Peak UVI revisited

A number I believe ....

Colleagues, led by Raul Cordero at Chile’s University of Santiago, recently published a paper reporting radiation measurements from at Altiplano region, a region we identified nearly 2 decades - ago at a NIWA UV Workshop in 2006 - as having the highest UV amounts anywhere on the planet.

Our result was based on an analysis of satellite data. Here’s an updated version of our plot (thanks to Ben Liley), using a new expanded UVI colour scale, and centered on New Zealand. The grid cell we found with the highest values was close to Cusco, Peru, altitude 3400 metres on the eastern flank of the Andes (latitude = 13.5S, longitude = 72W metres). I remember that paper making us very popular with our colleagues in Peru! We put them firmly ‘on the map’.

Peak UVI deduced from satellite for each 100 x 100 grid cell of the planet. The maximum UVI of 25 occurs in the Altiplano region

Cordero’s new paper, titled “Surface Solar Extremes in the Most Irradiated Region on Earth, Altiplano”, reports on results from nearby high-altitude measurement site called Chajnantor (altitude 5,150 m). That’s just across the border from Peru in Chile, a couple of hundred kilometres from Cusco.

I’ll summarise the relevant points here.

The total short wave radiation (strictly, irradiance, according to fussy physicists) from the Sun arriving at the top of the Earth’s atmosphere is about 1,362 Watt/m2. But at this clean high-altitude site, the observed values are frequently much higher than that. The enhancement over the clear-sky value is due to reflections from the edges of clouds. Cordero reports an instance in this paper of the irradiance reaching 2,177 W/m2 (an enhancement of about 70 percent over that arriving outside Earth’s atmosphere - i.e., a ‘cloud enhancement factor’ of 1.7). Apparently they didn’t have any UV measurements available on that particular day (Measurements of total radiation have been available at the site since 2016).

They date they identify as having the highest UVI is 22 January 2020: a day that was relatively clear in the morning, but much cloudier in the afternoon. The plot below, from their paper, shows the variation in both total solar radiation (blue curve) and the UVI (red crosses) over that day.

The total solar radiation data (blue line) are 1-minute averages, while UVI data are integrated values from a scanning instrument that takes several minutes per scan. The dashed blue line is a calculation of the total solar radiation.

The peak total solar irradiance on this day, which occurred during the cloudy afternoon, was about 1,500 Watts per square metre, about 20 percent more that the total outside the atmosphere. The peak UVI is similarly elevated. Its peak value is about 25.8, a little more than the all-time peak clear-sky value for nearby Cusco in our map above.

I was a little puzzled at the continued close match between short wave irradiance and UVI throughout the morning period. Because of its stronger absorption by ozone (and its larger losses due to Rayleigh scattering by air molecules), I’d expect a more rapid fall-off at lower sun elevation angles for UVI compared with total radiation. To demonstrate that, I used the very useful on-line calculator produced by my colleague Sasha Madronich, to calculate the expected clear sky variation at that location and date. The results are shown below.

UVI and shortwave radiation (400 to 920 nm) relative to their noon values calculated for clear skies.

As you can see from the plot, the UVI is much more strongly peaked to the midday value, and drops off more quickly at other times compared with the drop-off in total radiation. For example, 3 hours before or after local noon, that separation has become quite large. The UVI has halved to about 50 percent of its peak value whereas the total radiation has dropped only half as much to about 75 percent of its peak value, in good agreement with the dashed blue line in Cordero’s figure.

By contrast, no such difference is visible in the plot from Cordero’s paper. At 10:30 am, both the total radiation and the UVI have dropped to about 50 percent of the peak clear-sky value. The unexpected match in behaviour between the two measurements could be due to differences in the angular response of the instruments. Or it may be attributable to cloud effects. You can see from the noise in the signal that it wasn’t perfectly clear.

Frankly, I’m at a loss to explain it. Hopefully at least one of the authors will read this and post a response. It would be a lot easier to see what’s going on if they could provide additional UV measurements from earlier in the day.

Notes added over the period 23-26 July 2023.
A reader, Zim Sherman, noted that the figure has different lower ranges for the two y-axes, which goes some way towards explaining my confusion. The paper’s lead author, Raul Cordero, has now conformed that the left axis is non-linear (Thank you Raul). My guess is that it’s a power-law function. Rather unusual. I didn’t look closely enough at the lower axis numbers because the model curve ‘appeared’ to reach zero at about the correct time. Now that I look at that curve more closely, the shape seems a bit wrong for total short wave radiation and it’s more more like what the UVI variability would look like (see my plot above comparing those model calculations). I wonder if the dashed blue model curve from Cordero’s paper is actually modelled UVI - i.e., applies to the right axis - rather than modelled total short wave radiation as labelled. If so, it should also therefore have been drawn in red, rather than blue. On the other hand, the ‘wrong’ shape for that curve could also be a consequence of the highly non-linear y-axis for small irradiances. All very confusing.

The good news is that - unlike other results claiming extreme UV levels - I do (still) believe this new record value. For clear skies, I calculate a peak UVI = 20.2, in close agreement with that measured before the cloud enhancement period.

Incidentally, I passed my earlier post about that ridiculously high UVC radiation on to Herndon (the source of the bogus data), but haven’t yet received any response. His mail box must be overflowing 😊.