Instrumentation for measuring solar radiation reaching the ground falls into a few main classes, which have been described at length in specialized literature and reports. The earliest reliable device for quantitative measurements of solar radiation was a pyranometer (Kerr et al., 1967), which converted the sun's photon energy into heat and then into a voltage or current proportional to the energy-weighted integrated solar photon flux (Fig. 5.1(a)). Pyranometers are still deployed at many sites, some of which have the longest broadband data records for well-calibrated highly stable instruments. The basic pyranometer measurement is the solar irradiance in watts/m over a hemispherical field of view, which includes both direct and diffuse (scattered) photons, and a wavelength range from 280 nm to 3,000 nm. The detector is a black-coated thermopile with a nearly perfect cosine response for photons incident at different angles. The pyranometer's stability makes it ideal for detecting small long-term changes in solar irradiance caused by changes in the atmosphere (clouds and aerosols), changes in surface reflectivity, and of course, changes in solar output. The stability suggests an application for determining the amount of UV irradiance by combining pyranometer data for cloud transmission estimates with an instrument capable of determining the column amount of ozone Q, and radiative transfer calculations. Q is usually determined from the ratio of two or more measured narrow-band narrow FOV direct-sun viewing solar irradiances, and more recently by matching the entire absorption spectrum for a moderate wavelength range. This reduces two calibration problems: (1) the need for absolute radiometric calibration, and (2) the need for determining UV measuring instrument's cosine response to direct and diffuse radiances from different angles. While the combination is not a direct measurement of UV irradiance, it offers a solution to difficult calibration and stability problems inherent in direct measurements.

All other instruments that attempt to directly measure the direct plus diffuse photons (known as global flux) have a common calibration problem; namely, determining cosine response to the angle of the incident photons on a diffuser material (flat plate or some specially shaped diffuser). Most instruments rely on a laboratory-derived correction to their approximation of the ideal cosine response (e.g., Brewer spectrometers) instead of achieving a nearly perfect cosine response from hardware. An example of the latter is given by a commercially available UV spectrometer (290 nm - 400 nm with 1 nm resolution) developed at the National Institute of Water and Atmospheric Research in New Zealand. Unprotected diffuser materials (e.g., Teflon) are usually porous and susceptable to transmission changes from atmospheric pollution and from natural aging, which affects both the radiometric calibration and the cosine response. Such instruments require periodic recalibration to correct for changes in diffuser properties, detector sensitivity, and changes in the internal optics. One popular commercial instrument, the Multi-Filter Rotating Shadowband Radiometer (UV-MFRSR), uses an exposed diffuser element (no quartz window or dome in front of the Teflon diffuser) with the result that the optical transmission decreases from pollution effects and partly recovers after every rainfall. Hand-cleaning the diffuser with alcohol is only partialy effective.

Currently deployed instruments include broadband instruments, which measure the solar radiation from all directions using a diffuser over a specific wavelength range, which is usually less than viewed by a pyranometer. Most are tailored to have a response to different wavelengths that approximates some biological or application absorption spectrum (action spectrum A(A)). The most frequently approximated action spectrum is the erythemal action spectrum, which provides a measurement of the skin reddening effect for exposure to UV-A and UV-B (McKinlay and Diffey, 1987). Next are multi-spectral filter instruments that measure solar radiation from all directions, or in a narrow field of view, using several narrow wavelength bands (usually less than 10 bands from 1 nm to 10 nm wide) at the same time or sequentially (e.g., the Yankee Environmental System Ultraviolet-Multifilter Rotating Shadowband Radiometer (UV-MFRSR), and the CIMEL Sunphotometer, Fig. 5.3). These instruments require calibration for the detector response, filter transmission function, and diffuser element (if any). All of these instruments, except for the pyranometer, suffer from out of band wavelength responses, scattered and stray light in the optics and detectors, and noise in either analog amplifiers or analog to digital converters. These problems can affect data quality and instrument stability if the instruments are not frequently monitored and calibrated.

