Rotating Shadowband Spectroradiometer Research Articles

Moderate spectral resolution solar irradiance measurements, aerosol optical depth, and solar transmission, from 360 to 1070 nm, using the refurbished rotating shadow band spectroradiometer (RSS)

Abstract. This paper reports on a third-generation rotating shadow band spectroradiometer (RSS) used to measure global and diffuse horizontal plus direct normal irradiances and transmissions at 1002 wavelengths between 360 and 1070 nm. The prism-dispersed spectral data are from the Atmospheric Radiation Measurement (ARM) Southern Great Plains site in north-central Oklahoma (36.605∘ N, 97.486∘ W) and cover dates between August 2009 and February 2014. The refurbished RSS isolates the detector in a vacuum chamber with pressures near 10−7 torr. This prevents the deposition of outgassed vapors from the interior of the spectrometer shell on the cooled detector that affected the operation of the first commercial RSS. Methods for (1) ensuring the correct wavelength registration of the data and (2) deriving extraterrestrial responses over the entire spectrum, including throughout strong water vapor and oxygen bands, are described. The resulting data produced are archived as ARM data records and include cloud-screened aerosol optical depths, spectral irradiances and direct normal solar transmission, as well as normalized diffuse and global irradiances.

The SMARTS spectral irradiance model after 25 years: New developments and validation of reference spectra

A detailed literature survey of the diverse applications in which the SMARTS spectral radiation model has been involved during the last 25 years first provides an overview of the model’s influential status over a broad range of scientific disciplines. An historical perspective gives details about the founding principles and evolution of the model. Its main modeling algorithms, regarding both absorption and scattering from various atmospheric constituents are described, highlighting the new developments in the model’s latest version (2.9.8), and the possible verification of its atmospheric transmittance functions with observations from a Rotating Shadowband Spectroradiometer. The addition of new dust and smoke aerosol models now allows the quantification of the circumsolar irradiance within 10° of sun center for a wide range of scattering conditions. The circumsolar ratio for a 2.5° aperture angle is found ≈5 times larger with dust aerosols than with rural aerosols. The impact on the modeled spectral or broadband irradiance of selecting the newer default extraterrestrial (AM0) spectrum is found generally low, except in the UV where differences of ≈5% can be expected. The possibility of experimental realizations of the AM1.5D and AM1.5G standard reference spectra is investigated using spectroradiometers and other instruments at a high-elevation station in Colorado. The impacts on direct and global irradiance of using different input streams, different types of spectroradiometer, and two versions of SMARTS, are quantified. Usage of the SMARTS v2.9.8 is recommended for all demanding applications, possibly including future revisions of the standard reference spectra.

Parametrization of solar spectral irradiance between 300 and 1100 nm for photovoltaic devices

Most solar cells use the solar irradiance at a spectral region between 300 and 1100 nm to generate electricity. We refer to this irradiance as a cell band irradiance (CBI). However, solar energy power systems use the global horizontal irradiance (GHI) to manage the electricity generation rather than the CBI. This is because the CBI is generally not available from either observation or radiation modelling. We conduct a theoretical investigation to show that the ratio of the CBI to GHI varies significantly with atmospheric conditions, which means that using the GHI to manage the electricity generation will introduce a large uncertainty. In this article, we develop a parametrization of irradiance ratio as a function of water vapour, solar zenith angle, aerosol and cloud optical depth, which can be used to convert the GHI to the CBI used by solar cells. We further evaluate this parametrization using the high spectral resolution Rotating Shadowband Spectroradiometer (RSS) data measured at the Southern Great Plains (SGP) site of the US Department of Energy Atmospheric Radiation Measurement (ARM) programme. The results show that the parametrization is reasonably accurate in both clear‐ and all‐sky conditions. The relative mean biases for the estimated clear‐sky GHI and CBI are 1.50 and −0.44%, respectively. The relative r.m.s. differences are 7.38 and 7.68%. The relative mean biases for the estimated all‐sky GHI and CBI are −4.25 and −3.31%, and the relative r.m.s. differences are 16.04 and 18.48%, respectively.

