementation of optimal techniques for the measurement of diffuse and global irradiance also results in significant improvements, particularly under clear skies in winter (Gueymard and Myers 2007). Therefore, the development of empirical solar radiation models, or the validation of any type of radiation model, should only be based on optimal data.
All the modifications described above to the conventional measurements of diffuse and global irradiance (which were the norm only ten years ago) have induced noticeable effects on the operation of research sites. More sophisticated equipment is required, with redundancy and higher measurement frequency (e. g., 1 minute), and increasing cost. Calibration and maintenance are more stringent, requiring more skilled personnel. Finally, efficient techniques for quality assessment and dissemination of the measured data need to be established. These new constraints require significant resources, limiting the commissioning of these high-quality radiation sites to only a few in the world. These research-class, high-end sites have been made possible because of their key role in the current climate change context, in which the radiative forcing of the climate must be precisely understood and predicted. The next section gives an overview of the conventional and high-end networks that exist in the world.