The CIMEL sunphotometer, Fig. 5.3(a), is used by AERONET for worldwide aerosol characterization (440, 500, 675, 870, and 1,020 nm) for measuring atmospheric aerosol optical thickness, absorption, particle size, and a filter at 939 nm for measuring atmospheric water vapor. There are two additional channels (340 nm and 380 nm) that are only used for aerosol extinction optical depth. The

Figure 5.3 (a) Commercial AERONET CIMEL Sunphotometer for direct-sun and sky irradiances using two independent apertures, mounted on a sun-tracker; (b) Commercial modified UV-MFRSR (Shadowband) mounted on a fixed baseplate located at GSFC. The 300 nm channel has been replaced by 440 nm and a quartz dome is over the diffuser; (c) Commercial Brewer double monochromator measuring from 270 nm - 365 nm with a 0.5 nm resolution. This Brewer has been modified at GSFC with a curved quartz window and depolarizer to permit accurate measurements of sky radiances; (d) Miniature commercial spectrometer measuring from 270 nm - 525 nm with 0.5 nm resolution. The spectrometer is connected to an optical head by a fiber optic cable. This, or other commercial spectrometers, form the basis of the Pandora spectrometer system

Figure 5.3 (a) Commercial AERONET CIMEL Sunphotometer for direct-sun and sky irradiances using two independent apertures, mounted on a sun-tracker; (b) Commercial modified UV-MFRSR (Shadowband) mounted on a fixed baseplate located at GSFC. The 300 nm channel has been replaced by 440 nm and a quartz dome is over the diffuser; (c) Commercial Brewer double monochromator measuring from 270 nm - 365 nm with a 0.5 nm resolution. This Brewer has been modified at GSFC with a curved quartz window and depolarizer to permit accurate measurements of sky radiances; (d) Miniature commercial spectrometer measuring from 270 nm - 525 nm with 0.5 nm resolution. The spectrometer is connected to an optical head by a fiber optic cable. This, or other commercial spectrometers, form the basis of the Pandora spectrometer system instrument automatically determines and tracks the sun's position, and collects data in automated sequences of measurements, which include almucantar and principal plane with a 1.2° FOV. The CIMEL sunphotometers are calibrated at a central facility at Goddard Space Flight Center (GSFC) in the laboratory and against standard CIMELS, which are calibrated at Mauna Loa, Hawaii using the Langley method (Slusser et al., 2000). While the AERONET CIMELs have been extremely successful in determining aerosol properties (optical depth, absorption, and particle size distribution) in the VIS wavelengths, they cannot determine aerosol absorption at UV wavelengths.

The UV-MFRSR (Fig. 5.3(b)) measures global radiation in seven wavelength bands (300, 305, 311, 317, 325, 332, 368 nm with 2 nm bandpass) using a cosine approximating diffuser assembly over seven separate detectors. The instrument measures whole sky (2n steradians) with and without the sun blocked by a curved metal band, which permits estimating the direct sun irradiance from the difference. The U.S. Department of Agriculture UV-B Monitoring and Research Program (UVMRP) has developed and maintains the largest network of UV-MFRSR's (over 36), which are field calibrated using the Langley method (Slusser et al., 2000) to measure global and direct sun irradiance and derive aerosol optical depths and ozone amounts from the short wavelength channels (Goering et al., 2005). While the primary purpose was for the furtherance of agricultural productivity in the US, the network has also provided a long-term database useful for irradiance related human health studies and satellite validation of ozone and UV irradiance estimates along with other traditional Brewer and Dobson spectrometers (Gao et al., 2001).

