Introduction

Solar ultraviolet radiation transmitted through the atmosphere to earth's surface is known to induce various biological and chemical processes, many of which are harmful to living tissues and some materials (see UNEP, 2006, for a review). Examples of processes relevant to human health include skin-reddening (erythema), synthesis of vitamin D within skin, and induction of various skin cancers. The long-term geographical distribution of surface UV radiation is of considerable interest towards understanding these effects. However, environmental UV levels are highly variable due to daily and seasonal cycles at different latitudes, and to variations in atmospheric transmission (mainly attributable to variations in ozone, clouds, and aerosols) and surface reflections. Ultraviolet radiation measurements by ground-based instruments are too few, and their record relatively short, to construct a unified picture of its average global distribution.

An alternative method of estimating surface UV levels with long-term global coverage relies on satellite-based observations of earth's atmosphere and surface, combined with a computer model of the propagation of UV radiation through the atmosphere. This methodology is already in use on a NASA website (http://jwocky.gsfc.nasa.gov/ery_uv/ery_uv1.html, which uses data from the TOMS ozone-monitoring satellite instruments to generate maps of erythemal UV for specific days. Other applications of the technique have illuminated interesting aspects of the problem, i.e., estimation of zonal mean irradiances at different UV wavelengths, of trends due to ozone changes, of cloud effects, and of geographical distributions based on monthly averaged ozone and clouds (e.g., Frederick and Lubin, 1988; Madronich, 1992; Eck et al., 1995; Frederick and Erlick, 1995; Herman et al., 1996a; Lubin et al., 1998; Herman et al., 1999; Sabziparvar et al., 1999; Herman et al., 2000; McKenzie et al., 2001). Here we use satellite-based observations of atmospheric ozone and clouds to derive a climatology of erythemal UV radiation with nearly global coverage (excluding the polar regions), averaged over the years 1979 - 2000. We developed a fast method for the explicit calculation of UV daily doses for each day of the whole time period. Averaging daily UV doses, rather than calculating monthly doses on the bases of monthly-averaged cloudiness and ozone, reduces possible uncertainties connected with the non-linear relationship between atmospheric parameters (e.g., total ozone and clouds) and surface UV radiation. Comparisons with long-term measurements at 22 UV monitoring stations allow some assessment of the reliability of this technique. Climatologies such as those presented here can be useful in epidemiological studies that assess the role of long-term environmental exposure to ultraviolet radiation, such as those discussed in Chapter 2 (McKenzie and Liley).

1.2 Method

UV broadband irradiances (W m ) are computed as integrals over wavelength A (nm) of spectral irradiances E(A) (W m-2 nm-1) weighted by appropriate spectral functions 5(2) (typically unit-less):

Irradiance =

E(A) is a function of solar zenith angle (SZA) and surface elevation, as well as optical depth profiles of atmospheric absorbers and scatterers (e.g., ozone and clouds). The values of S(A) are unity for UV-A and UV-B in the respective wavelength ranges of 315 nm - 400 nm and 280 nm - 315 nm, and zero outside these ranges. Figure 1.1 shows the wavelength dependence for three action spectra with relevance to human health: (1) erythema (McKinlay and Diffey, 1987), (2) pre-vitamin D3 production in human skin (Holick et al., 2006, after MacLaughlin et al., 1982), and (3) photocarcinogenesis of non-melanoma skin cancers (CIE, 2006). The erythema action spectrum has been accepted for the calculation of the instantaneous UV index (defined as the UVery irradiance multiplied by 40 (ICNIRP, 1995; WMO, 1997)), and the time-integrated standard erythemal dose (SED = 100 J m-2 (CIE, 1998)). In practice, use of this CIE spectrum emphasizes the ozone-sensitive region of 295 nm - 320 nm, peaking near 305 nm with minor contributions from longer wavelengths (Madronich et al., 1998; Micheletti et al., 2003). The other two functions are somewhat similar, in that they maximize at around 305 nm wavelength, and decrease by several orders of magnitude by 330 nm.

280 300 320 340 360 380 400 Wavelength (nm)

Figure 1.1 Spectral functions for erythema: solid line (McKinlay and Diffey, 1987); synthesis of pre-vitamin D3: dashed line (MacLaughlin et al., 1982; Holick et al., 2006); and non-melanoma carcinogenesis: dotted line (CIE, 2006)

Compilation of a global UV climatology is computationally intensive, requiring the calculation of E(X) at all relevant wavelengths, at each geographical location, and over diurnal cycles for each day of each year. To reduce computational time, we used the TUV model (Madronich and Flocke, 1997) to pre-tabulate values of weighted UV irradiances as a function of SZA (0° to 96° in 1° steps), ozone column (43 DU - 643 DU in steps of 10 DU), and surface elevation (0, 3, and 8 km above sea level), for cloud-free and aerosol-free conditions. The omission of UV absorption by aerosols can lead to overestimates of irradiance for polluted locations; this limitation will be discussed in more detail later. E(X) at earth's surface was computed at 1 nm steps from 280 nm - 400 nm. The spectral irradiance

280 300 320 340 360 380 400 Wavelength (nm)

Figure 1.1 Spectral functions for erythema: solid line (McKinlay and Diffey, 1987); synthesis of pre-vitamin D3: dashed line (MacLaughlin et al., 1982; Holick et al., 2006); and non-melanoma carcinogenesis: dotted line (CIE, 2006)

incident at the top of the atmosphere was taken from the Atlas3/SUSIM measurements (D. Prinz, pers. comm., 1998). Vertical profiles (appropriate for mid-latitude, annual average conditions) for air density, temperature, and ozone were taken from the U.S. Standard Atmosphere (USSA, 1976) with, however, the ozone profile re-scaled to the actual ozone column (see below). The propagation of solar radiation through the atmosphere was computed with a 4-stream discrete ordinates method (Stamnes et al., 1988), with pseudo-spherical correction for improved accuracy at low sun conditions (Petropavlovskikh, 1995). A Lambertian surface albedo of 5% was assumed at all wavelengths.

