Clouds roll in. The glare eases. People take the hat off.

That is the wrong cue. Thick overcast that hides the sun does cut UV. Broken, puffy cloud that still leaves a sharp shadow can do the opposite: cloud walls bounce extra ultraviolet down, and the index at the ground can sit above the same hour on a cloudless day.

How the boost works

The sun’s disc is still in the open. Direct UV keeps arriving. Bright cloud sides then scatter more down on top of it. WHO notes that light or thin cloud has little effect and may even enhance UV because of that scattering. INTERSUN’s public line is the same physics in a different number: over 90% of UV can get through light cloud. NOAA’s Climate Prediction Center is blunter about the geometry: UV reflects off the sides of towering cumulus, and those conditions produce actual enhancements at the surface.

NIWA’s public UV work puts a size on the spike:

• Puffy cloud, sun still visible — about 20% or more above the clear-sky index; nearby cloud can push it to about 30%
• Clear-sky peak of 10 — a 20% bump is 12; the real-world peak can occasionally exceed 12
• Sun fully hidden (no shadow you can pick out) — typically about half the clear-sky index, with a wide range

NIWA’s smartphone-UVI notes match the top of that range: clouds near the sun can enhance the index by up to ~30%, but only for short periods. The minutes you stand in the gap are not the day. A separate NIWA briefing puts the average the other way: heavy overcast, more than 90% cut; on average, clouds take about 30% off clear-sky UV.

Field measurements of those bursts most often last a few minutes. In Río Gallegos, Patagonia, 62% of UV-enhancement events lasted 1–5 minutes; 2.7% ran 25–30 minutes; fewer than 1% lasted longer than 30 minutes. Enhancements there reached 25–30%. A long stretch outdoors still adds up.

Average cloud is not the gap

The EPA’s average cloud factors used in U.S. forecasts go the other way—scattered cloud about 89% of clear-sky UV, broken about 73%, overcast about 31%. Those are average cuts across a sky. They do not describe the minutes when the disc is open and the cloud walls are bright. A forecast that assumes broken cloud at 73% is smoothing the hour. Your face is in the 20–30% bump.

McKenzie and colleagues, looking at cloud modification factors at Lauder and two other NDACC sites, found enhancements (CMF > 1) do happen—and CMF stayed below 1.3 in 99% of cases. Treat 20–30% as the planning band for a bright gap, not as a reason to ignore the live number.

Why you drop cover

Three traps:

1. Cooler air, softer glare. Clouds block heat and visible light better than they block UV. Comfort is not the index.
2. “It’s not full sun.” If you still have a sharp shadow, the disc is not covered. That is the enhancement setup, not shelter. The shadow test for sun height is a different check. This one is simpler: can you pick the shadow out at all.
3. A peak you already checked. A clear-sky peak of 8 can behave like about 10 in a bright gap. Keep cover on for the break, not for the grey in between.

Open shade still leaves scattered UV. That is the shade problem. Cloud walls add a different extra: they can raise the open-sky number itself.

When you still have a shadow

Use the live UV index first, then treat the sky as a switch, not a shield:

1. Check today’s peak. If it is High (6–7) or above, keep cover on through the bright gaps. A 20–30% bump can move a High hour into Very High.
2. Use the shadow. Sharp shadow: treat it as at least full sun, maybe more. No shadow you can pick out: UV is usually down, often toward half—still not zero on a High-peak day. NIWA’s rule of thumb is that 50% cut if you cannot discern a shadow.
3. Do not wait for the next break to reapply. The burst is the gap. Hat, shirt, and sunscreen stay on between clouds. Most bursts die in a few minutes. Some run toward half an hour.
4. Dark storm cloud is the exception. Heavy cover that kills the shadow is a real cut—NIWA’s >90% overcast drop. Puffy fair-weather cloud is not.
5. Recheck when the sky opens. The live reading moves faster than the look of the afternoon.

Sources

Short-term cloud enhancement (often ~20% or more, up to ~30%; a clear-sky 10 can briefly behave like 12) and the ~50% cut when no shadow can be discerned: NIWA, Calculation of UVI for Smartphone Apps. Heavy-overcast >90% cut and ~30% average reduction: Liley, NIWA. Light cloud may even enhance UV: WHO UV radiation Q&A. Over 90% of UV through light cloud: WHO INTERSUN. Towering-cumulus side reflection: NOAA CPC, Effects of Clouds. EPA forecast averages (89% / 73% / 31%): Learn About the UV Index. Burst lengths (62% at 1–5 min, some 25–30 min) and 25–30% enhancements: Wolfram et al., Río Gallegos. CMF < 1.3 in 99% of cases: McKenzie et al., NDACC poster. EPA percentages are forecast averages, not the size of one bright gap.