The combination of global and direct-sun irradiance can be used to deduce UV aerosol attenuation and absorption optical depths when the UV-MFRSR calibration is matched to the CIMEL sunphotometer. At GSFC and Colorado State University, the UV-MFRSR has been modified to improve its performance for aerosol detection by replacing the 300 nm channel with one for 440 nm to match one of the CIMEL sunphotometer's almucantar wavelength channels. With this modification, the UV-MFRSR can extend the CIMEL aerosol characterization into the UV wavelengths while matching the CIMEL results at 440 nm. The combined use of the CIMEL and the modified UV-MFRSR resulted in finding that the absorption of many urban aerosols have spectral dependencies in the UV that are different than those extrapolated from the visible CIMEL channels (Krotkov et al., 2005; Cede et al., 2006). The differences can be used to identify whether the aerosols are black carbon or the more strongly absorbing organic hydrocarbons. The results can also be used to calculate the reduction of UV irradiance by absorbing aerosols and to improve the estimation of tropospheric photolysis rates.

While useful work is still being done with filter spectroradiometers and broadband instruments, much more information can be derived from high spectral resolution spectrometers (e.g., the global network of Brewer spectrometers represented in the U.S. by the NOAA-EPA network of single-grating Brewers, the NSF/Biospherical network, at NASA by a modified (polarization insensitive) double-grating Brewer (Fig. 5.3(c)) (Cede et al., 2006), and by recently developed instruments using high quality commercial Charge Coupled Device (CCD) and Complementary Metal Oxide Semiconductor (CMOS) spectrometers (Fig. 5.3(d)).

The most versatile UV-VIS instruments are full spectrometers (prism, single-and double-grating dispersion elements) that are able to measure continuously in wavelength over a specified range in both narrow (less than 2°)and whole-sky FOV. The best known of these are the widely deployed Brewer spectrometers (single- and double-scanning grating versions with a single detector), which are capable of both narrow FOV and whole-sky viewing through separate ports (Fig. 5.3(c)). The Brewer double grating instrument (283 nm - 364 nm) with resolution of about 0.5 nm full width half maximum (FWHM) is especially capable for measurements of ozone in the 300 nm - 315 nm range because of its extremely low amount of scattered or stray light (photons of one wavelength affecting the count rates of another wavelength). These instruments measure one wavelength at a time and can suffer random errors from the effects of a changing atmosphere during the duration of a measurement. This effect is minimized, but not eliminated, by a slit mask used to measure six closely spaced wavelengths in a short period of time. At GSFC, each Brewer wavelength is integrated multiple times for 0.1147 seconds, or about 0.7 seconds for all six wavelengths anywhere in the Brewer spectrometer's range. A modified version of the Brewer spectrometer has been built at GSFC, which removes the polarization sensitivity of the narrow FOV port by adding a depolarizer in front of the grating and a curved fused silica window, to replace the standard flat-plate window so that the viewing direction is always perpendicular to the window surface (Fig. 5.3(c)). This eliminates the Fresnel effect for diffuse polarized skylight. The modification permits the measurement of accurate sky radiances in the presence of an unknown amount or type of aerosol. The GSFC Brewer has successfully used measured sky-radiances to determine ozone profiles in the troposphere and stratosphere every 20 minutes throughout the day (Tzortziou et al., 2008). Another commercially available moderate size double monochromator spectrometer system has been developed in New Zealand by the National Institute of Water and Atmospheric Research (NIWA) that is technically competitive with the Brewer spectrometer. It has a wider spectral range (280 nm - 450 nm), but a coarser spectral resolution (1 nm FWHM).

There is a new class of portable narrow or wide FOV instruments that simultaneously measure all wavelengths in a specified range using a single grating as the dispersion device and a multi-pixel CCD or CMOS detector. When carefully characterized and calibrated, the narrow FOV CCD/CMOS spectrometer-based systems can measure UV and VIS radiances and derive trace gas amounts (O3, SO2, HCHO, BrO, NO2, H2O), aerosol properties (optical depth, absorption, and particle size) (Herman et al., 2009a), and ozone altitude profiles.