The atmospheric ozone column and cloud reflectivity at 380 nm (R) were taken from the TOMS data from three satellites: (1) Nimbus-7, Level 3/Version 8 (McPeters et al., 1996), Nov. 1, 1978 to Dec. 31, 1992; (2) Meteor-3, Level 3/Version 8 (Herman et al., 1996b), Aug. 22, 1991 to Dec. 11, 1994; and (3) Earth Probe, Level 3/Version 8 (McPeters et al., 1998), July 7, 1996 to June 30, 2000. The geographical resolution of the measurements was 1.25° longitude by 1.00° latitude. For each grid point, only one satellite overpass per day occurred (ca. local noon). We therefore assumed constant ozone and reflectivity values for the entire day. Local values of the ozone column, SZA and surface elevation were used to compute the clear-sky irradiances at 30-minute intervals over half days by interpolation of the pre-tabulated values. Assuming symmetry about local noon, these data were integrated over 24 hours to obtain the daily UV-A, UV-B, and erythemal doses. A correction for variations in the earth-sun distance was applied as a function of date. A reduction factor F for cloud cover, identical to that used by Eck et al. (1995), was then applied:

For cloud-free and aerosol-free conditions, total reflectivity at 380 nm is dominated by Rayleigh scattering and surface reflections, the latter being rather small at UV wavelengths unless snow or ice is present. The TOMS algorithm attributes excess reflectivity to clouds or scattering aerosols, without distinguishing between the two. When high surface albedo is encountered (e.g., snow or ice), this method erroneously interprets the high surface reflectivity as cloud cover, thus artificially reducing surface UV irradiance. Polar regions are therefore excluded from our analysis. For non-polar regions, including mountainous regions, we did not attempt to correct for snow cover. The calculated UV doses for such areas should therefore be considered as lower limits.

The calculation of UV doses should in principle be carried out for each location and each day over the satellite record (ca. 1979 - 2000). However, gaps in the satellite record exist, so that for some days and/or locations, no doses could be computed. These missing days require some consideration to avoid biases in any

long term averages and trends. For each location, monthly averaged doses were calculated for each of the 247 months in the combined dataset, but were considered valid only if at least half of the days in that month had data. No attempt was made to discriminate between the months in which data gaps typically occurred during the early part of the month and when they typically occurred during the latter part of the month. In some cases, measurements for the same location and days were available from two different satellites; in such cases, monthly means for each satellite were computed, then averaged together to obtain a single mean for that month.

Climatological monthly values were computed for each location by averaging all valid values for that month over multiple years (e.g., climatological January is the mean of all valid January values over 1979 - 2000, etc.). For most of this chapter, we consider averages over the full 22 years (1979 - 2000), but for some of the discussion below, we also considered the time periods 1979 - 1989 and 1990 -2000 separately. Climatological annual values were computed as the mean of all valid climatological monthly values, specifically (mean of all Jans. + mean of all Febs. h-----h mean of all Decs.)/12.

The second period (1990 - 2000) is missing some data (all of 1995, Jan - Jun 1996, Jul - Dec 2000). We tested the effects of these missing data on the calculated changes by temporarily removing the analogous months from the 1979 - 1989 record and comparing the resulting climatology to that of the complete 1979 - 1989 period. Differences of < + 0.2% were obtained. This is on the order of ~1/10 of the clear sky changes between the two periods 1979 - 1989 and 1990 - 2000, and on the order of < 1/10 of the changes in the "all sky" values between these two periods.

For a comparison with the satellite-derived estimates, we used measurements of UV irradiances by ground-based spectroradiometers, obtained from the World Ozone and UV Data Center archive (WOUDC; data downloaded June 2002). Measured UVery doses are reported as daily integrals of spectral observations integrated over wavelength with the McKinlay and Diffey (1987) erythemal action spectrum weighting. The archives include 22 non-polar stations; 10 in Canada (Meteorological Service of Canada, MSC); 4 in Japan (Japan Meteorological Agency, JMA); 2 in the Taiwan region ("Central Weather Bureau of Taiwan, CWBT"), and 1 each in Obninsk, Russia (Institute of Experimental Meteorology-Scientific Production Association (IEM-SPA)), Poprad-Ganovce, Slovakia (Slovak HydroMeteorological Institute (SHMI)); Mauna Loa, HI (MSC); San Diego, CA; Ushuaia, Argentina; and Palmer Station, Antarctica (all US National Science Foundation (NSF) sites). The NSF sites operated double monochromators (Biospherical Instruments, Inc), while all other sites operated Brewer single monochromators. Our satellite-based irradiance values for station locations were derived for the locations and altitudes of the ground-based stations.

Was this article helpful?

0 0
How To Prevent Skin Cancer

How To Prevent Skin Cancer

Complete Guide to Preventing Skin Cancer. We all know enough to fear the name, just as we do the words tumor and malignant. But apart from that, most of us know very little at all about cancer, especially skin cancer in itself. If I were to ask you to tell me about skin cancer right now, what would you say? Apart from the fact that its a cancer on the skin, that is.

Get My Free Ebook


Post a comment