Jeff Sabburg

Feature Extraction from Whole-Sky Ground-Based Images for Cloud-Type Recognition

Several features that can be extracted from digital images of the sky and that can be useful for cloud-type classification of such images are presented. Some features are statistical measurements of image texture, some are based on the Fourier transform of the image and, finally, others are computed from the image where cloudy pixels are distinguished from clear-sky pixels. The use of the most suitable features in an automatic classification algorithm is also shown and discussed. Both the features and the classifier are developed over images taken by two different camera devices, namely, a total sky imager (TSI) and a whole sky imager (WSC), which are placed in two different areas of the world (Toowoomba, Australia; and Girona, Spain, respectively). The performance of the classifier is assessed by comparing its image classification with an a priori classification carried out by visual inspection of more than 200 images from each camera. The index of agreement is 76% when five different sky conditions are considered: clear, low cumuliform clouds, stratiform clouds (overcast), cirriform clouds, and mottled clouds (altocumulus, cirrocumulus). Discussion on the future directions of this research is also presented, regarding both the use of other features and the use of other classification techniques.

Optical properties of poly(2,6-dimethyl-1,4-phenylene oxide) film and its potential for a long-term solar ultraviolet dosimeter

The optical properties of poly(2,6-dimethyl-1,4-phenylene oxide) (PPO) film have been characterized in order to develop an alternative method for UV dosimetry with a focus on long-term human exposure measurements. The dynamic range of PPO film was found to extend to 2 MJ m(-2) of broadband UV exposure independently of film thickness, providing an exposure range of roughly four summer days at subtropical latitudes. The sensitivity of the film to UV exposure was positively related to film thickness in the 20-40 microm range. Films of 40 microm thickness proved to be the most suitable for long-term human UV exposure measurements. The temperature independence of the response of 40 microm PPO film was established from 1.5 degrees C to 50 degrees C within a dosimeter response uncertainty of 6.5%. Dose-rate independence was also demonstrated within 8% of the mean dosimeter response. The spectral response approximates the CIE erythemal action spectrum between 300 and 340 nm, with a peak response at 305 nm. A large deviation from this action spectrum was observed at shorter wavelengths. Investigation of the angular response in both the azimuth and altitude planes showed a cosine error of less than 6.2% between 0 degrees and 40 degrees, and did not exceed 13.3% at any angle greater than 40 degrees. These results indicate that PPO film satisfies the requirements for use as a UV dosimeter, and may be employed in long-term human exposure measurements.

Enhanced spectral UV irradiance: a 1 year preliminary study

The UV spectra on cloudy days were compared to those on cloud free days to determine which part of the UV spectrum has the greatest enhancement due to the cloud compared to both corresponding measured clear-sky spectra as well as other enhanced spectra. In this preliminary study, cloud enhanced UV spectra selected for maximum UVA enhancement compared to a clear-sky UV spectrum at similar solar zenith angle (SZA) and ozone values, showed that the ratio of the two sets of spectral irradiances was approximately wavelength independent (approximately 1.1) above the cut-off wavelength of approximately 306 nm. Similarly, above 306 nm the average ratio of the spectral irradiances of a maximum UVB enhanced UV spectrum compared to a clear-sky spectrum was 1.2 with maximum values generally above this average between 316 and 344 nm and generally below 1.2 above the wavelength of 344 nm. The UVA and UVB enhanced spectra were separated into five SZA ranges and the irradiance at each wavelength averaged for each range and compared to clear-sky spectra in each of the ranges. Above approximately 306 nm, the ratios are wavelength independent for all SZA. However, with the exception of the SZA range centred on 20 degrees, there is an increasing dependency with shorter wavelengths below the 306 nm. Also there appears to be two distinct groupings of the average irradiance ratios, corresponding to the SZA range centred on 20, 37 and 49 degrees (ratio of 1.2) and 32 and 42 degrees (ratio 1.0), the latter cases suggesting that on average there is no enhancement for these SZA, except for wavelengths less than 306 nm.

Scattered UV Beneath Public Shade Structures During Winter

Broadband field measurements were conducted beneath three different‐sized public shade structures, small, medium and large, during winter in the Southern Hemisphere. These measurements were compared with the diffuse UV to quantify the relationship of the UV under and around the shade structures to the diffuse UV. For the shade structures, a relationship between the diffuse UV and the UV in the shade has been provided for clear skies and solar zenith angles (SZA) of 49–76°. This allows the prediction of the UV in the shade of these structures if the diffuse UV is known. The ultraviolet protection factors for the three shade structures ranged from 1.5 to 5.4 for decreasing SZA. For the greater SZA of 70–76°, the erythemal UV in the shade was 65%, 59% and 51% of that in full sun for the small, medium and large structures, respectively. For the smaller SZA of 50–53° the erythemal UV in the shade was 35%, 41% and 18% for the small, medium and large shade structures, respectively. From this research it can be concluded that the UV radiation levels in the shade in winter could cause erythema and other sun‐related disorders.