A small portable spectrometer based system (Pandora, Figs. 5.3(d) and 5.4(a)) that is accurate, but inexpensive, has been developed at GSFC. The CCD version consists of a temperature controlled 1/8 m symmetric Czerny-Turner commercial spectrometer (Fig. 5.3(d)) using a backthinned 2048 x 14 pixel CCD detector (270 nm to 525 nm with a resolution of 0.5 nm FWHM) coupled to a small custom-built optical head (Fig. 5.4(a), 1.6° circular FOV collimator, lens and filter wheel system, and electronics) by a 10 m UV-VIS transmitting fiber optic cable. The head is mounted on a computer controlled small sun-tracker device, which also permits pointing to anywhere in the sky (the fiber optic cable depolarizes the input). The plans, machine drawings, specifications, and software are freely available upon request.

For an accurate detection of trace gas amounts, the slit function, wavelength, and radiometric calibration must be accurately determined and maintained during operation by an active temperature control (20°C± 1°C) of the entire spectrometer. High signal to noise (SNR) and precision are achieved by averaging successive measurements (a few milliseconds exposure) over several seconds. The small automated sun tracker permits the instrument to operate in direct-sun viewing, almucantar, and principal plane modes for determination of aerosol properties in the same manner as the CIMEL, but as a continuous function of wavelength in both the VIS and UV ranges. The narrow FOV precludes measuring irradiance directly; but indirectly, the irradiance can be deduced by combining solar irradiance and sky radiance measurements with radiative transfer calculations.

A much larger portable spectrometer system (MFDOAS, Fig. 5.4(b); Herman et al., 2009a) with even higher precision for a given exposure time, but with the same accuracy, has been developed at Washington State University. It measures both sun and sky radiances (see Tables 5.1 and 5.2). The MFDOAS instrument consists of separate telescopes for direct sun or scattered sky, a filter wheel containing various filters and polarizers, a spectrometer/CCD-detector system, a pointing device, a computer, and a water-based cooler to maintain constant temperature.

MF-DOAS incorporates a modified commercial, single-pass Czerny-Turner spectrograph with a focal length of 300 mm (f/4). A 400-groove/mm diffraction grating, with a blaze wavelength of 400 nm, is used to disperse light. The spectrometer has a fixed 100 |im entrance slit, resulting in a 0.83 nm (7.8 pixels

Figure 5.4 (a) Pandora optical head and sun-tracker mounted on a tripod and attached to a small spectrometer (Fig. 5.3(d)) with a 10 m fiber optic cable; (b) MFDOAS spectrometer mounted on its base and sun-tracker. The water-cooled spectrometer is controlled by with a computer in its base

Figure 5.4 (a) Pandora optical head and sun-tracker mounted on a tripod and attached to a small spectrometer (Fig. 5.3(d)) with a 10 m fiber optic cable; (b) MFDOAS spectrometer mounted on its base and sun-tracker. The water-cooled spectrometer is controlled by with a computer in its base

Table 5.1 Pandora characteristics

Wavelength interval

270 nm - 520 nm

Spectral resolution (FWHM)

0.42 nm - 0.52 nm


8 pixels

Spectrometer slit width

50 pm

Grating line density

1200 mm 1

Grating blaze wavelength

250 nm

CCD array size (pixels)


CCD pixel size (pm)


80% full well S/N (1 sigma)


CCD dynamic range

16 bit

Minimum integration time

4 ms

FOV (sun and sky)


Table 5.2

MFDOAS characteristics

Wavelength interval

282 nm - 498 nm

Spectral resolution (FWHM)

0.83 nm


7.8 pixels

Spectrometer slit width

100 pm

Grating line density

400 mm 1

Grating blaze wavelength

400 nm

CCD array size (pixels)


CCD pixel size (pm)

13.5 x 13.5

80% full well S/N (400 rows)


CCD dynamic range

16 bit

Integration time

20 ms - 60 s

FOV (sun)


