A range of air mass values is important because air mass is the path length of the solar beam through the atmosphere (see Figure 3-1}. Air mass increases with increasing solar zenith angle and thus changes with location (latitude), season, and time of day. As air mass (the path length) increases, there is increased spectral absorption and scattering of solar radiation by atmospheric constituents, such as aerosols and water vapor, and the spectral content of solar radiation at the collector surface changes.
The most important atmospheric variables that determine broadband and spectral solar radiation are cloud cover, atmospheric turbidity (aerosols), precipitable water vapor, and barometric pressure (number of air molecules in the path of the solar beam). To properly document the spectra, it is important to measure or characterize these variables. Several methods can be used, depending on the instruments available at the measurement sites:
The spectral solar radiation measurements were made in several different measurement modes corresponding to solar collector configurations. Direct normal solar radiation (radiation from the solar disk in about a 5° field of view) is used by concentrating (focusing) collectors; global-normal (direct plus scattered radiation on a surface normal to the solar disk) is used by two-axis, sun-tracking, flat-plate collectors; and global-tilt is used by fixed-tilt flat plates, such as south-facing collectors tilted at the latitude angle, or single- axis tracking flat plates. Global-horizontal (direct plus scattered radiation on a horizontal surface) is used to develop and test models that convert global radiation to radiation on a surface of any orientation, such as building walls and windows; the photosynthetically active region of the spectrum is also important for biomass applications. The diffuse spectra (sky radiation, with the solar disk blocked) included in the data base were measured as part of SERI's research data collection to examine air pollution effects in the Denver/Golden area.
Spectral solar radiation data were acquired using spectroradiometers with the following characteristics:
The spectroradiometers were calibrated at SERI every six months against standard lamps traceable to the National Institute of Standards and Technology [NIST, formerly National Bureau of Standards (NBS)]. In this process, the spectroradiometers were calibrated in the laboratory and then compared with one another outdoors (Figure 3-5) to determine spectral uncertainty limits, precision (or repeatability) errors, and bias error estimates.
At the measurement sites, the instruments were checked monthly against a reference lamp (LI-COR Optical Radiation Calibrator) to monitor instrument stability; these results were reported to SERI. (Measurement uncertainty is reported in detail in references 15, 16 and a summary is given in Section 5.)
Table 3-1 lists the data collected to calculate or specify the atmospheric conditions and properly document each spectrum in the data base. The broadband solar radiation data were measured to correspond with the different spectroradiometer measurement modes and to calculate atmospheric descriptors, such as Kt. Ground albedo was included because it affects both solar radiation measured on tilted surfaces and ground-to-sky reflections (especially under cloud cover). Wind speed and ambient temperature were recorded because they may be useful, along with the spectral data, to predict collector performance. The broadband and meteorological sensors were maintained and calibrated by the measurement sites as documented in Section 5.
Broadband solar radiation measurements were made immediately (within 15s) before and after the spectral solar radiation scan to allow evaluation of atmospheric stability at the time of the scan. In addition, the data-acquisition software was designed so that global- horizontal solar radiation was sampled six times during the spectral scan using a silicon-detector pyranometer; if global solar radiation varied by more than 2%, the spectral scan was not recorded. Three attempts were made to acquire a spectral scan during unstable conditions (such as partly cloudy skies). If a scan was not obtained after three tries, the broadband and meteorological data were recorded and then the data acquisition system waited until the next regularly scheduled data acquisition time (usually an hour) to attempt another spectral scan. Because of the stability monitoring during the spectral scans, we believe there are very few cases when the spectral scan was acquired during variable solar radiation conditions. We found three examples of distorted spectra (see Figure 3-6) that possibly were caused by unstable conditions.
Software to test the atmospheric stability and acquire the data in a specific format was integrated into the field operations by FSEC and PG&E. The format included fields for site identification, date and time, broadband solar radiation values, atmospheric data, spectral solar radiation data, and comments. In addition, an instrument description and configuration file was stored for each data file.
The data acquisition software was designed to accept manually entered data from sunphotometers, cloud-cover estimates, and National Weather Service precipitable water vapor measurements. However, the personnel, instruments, and measurements were generally not available to enter these data each hour at the measurement sites.
The schedule for data collection was flexible to accommodate individual site requirements and constraints. The goals, which were accomplished, were to collect data over at least one year at each site to capture seasonal variations and to cover a wide range of atmospheric conditions for each measurement configuration.
Measured spectra and supporting data were sent to SERI on floppy disks from PG&E and on magnetic tape from FSEC. Daily log sheets that describe the spectroradiometer operation and operator comments were included with the data. All-sky photographs corresponding to many of the FSEC spectral data sets were provided by the FSEC Fenestration Energy and Illumination Performance Research Program (Figure 3-7). All-sky photographs were also taken for each of the SERI spectral data sets (Figure 3-8). A camera system was installed at PG&E late in the data collection period, but no photographs are available.
At SERI, the data were processed through an interactive quality-control procedure during which the data were visually inspected for obvious problems. First, the instrument configuration file (see example in Figure 3-9) was examined against a template from the previous data set to check for any instrument changes or a change in the mode of operation. After the operator approved the configuration table, the template was updated and the information was written to the data base.
Figure 3-10 shows an example of the computer terminal output during the second step of the quality-control session.
The information displayed (see circled numbers in Figure 3-10) included the following:
We chose to include all data in the data base and to be as specific as possible about the measurement uncertainty, rather than set stringent criteria for excluding data sets. This allows users to select data based on their specific application and accuracy requirements, or to use portions of the data sets, such as broadband data or spectral solar radiation segments. None of the data should be used without referencing the measurement uncertainty.
Table of Contents