Using EOF analysis to qualitatively analyze, and identify inhomogeneities in, data from ground‐based aerosol monitoring instruments

Empirical Orthogonal Function (EOF) analysis is performed on ground‐based shadowband radiometer and Sun photometer data. The data come from Multifilter Rotating Shadowband Radiometer (MFRSR) instruments located at the central and extended facilities of the Southern Great Plains (SGP) research site of the Department of Energy's Atmospheric Radiation Measurement (ARM) program, as well as the Rotating Shadowband Spectroradiometer (RSS) 102 and 105 and the CIMEL Sun photometer CSPHOT located at the central facility at SGP. The EOFs show the variability in total aerosol optical depth and provide some qualitative information on the separation of the coarse and fine aerosol modes. In addition, as has been demonstrated previously with satellite data, EOF analysis also exposes several flaws and inconsistencies within the ground‐based data sets. These inhomogeneities include optical depth anomalies in some MFRSR filters, wavelength shifts in the RSS, and problems with the data transmitter of the CIMEL instrument. Therefore EOF analysis is shown to be a quick and effective means not only of assessing the general aerosol behavior in the air above a particular monitoring instrument, but also of identifying both known and unanticipated influences on the data coming from within the instrument itself.

Photosynthetically active radiation from Clouds and the Earth's Radiant Energy System (CERES) products

We describe a method that retrieves surface photosynthetically active radiation (PAR) and its direct and diffuse components from the Surface and Atmospheric Radiation Budget (SARB) product of Clouds and the Earth's Radiant Energy System (CERES). The shortwave spectrum in the SARB Edition 2 is calculated in 15 bands, 4 of which are used to develop the PAR, in conjunction with the look‐up tables described in this paper. We apply these look‐up tables to existing CERES Terra Edition 2 products. The new retrieved surface PAR is validated with LI‐COR PAR measurements at seven Surface Radiation Budget Network (SURFRAD) sites using data from March 2000 to June 2005. The relative bias of retrieved all‐sky PAR at the SURFRAD sites is 4.6% (positive sign indicating retrieval exceeds measurement), and 54% of the all‐sky samples are within the ±10% uncertainty of the LI‐COR PAR measurements. The satellite field‐of‐view (FOV) is more representative of the ground instrument FOV under clear conditions, so 89% of clear‐sky retrievals are within the uncertainty of the LI‐COR PAR measurements at SURFRAD sites with positive biases at most sites. The retrieved PAR is also validated at the Atmospheric Radiation Measurement (ARM) Southern Great Plains Central Facility (CF) site using data from October 2003 to June 2004 for those FOVs having both LI‐COR and Rotating Shadowband Spectroradiometer (RSS) ground measurements; for this small domain, all‐sky relative biases are again positive (1.9%) for LI‐COR but negative (−4.2%) for RSS. The direct‐to‐diffuse ratio derived from CERES is smaller than that from RSS for both clear and cloudy conditions. CERES also retrieves the broadband shortwave insolation, and the relative biases for the broadband retrievals are much less than those for PAR at the above sites. It appears that some of the ground‐based measurements of PAR do not have the fidelity of those for broadband shortwave insolation.