FWHM) average spectral resolution. The wavelength coverage is from 282 nm to 498 nm. The CCD is thermoelectrically cooled to - 70°C. Direct sunlight enters the instrument through a 25 cm black anodized, baffled snout with a quartz window for weather control. A biconvex lens focuses the sunlight onto an 8 cm spectralon integrating sphere that moves into the spectrometer's FOV during the direct-sun measurements. A series of baffles inside the telescope limit the direct-sun field of view to about 1.5°.The sky FOV is a narrow rectangle that is 1 ° wide. Use of an integrating sphere assures equal illumination of the optics and minimizes the effect of pointing errors, eliminating a critical source of spectral residual error. Light from the scattered sky or the direct sun passes through two filter wheels containing optional depolarizers, a spectral flattening filter, and UV-transmitting and UV-cutoff filters.

The two independently developed instruments have been compared during two 118

campaigns at GSFC and JPL's Table Mountain Facility in California (TMF) and found to agree very closely (Herman et al., 2009a). During these campaigns, both instruments measured C(NO2), the column amount of NO2, in the atmosphere using direct-sun observations of atmospheric absorption. The precision achieved was 0.01 DU (1 DU = 2.67 x 1016 molecules/cm2) with an accuracy of ± 0.05 DU compared to the minimum C(NO2) in the atmosphere of 0.1 DU. This level of precision and accuracy is sufficient to measure UV and VIS radiances, as well as trace gas amounts (O3, SO2, NO2, H2O and aerosols). The measured trace gas and aerosol amounts can be used to calculate UV irradiances and radiances over the entire 300 nm to 400 nm range.

The large dynamic intensity range as a function of wavelength can cause a well-known spectrometer system problem of stray light from longer wavelengths affecting the radiometric response of the shorter wavelengths. This problem is minimized by using bandpass filters (e.g., UV-340 filter) that block the visible and pass the UV so that ozone amounts can be measured (305 nm - 320 nm) by the same instrument that measures NO2 (400 nm - 450 nm). The best performance is obtained by use of a double filter wheel with one wheel containing neutral density ND filters, open hole, and blocked hole, and the other containing function filters (e.g., UV-340, polarization filters, open hole etc.). The blocked hole is essential for the measurement of spectrometer dark current correction in between each measurement of radiance for the same exposure time as the radiance measurement. The use of fiber optic cables to connect the optical head to the spectrometer (Pandora) has advantages (modularity, light weight, depolarization, and remote operation) and disadvantages (radiometric stability, calibration, controlling the FOV, and some fragility) compared to direct coupling of the optical head to the spectrometer (MFDOAS). While these problems are not trivial, they have been mostly overcome.

For all of these instruments, the key factors are the absolute accuracy of the calibration and the long-term stability (precision) of the instrument. Absolute accuracy is achieved by measuring the output of a standard lamp that has been verified by an appropriate standards organization (e.g., the National Institute Standards and Technology (NIST), in the U.S., or one of the European standards laboratories). The quality of the transfer of calibration is dependent on a meticulous laboratory setup between the standard lamp and the instrument in question. Currently, the best transfer of calibration produces an accuracy of between 2% and 3% under the assumption that the instrument has been fully characterized in the laboratory before the use of a standard lamp (slit functions, filter transmission functions, wavelength calibration, temperature sensitivity, stray and scattered light, etc.). The achievable accuracy will be a combination of the instrument characterization errors and the lamp transfer radiometric calibration. Note: To determine the wavelength dependent slit functions over the entire wavelength range or the filter functions, the detector must have many light sensitive elements within the projection of the grating's spectral resolution onto the detector at its full width at half the maximum height (FWHM) or the width of the filter bandpass at FWHM. For wavelength scanning spectrometers, the scanning step size must be much less than the spectral resolution FWHM to accurately determine the slit function.

