Improved techniques for the spatial and temporal measurement of ultraviolet radiation

One of the more difficult tasks confronting atmospheric researchers today is the acquisition of long-term radiometric measurements that encapsulate variability in the sky hemisphere as well as time. High quality spatial measurements would allow for a greater understanding of atmospheric processes, both specific to the site and between locations. This information would lead to further development of better spatial models for predicting future trends. This study develops a timer-controlled variable sky-view platform (or VSP), designed to encapsulate both the spatial and temporal variation of radiation in the sky hemisphere for Hobart (42.90 S; 147.33 E). The VSP essentially allows for a sensor to move up and down inside a fixed cylinder, therefore causing the sky view factor to vary. It is currently in use with the erythemal ultraviolet (UV) biometer 501-A manufactured by Solar Light Company Inc. (SLC). As such, the VSP provides an additional source of UV sky distribution data to the Hobart UV-climatology database. Before VSP deployment of the SLC instrument, it was first necessary to fully characterise its cosine response in an accurate calibration scheme. Here, calibration of the detector is dependent upon wavelength as well as solar zenith angle, due to the discrepancy between the actual instrument relative spectral response and the erythemal response curve. In this study, measurements were compared for the purpose of characterising a new cosine response function (CRF) for the broadband detector. Concurrent measurements of spectral and broadband erythemal UV data have been undertaken at the University of Tasmania in Hobart, Australia. Measurements spanned solar zenith angles (SZA) of 21°-74°, total ozone values of 220 to 320-DU and relatively small aerosol optical depths. The CRF is both wavelength and SZA dependent and applicable to all broadband UV-B instruments based on the original Robertson-Berger design. A methodology is presented for incorporation of the CRF into the calibration scheme. The concept of spectral multipliers is introduced, which are unique to each broadband instrument and correct for the wavelength dependence of the CRF. A method is then described for ongoing calibration of the detector based on model data alone. The scheme is then tested with real data to validate its reliability. Overall, by nature of the simultaneous characterisation of both spectral and cosine response, the CRF allowed for a more robust calibration for the VSP deployed sensor. A full description of the VSP design, electronics and construction is then given. Emphasis is on high precision long-term operation under all weather conditions. A modelling scheme and output results illustrate the advantages of the VSP over current sky observing instruments. It can accurately parameterise the direct and diffuse erythemal irradiance as well as the azimuthally-integrated erythemal sky radiance distribution for every two minutes duration (at the current setting). In addition, all data can be obtained for clear-sky and cloudy conditions. Analysis of the clear-sky VSP sky distribution, in conjunction with spectroradiometer radiance measurements, allows for derivation of the azimuthally-integrated circumsolar component for each zenith angle. The thesis concludes with an application of the VSP for the all-sky calibration of other radiometric instruments. Occulting disk or shadowband arrangements are often applied to both spectral and broadband sensors in order to retrieve the total diffuse irradiance. Here, extraction of the diffuse component is simply calculated as a difference between the measured global and direct components. However, difficulty lies in the application of a suitable diffuse calibration due to the instrument cosine response. Generally, the sky distribution is assumed isotropic during the correction, whereas it has been shown in most studies to be nearly always anisotropic. Hence, without in situ sky distribution measurements it is difficult to obtain the correct total diffuse measurement. This study utilises VSP-extracted erythemal sky distribution data. Results are used to correct for the cosine error of instruments that measure diffuse erythemal radiation data, for both clear-sky and cloudy conditions. A comparison was then made between corrected diffuse values resulting from application of an isotropic versus the real sky distribution. Results showed that an isotropic diffuse correction overestimated the real diffuse signal by a maximum of 4 to 6% for clear skies, with this magnitude decreasing with an increase in SZA. This is in accordance with the increase in sky isotropy at high SZAs. In contrast, isotropic corrections for increasing cloud cover displayed greater fluctuation in the overestimate statistic over relatively short time periods. Here, variability at low to mid SZAs (for both cloud types) ranged from approximately 2 to 5% for moderate cloud cover and zero to 10% for high cloud cover. Furthermore, results can again differ between instruments for the same cloud and SZA conditions due to cosine response differences. Finally, the overestimate variability under cloudy skies tended to increase with SZA such that the overall trend displayed less SZA dependence than that for clear skies. Here, cloud distribution masks the effect of increased sky isotropy.