Aerosol retrievals using rotating shadowband spectroradiometer data

The rotating shadowband spectroradiometer (RSS) is a high‐resolution device that measures the total, direct, and diffuse intensity of sunlight at 1016 different wavelengths. RSS data can be used to retrieve gas amounts and aerosol properties as well as to assess the accuracy of retrievals using data from lower‐spectral resolution instruments, such as the multifilter rotating shadowband radiometer (MFRSR). An algorithm to retrieve aerosol and gas amounts and the aerosol size distribution from RSS data, using the full resolution and using selected wavelengths, has been developed (Gianelli, 2004). The results of the retrievals, applied to RSS data from the southern Great Plains (SGP) site, indicate a number of things about the aerosol size distribution and our ability to retrieve aerosol information accurately. First, we show that the aerosol size distribution at SGP is at least bimodal. The notion that a more complex size distribution can be modeled with an appropriately selected monomodal distribution produces an unacceptable fit to the data once the separation of aerosol extinction from nitrogen dioxide absorption at short wavelengths is taken into consideration. Second, we find that the amount of retrievable aerosol information is limited by the wavelength range of the data. This is exemplified by the indeterminacy of the coarse‐mode effective radius and the interdependence in the retrievals of the fine‐mode effective radius and effective variance. Third, when the fine‐mode effective variance is constrained, an annual cycle in fine‐mode effective radius values emerges from the retrieval results, with a maximum in March and a minimum in September. Conceivably, changes in the effective variance could influence the observed pattern as well. The large margin of error in the coarse‐mode effective radius leads to relatively small error bars for the coarse‐ and fine‐mode optical depths, the fine‐mode effective radius, and ozone, except on those days when the coarse‐mode optical depth is high. Examining the retrieval results for different wavelength combinations of five RSS channels allows us to investigate whether or not the MFRSR channels are optimized to retrieve aerosol information or if a different filter set would increase the robustness of the retrievals. We show that replacing the 670 nm channel with one at either 375 or 1034 nm improves the retrieval accuracy. In particular, retrievals with the 1034 nm channel very closely reproduce the full RSS retrieval results, provided that the NO2 column amount can be determined by other means. The precision of the retrieved values of the fine‐mode effective radius is shown to be strongly sensitive to the precision of measured NO2 amounts; a margin of error in column NO2 of 0.3 Dobson units results in a corresponding margin of error of 0.04 μm in the fine‐mode effective radius. This confirms that aerosol sizes cannot be inferred accurately from optical depth spectra alone if either the amount of NO2 above a site is unknown or an inaccurate value is assumed.

Extraterrestrial solar spectrum 360–1050 nm from Rotating Shadowband Spectroradiometer measurements at the Southern Great Plains (ARM) site

Two years of Langley extrapolations made from the Southern Great Plains (SGP) site of the Atmospheric Radiation Measurement (ARM) Program using two very different RSS instruments and a NIST‐derived irradiance scale show larger extraterrestrial solar irradiances in the 400 to 600 nm domain by as much as 4.5% compared to the Neckel and Labs [1984] data. Our results are more congruent with that of Thuillier et al. [1998a, 1998b] in this domain but do not show larger irradiances than the Neckel and Labs data at wavelengths less than 400 nm. In addition to the question of the true extraterrestrial solar spectrum, the data we present here are directly useful for the interpretation of spectral measurements calibrated with current National Institute of Standards and Technology (NIST) irradiance standards, for which we believe this represents the most extensive determination of the solar irradiance.

Column water vapor from diffuse irradiance

We investigate the possibility of measuring water vapor column from diffuse irradiance, and thus the extension of optical retrievals to cloudy days. Data from the rotating shadowband spectroradiometer (RSS) during its winter deployment at the North Slope of Alaska (NSA) site are used: 20 days in March 1999 that include clear, partly cloudy and overcast days. During these days water vapor column varied between one and five mm according to the NSA site's microwave radiometer (MWR). The diffuse transmittance in the 760‐nm oxygen absorption band is used to obtain the effective air mass. The end result is a correlation of 0.97 and 0.95 between MWR and RSS retrievals using 820‐nm and 940‐nm diffuse transmittances, respectively. These results, which are based on empirical data analysis only, imply that the diffuse irradiance may contain sufficient information to retrieve water vapor column.