Good long-term stability of the instrumentation is needed to detect possible changes in the amount of radiation reaching the ground. The stability can be determined by repeated checks against standard lamps in the laboratory or by techniques applied to the data obtained in the field. The most common of these is a procedure that attempts to derive the solar irradiance above the atmosphere by measuring solar irradiance at the ground as a function of SZA (Langley method). This gives a series of measured instrument counts as a function of the amount of atmosphere between the instrument and the sun. If the atmosphere is approximately stable over the period of the measurements, the series of measurements can be extrapolated back to zero atmospheres. The details of a modified Langley method are described in Herman et al. (2009a). Repeating this procedure frequently for the life of the instrument can give a measure of the instrument's stability or precision. The best Langley procedures are performed at extremely clean sites such as Mauna Loa, Hawaii, where the interference from changing aerosol and boundary layer ozone amounts is at a minimum. The combination of frequent standard lamp calibration and in the field Langley procedure provides a precision between 2% and 3%. Some instruments, such as a well-maintained double Brewer spectrometer, are known to be very stable against standard lamps and can produce a radiance precision near 1% if the instrument is modified to be polarization insensitive (Fig. 5.3(c)).

In addition to ground-based instruments, there have been several spectrometers located on satellites since 1979, which are able to provide accurate estimates of ozone and cloud transmission T (through measured cloud and aerosol Lambert Equivalent Reflectivity R (Herman et al., 2001a; 2009b; see Section 5.3.4), where T is approximately 1 -R). The estimate for T has been improved by detailed radiative transfer solutions for plane parallel clouds of different optical thickness. The resulting cloud transmission factor CT (Krotkov et al., 1998) gives results that are close to the simple expression CT = (1 - R)/(1 - RG), where RG is the measured surface reflectivity and RG < R < 1 (Herman et al., 2001b). This expression for CT has been given by Krotkov et al. (1998), and can be derived from a Stokes calculation (see Section 5.3.5) for a transmitting and reflecting surface over a reflecting ground (Herman et al., 2009b). Ultraviolet and VIS irradiance, at ground-level or within the atmosphere, can be accurately estimated using radiative transfer calculations that include the appropriate absorbing gases, aerosol scattering and absorption, surface reflectivity, and cloud transmission. Frequently, the calculated amount of radiation is overestimated by 10% to 20% because of the lack of knowledge regarding aerosol absorption, even when the aerosol optical depth is known, especially in urban areas (Herman et al., 1999).

Instrumental requirements for making long-term UV irradiance measurements are well understood in terms of calibration and stability for both spectrometers and broadband radiometers (WMO, 1999; 2003; 2007). The degree to which further improvements are made to a given instrument type should be dictated by the scientific goals of an application. For example, different accuracy and precision are needed for estimating transmission through the atmosphere in the presence of trace absorbing gases, changes in UV and VIS radiances and irradiances needed for estimating changes in atmospheric chemistry photolysis rates, climate change studies, or for use in various applications (estimations of solar radiation amounts vs. cancer incidence, material damage, atmospheric photochemistry, and biological productivity).

Biological or material damage applications are usually expressed in terms of "action" spectra, which are an estimate of the relative strength of a given process (e.g., skin reddening) for a given wavelength per unit of incident solar radiation. The biological action spectra are usually poorly known and have a wide variation within a species depending on skin or surface type, genetics, species adaptation to its local environment, and from species to species. In other words, knowing the solar radiation change to within a few percent, or even 10%, is good enough for most biological and materials damage process studies. Of more importance is determining the changes in exposure to solar radiation either in the same location or by moving to lower latitudes where there is usually more solar irradiance.

Higher accuracy and precision are needed for understanding physical processes, such as contributions to atmospheric energy balance related to global warming or for estimations of photolysis rates for atmospheric chemistry studies. The estimated surface heat imbalance between the surface re-radiation and the atmosphere is about + 9 watts/m2 (Lin et al., 2008), so that a contribution of 1 watt/m2 would be a significant portion of the uncertainty. For the UV range from 300 nm to 400 nm, there are approximately 100.7 watts/m2 at the top of the atmosphere and about half that at the surface, so that detecting a 1-watt/m change in the UV band over a decade would require 2% long-term precision.

Was this article helpful?

0 0

Post a comment