Evaluation of a ground-based sky camera system for use in surface irradiance measurement

This paper describes the evaluation of a ground-based sky camera system for studying the effect of clouds on the level of the ambient ultraviolet radiation. The system has been developed for research in the characterization of the effect of clouds around the sun. It is the first sky camera system to be used for the assessment of cloud conditions in the vicinity of the sun, rather than a whole-sky assessment. The system features a sun-tracking sky camera with an integrated measurement of horizontally received radiation at the same location. The imageprocessing algorithm uses solar radiation readings to reduce reflections from the sun on the camera system being mistaken for cloud in the images. Cloud amount was estimated in an angular region of between 12.58 and 37.58 around the sun. The algorithm also estimates the amount of solar obstruction by cloud (sun not covered, partially or totally covered). The system was evaluated during September 1997 at Toowoomba, Australia (27.68S latitude). Compared to manual assessment of 592 images, 76.5% were identified correctly by the algorithm for the degree of solar obstruction and 81.9% for cloud amount. The behavior of ultraviolet radiation levels with cloud conditions is discussed.

Understanding the UVA environment at a sub-tropical site and its consequent impact on human UVA exposure

Daily UVA and erythemal irradiance data on a horizontal plane at a sub-tropical site were measured during a period from March 2000 to February 2001. On a relative basis, UVA radiation was shown to be a greater concern to human exposure during the winter months than summer months. In summer (December to February), the peak daily UVA exposure was 205 J cm(-2) and in winter (June to August), the minimum daily value was 19 J cm(-2). The peak daily UVery exposure was 37 MED in summer and the winter minimum was 4 MED. The occupational work day UVA exposure to the vertex of the head was estimated using the collected UV data. The outdoor workers received 89% of the available UVA radiation whilst the home workers received 18% of the available ambient UVA radiation. This result parallels the exposure patterns of these two population groups, with the outdoor workers spending most of the working week outdoors, whilst the home workers spend small, intermittent time periods outdoors in the sun.

Glass Filtered Solar UV

Transparent screens such as those in vehicle window glass and windscreens, in office and home window glass and in other structures act as a barrier to some of the shorter solar UV wavelengths. As a result, the spectrum of filtered UV may be substantially different from that of the unfiltered solar UV spectrum and the relative ratio of UVA to UVB irradiances can change. Additionally, the spectral transmission of different types of glasses varies with the type of glass and the angle of transmission of the UV through the glass (Bartels and Loxsom, 1995). These changes in the UV spectrum need to be investigated in terms of the spectral dependence of the action spectrum of the biologically effective process being considered.

Personal Solar UV Exposures in Diffuse UV Settings

The previous chapter has described the details of the spectrum of diffuse ambient solar UV radiation in a number of different environments. This chapter outlines the broadband erythemal UV, UVA and the visible irradiances in the shade of trees. The percentage of these irradiances in the shade compared to those in the sun may change with season and the time of day due to the relative proportions of the direct to diffuse radiation changing with SZA. Consequently, the data from the measurements presented in this chapter will be for different times of the day in each of the seasons. Additionally, the percentage of diffuse broadband erythemal UV and UVA radiation in both full sun and in the shade of trees is compared. The resulting amount and distribution of the personal solar erythemal UV exposures to humans in the tree shade will be discussed.

Diffuse Ambient Solar UV

The direct, scattered and filtered solar UV radiation leads to the global UV radiation environment that humans are exposed to being comprised of both the direct and the diffuse components. The diffuse component provides a significant proportion of the human UV exposures as it is incident from all directions and difficult to reduce with the usage of hats, tree shade and shade structures. The anatomical distribution of UV exposures to humans is influenced by the relative proportion of diffuse UV. Additionally, the UV exposure to a surface is influenced by the angle of inclination and orientation of the receiving surface. In shade or under cloud, it may possibly be accompanied by an increase in exposure time to humans because of lower temperatures and reduced thermal discomfort and a possible incorrect perception by the public that there is no UV in shade or under cloud.

UV Enhancement by Cloud

In Chapter 4 the effect of clouds on solar U V radiation has been studied. Figure 4–1 illustrated the effect of clouds on UV radiation during 1 year of continuous daily erythemal UV measurements at Toowoomba, Australia (27.6°S, 151.9°E, 693 m above sea level). The effect of clouds could be seen by the general decrease of the UV levels from the clear-sky levels (i.e. the maximum envelope of the data points). It has also been mentioned that since the 1960s the measurement of UV radiation at the surface of the earth during cloudy conditions, has sometimes revealed levels exceeding equivalent clear-sky values under cloudy conditions. This is commonly referred to as cloud enhanced UV, the subject of this chapter, and has been reported over different timescales and for different wavelengths. Figure 5–1 illustrates possible cases of UV enhancement on a daily basis, i.e. those data points shown above the equivalent modeled clear-sky values. Sometimes these data points are not confirmed to be actual cases of enhancement, e.g. Bais et al. (1993) found a slight increase of the UV irradiance during partly cloudy conditions compared to clear-sky conditions, however, they could not confirm if this was actual enhancement as the increase was within the instrument’s uncertainty limits.