Impact of surface inhomogeneity on solar radiative transfer under overcast conditions

The goal of this study was to assess the ability of the Moderate‐Resolution Transmittance 4 (MODTRAN‐4) code to simulate high‐resolution shortwave (SW) fluxes given detailed and complete input information under overcast conditions. The study underlines the impact of surface inhomogeneity on the closure of SW radiative transfer. It also leads to a method of estimating surface spectral areal‐mean albedo from downwelling solar transmittance measurements. The investigation made use of ample Atmospheric Radiation Measurement (ARM) field data collected by a suite of instruments, including broadband and narrowband radiometers and spectrometers, cloud radar and lidar, microwave radiometer, atmospheric sounding instruments, and satellite data. Furnishing the MODTRAN‐4 code with observed atmospheric, cloud, and surface parameters generates spectral solar transmittance at the surface and reflectance at the top of the atmosphere (TOA). The transmittances were compared with the Rotating Shadowband Spectroradiometer measurements and showed significant discrepancies in the near‐infrared (NIR) region, the bulk of which was attributed to the use of unrepresentative surface spectral albedos. A field campaign was undertaken to collect surface albedo data for a wide variety of land cover types near the ARM Central Facility. The sampled data were combined with thematic mapper/Landsat‐based land cover classification data to map surface spectral albedo. Substitution of the derived areal‐mean spectral albedo into the MODTRAN‐4 model eliminates major discrepancies in the NIR, and also leads to good agreements with surface solar broadband fluxes and TOA satellite spectral reflectance. On the basis of these findings, one may use downwelling spectral transmittance data, together with detailed cloud and atmospheric information, to estimate surface effective areal‐mean albedo. The estimated values agree well with those derived from the ground survey data. Following the method, a data set of effective areal‐mean spectral albedo throughout a year was obtained.

A robust retrieval of water vapor column in dry Arctic conditions using the rotating shadowband spectroradiometer

A method to retrieve water vapor column using the 940‐nm water vapor absorption band in dry Arctic conditions is presented. The retrievals with this method are stable with respect to uncertainties in instrument radiometric calibration, air pressure, solar source function, and aerosols. The water vapor column was retrieved with this method using spectra obtained with the rotating shadowband spectroradiometer (RSS) that was deployed during an intensive observation period near Barrow, Alaska, in March 1999. A line‐by‐line radiative transfer model was used to compute water vapor transmittance. The retrievals with this method are compared with retrievals obtained from three independent measurements with microwave radiometers. All four measurements show the same pattern of temporal variations. The RSS results agree most closely with retrievals obtained with the millimeter‐wave imaging radiometer (MIR) at its 183 GHz±7 double‐side band channel. Their correlation over a period of 7 days when water vapor column varied between 0.75 mm and 3.6 mm (according to RSS) is 0.968 with MIR readings 0.12 mm higher on average.

A differential technique to retrieve column water vapor using sun radiometry

Techniques for retrieving column water vapor from Sun radiometer measurements involving the 940‐nm water vapor absorption band have been around for the better part of a century. Arguably, the best method to use for this retrieval is the modified Langley technique. However, to apply this method one must obtain the instrument response at the top of the atmosphere using modified Langley plots on clear days with a very stable water vapor column. Using subsequent measurements in this filter, ratioed to the top‐of‐the‐atmosphere response allows one to determine the transmission in the 940‐nm water band. In this paper, we present an approach that does not require an absolute knowledge of the extraterrestrial instrument response. The method discussed here relies, instead, on relative measurements of a calibration lamp and the extraterrestrial spectral irradiance within and just outside the 940‐nm absorption band. We execute these retrievals for the rotating shadowband spectroradiometer (RSS) on 3 days during the Department of Energy's Atmospheric Radiation Measurement program's 1997 Water Vapor Intensive Observation Period. We compare the results to those retrieved from a colocated multifilter rotating shadowband radiometer (MFRSR) that uses an empirical calibration and from a colocated microwave radiometer. Since our optical method of retrieving column water vapor from RSS measurements does not depend on a calibration performed against another water vapor measurement, it contributes an independent estimate in the search for absolute accuracy. The major contributors to the uncertainty of this retrieval are the water vapor band strength calculations, the difference in aerosol extinction in and near the water vapor band, the relative spectral irradiance output of the calibration lamp and the Sun at the nonabsorbing and band‐centered wavelengths, and the stability of the spectral response of the instrument, which will be discussed in detail.

