UV in the twilight zone
No great risk ...
A couple of weeks back, I put out a call for UV requests on Substack’s Chat (you need to download their free app to see it). A reader, Steve, who has already suffered extensive skin damage from UV radiation responded.
I know that UVB drops off quickly at twilight, but I am interested to know whether there’s still risk of skin damage from any UVA that remains after sunset when some daylight is still present.
An interesting question, but I think the plots below should alleviate his concerns. The first ones show the progression - over sunny days near the summer solstice - of sunburning UV (UVI, left panel) and skin-wrinkling UVA (right panel). The upper pair is from Lauder, New Zealand (45°S) and the lower pair is from Alice Springs, Australia (24°S). Measurements are shown in red, while corresponding calculations for pristine skies are in black. There’s pretty good agreement on these clear days.
As you can see, the curves for UVA (at right) are wider, showing it does persist longer into the twilight period, but there’s still little of before sunrise and after sunset: 05:20 and 20:21 at Lauder, 05:47 and 19:15 at Alice Springs.
Below, I’ve plotted the variability of UVI and UVA for Lauder as a function of solar zenith angle (SZA), which is just the complement of the elevation angle (i.e, SZA = 90 - Solar Elevation Angle). The plots include all data obtained over a period of two decades. The stripes at 5-degree steps and the relative lack of data at larger SZAs arise because spectra are routinely measured at 5-degree steps in SZA up to SZA = 95° (which is 5 degrees below the horizon), and at 15 minute intervals over the 4 hours centred on local noon. It’s only on rare occasions that different sampling intervals were used. The largest noon sun elevation at this latitude is about 68 degrees, corresponding to SZA = 22°. Sunrise and sunset measurements are at SZA = 90°.
Incidentally, the regions higher densities of data points correspond to clear-sky data. In the left panel, you can see a double banding of those, with clear-sky days in spring having lower UVIs than in autumn because of the seasonal changes in ozone (which has a spring maximum and autumn minimum). The banding isn’t present in the UVA data because it’s independent of ozone. Note that the highest values for each SZA don’t occur under clear skies. They are characteristic of partly cloudy skies with the sun unobscured by clouds.
Again, you can see that the UVI (left plot) drops off more steeply than the UVA (right plot). The UVI - which comprises mainly UVB wavelengths but also includes a small contribution from UVA wavelengths - is essentially zero for all SZAs larger than about 85°, whereas for pure UVA there’s still about 1 percent of the summer maximum at sunset (SZA = 90°). A scattered light contribution persists even after sunset, though it has essentially all disappeared by the time the sun is 5 degrees below the horizon.
While this is all very interesting (thanks for raising it, Steve), I don’t think we need to worry too much about those UVA amounts at twilight. They are VERY small compared with the summer maximum. I think we can safely assume they pose no risk, especially given that the action spectrum for skin damage is much greater at UVB wavelengths.
In fact, the sharper drop-off in UVI reassures me that there’s little risk of UV damage whenever the sun is less than 10 degrees above the horizon (i.e., for all SZA > 80°). Over that range of SZAs, the UVI is never more than about 0.3. For a constant UVI of 3, it would take about an hour to to cause skin damage for the most sensitive skin types, so given the mean UVI amounts over the twilight period (say UVI = 0.1) , it would take an impossible 30 hours of exposure or more to cause any discernible reddening in the most sensitive skin types. That’s compared with less than 15 minutes at noon in summer.