Joint statistics of photon path length and cloud optical depth: Case studies

We show the joint statistics of photon path length and cloud optical depth for cloudy sky cases observed at the Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP) site between September and December 1997. The photon path lengths are retrieved from moderate resolution oxygen A‐band observations taken by a rotating shadow band spectroradiometer (RSS). For high optical depth cloud cases, two different populations in the scattergram of the path length versus cloud optical depth are apparent. One population is a result of single‐layer cloud cases that exhibit a small variation of path length enhancement over a large optical depth range, together with a strong correlation between the radiation field and the cloud liquid water path, while the second population is attributed to multiple‐layer cloud cases with large variability of enhanced photon path lengths. When the optical depth is less than 5, the population of cases appears to bifurcate as the solar air mass increases, with the lower branch exhibiting pressure‐weighted path lengths shorter than the direct beam path lengths at these larger solar zenith angles. Using information from a millimeter‐wave cloud radar, together with lidar and balloon‐borne sonde data to further analyze these cases demonstrates that this bifurcation is caused by the altitude of the scattering; thin clouds aloft produce the lower branch and low‐level aerosols produce the upper branch.

Comparison of spectral direct and diffuse solar irradiance measurements and calculations for cloud‐free conditions

Ground‐based spectral measurements of direct and diffuse solar irradiance from the Rotating Shadowband Spectroradiometer, taken in cloud‐free conditions in Oklahoma in the fall of 1997, are compared over the spectral range 10000–28500 cm−1 to corresponding calculations by an accurate multiple‐scattering radiative transfer model. For each case analyzed, the aerosol optical depths used in the calculation were determined by fitting an Angstrom relation based on the ratio of the direct‐beam measurements to a direct‐beam calculation with no aerosols present. Also used in the calculation was a spectrally‐independent aerosol single‐scattering albedo chosen to provide agreement with the diffuse measurements. The spectral agreement between the measurements and calculations for the direct and diffuse irradiances is very good, providing strong evidence that in this spectral range there are no unmodeled molecular absorbers of significance to the atmospheric energy balance. Especially notable is the correspondence between the observations and calculations for a case characterized by a large amount of water vapor in the direct‐beam path, directly contradicting the suggestion that water vapor absorbs more shortwave radiation than is represented in radiative transfer models.

O2‐O2 absorption band identification based on optical depth spectra of the visible and near‐infrared

The first implementation of the rotating shadowband spectroradiometer allows the routine collection of direct normal solar spectral irradiance data at 512 wavelengths. These data are used on clear days to calculate spectra in optical depth. A persistent absorption band near 477 nm in the first optical depth spectra collected led to the eventual identification of six bands in the 400 to 1080 nm range resulting from absorption by O2 collision pairs. All of these O2‐O2 bands were discovered earlier using atmospheric long pathlength or laboratory high pressure techniques, however, it is notable that this method's sensitivity permitted their detection at one air mass. These data suggest that there are no unexplained narrowband absorption features in the visible/near‐infrared, that excess optical depth measured near 1000 nm may now have an explanation, and that the linearity of the O2‐O2 absorption may be exploited to investigate mean pathlengths in the atmosphere.

Joint statistics of photon pathlength and cloud optical depth

A mean pressure‐ and temperature‐weighted photon pathlength in the atmosphere can be inferred from moderate resolution measurements in the O2 A‐band. We show a pathlength retrieval method and calibration results for measurements from a Rotating Shadowband Spectroradiometer (RSS), and present the joint statistics of pathlength and cloud optical depth for cloudy skies observed at the Southern Great Plains (SGP) site from Sept. 30 to Dec. 22, 1997. Two different population branches are apparent in the scattergram of the pathlength vs. cloud optical depth; we attribute these to 1) single layer cases exhibiting small variations of pathlength enhancement over large optical depth ranges; 2) multiple layer cases with large variances of enhanced photon pathlengths.