The Namib is one of the driest places on Earth, and it runs on fog. Cold water welling up in the Benguela Current chills the air over the South Atlantic, and on most mornings a grey wall of it slides in over Walvis Bay and Swakopmund and keeps going, across the gravel plains and the dunes, sometimes a hundred kilometers inland.
Everything out there has learned to drink it. Lichen fields. Welwitschia. A beetle that climbs a dune crest before dawn, faces the wind, and stands on its head so the fog condenses on its back and runs down to its mouth.
For a long time the question about fog was what it carries: water, salt, nutrients. This year the question changed. It turns out fog may also be a place where things live.
Something is alive in there
In May 2026, a team working with Ferran Garcia-Pichel at Arizona State published a study in mBio that sounds like the opening of a horror film. They sampled fog in Pennsylvania and found bacteria, mostly Methylobacterium, that weren’t just riding along in the droplets. Under the microscope they were getting bigger and dividing. And they were eating: breaking down formaldehyde, a common air pollutant that dissolves into fog, and turning it into carbon dioxide.
Fewer than one droplet in a hundred holds a bacterium, but fog water ends up about as crowded with microbes as seawater. Garcia-Pichel’s framing is the part that stuck with me: “If we harvest fog, we are getting rid of our little friends in the air. We don’t know if that’s going to make a big impact or not, but we should be considering that.”
The Namib was already in this story. Back in 2019, Sarah Evans and colleagues compared fog microbes on the coast of Maine with fog at two Namib sites 50 and 55 kilometers inland, right around the Gobabeb research station. Different continents, different oceans, and the same result: microbial diversity went up during and after fog, and fog moved living cells from the sea onto the land.
If fog is a habitat, it needs a habitat map. So I built one.
Mapping the fog
The first question any field team asks is simple: where and when is the fog reliable? You can’t sample something that shows up twice a year.
FOG-WATCH answers that from free data. It pulls ten years of hourly weather from ERA5, the European reanalysis, through Open-Meteo’s archive, no API key required. An hour counts as foggy when low cloud covers at least 70% of the sky and the air two meters off the ground is at least 95% saturated. A GitHub Action reruns it every month.
I laid a line of points due east from the coast at Gobabeb’s latitude and let it run. Here’s what came back, as the share of all hours, day and night, year-round:
| Place | Distance inland | Foggy hours |
|---|---|---|
| Swakopmund | coast | 8.9% |
| Walvis Bay | coast | 6.8% |
| Transect | 11 km | 5.5% |
| Transect | 51 km | 2.4% |
| Gobabeb | 56 km | 1.8% |
| Transect | 92 km | 0.5% |
| Transect | 153 km | 0.1% |
1.8% at Gobabeb sounds like nothing until you do the arithmetic. It’s a share of every hour in the year. Forty-odd fog mornings of three or four hours each gets you right there. In a place where rain comes to a couple of centimeters a year, when it comes at all, that’s the water supply.
The hour-by-month view is where it gets good. At Gobabeb the fog is a dawn thing. It builds after 2 a.m., peaks between 6 and 7 in the morning, when about one morning in seven is foggy from October through February, and it’s gone by 10. May is almost fog-free. Out on the gridded map, the foggiest square of all isn’t on land. It sits offshore, south of Sandwich Harbour, where the fog bank forms over the cold current before the wind pushes it inland.
A note on method, because the honest version matters. ERA5 is a reanalysis: a weather model constrained by decades of observations, laid on a grid about 25 km across. It isn’t a fog sensor. My rule for “foggy” is a proxy, and it catches the low stratus deck that becomes fog when it touches the ground as well as fog itself.
That trade is the whole point of the project. A proxy that’s consistent everywhere lets you compare the Namib to the Atacama to San Francisco with the same ruler. Station records are better in one place. A grid is better at telling you where to put the stations.
A paper that got there first
Here’s the moment that made me trust the map. Earlier this year, a team led by Deepanshu Malik and Jan Cermak published a study in Atmospheric Chemistry and Physics on exactly this fog, built around direct measurements of how high the cloud base sits.
Their findings, next to mine:
- Fog in the central Namib stays within about 100 km of the coast, and only rarely goes further. My transect drops to half a percent at 92 km.
- There’s a high-fog season from August to March, when the cloud deck sits higher and pushes about 60 km inland, and a low season from April to July, when it hugs a strip under 15 km wide. My Gobabeb heatmap lights up October through February and nearly goes dark in May.
- Fog is a night and early-morning phenomenon. The cloud base sinks through the night and lifts fast after sunrise, especially inland. Same pattern, same hours.
Two completely different methods, measurements on the ground and a reanalysis in the cloud, drew the same fog.
The beetle that got the credit
You can’t write about Namib fog without the beetle, and the beetle comes with its own correction.
The headstand is real. In 1976, William Hamilton and Mary Seely described it in Nature: Onymacris unguicularis climbing to dune crests on foggy mornings and tilting head-down into the wind to drink.
Then, in 2001, a different beetle became famous. Stenocara, with a bumpy back said to collect water on hydrophilic peaks between water-repellent valleys, launched twenty years of “beetle-inspired” fog-harvesting materials. You’ll still see it in design articles and podcasts.
In 2010, Thomas Nørgaard and Marie Dacke put four Namib beetles in a fog chamber. Only Onymacris did the headstand. Stenocara didn’t fog-bask at all, and every wing case they tested was fully water-repellent, bumps included. Their conclusion was that the behavior matters more than the surface.
Bumpy-surface materials still work in the lab, and the biomimicry field has moved on from needing Stenocara to be literally correct. But if you read that a Namib beetle drinks the fog with its bumps, the beetle doing the drinking is Onymacris, and it does it by standing on its head.
Three fog coasts
The Namib has cousins. Anywhere a cold current runs along a desert coast, you get the same setup: the Atacama’s camanchaca, where villages string up mesh nets and pull thousands of liters a day out of the air, and California’s summer marine layer, which is why San Francisco is cold in July.
FOG-WATCH runs all three with the same ruler. The gridded map shows the California fog peaking in July, at 31% of all hours across the Golden Gate region. That’s the marine layer, right on schedule.
My single “Golden Gate” point peaks in December, not July. Its grid square probably sits where winter radiation fog satisfies the 95%-humidity rule more often than the summer marine layer does. It’s a reminder that a 25 km square and a fixed threshold can both be wrong in small, specific ways, and that the grid view is the one to trust for patterns.
Where this goes
The next layer is satellites. GOES and MODIS can see fog directly, at much finer resolution than a reanalysis. After that comes the actual goal: ranking places that are foggy, reliable, and reachable by road as candidate sampling sites for the people now trying to figure out what lives in the fog.
The Namib has been running this experiment for millions of years. Beetles, lichens and welwitschia worked out where the fog lives long before anyone had a map. We’re only now finding out that the fog itself is alive.
Explore it yourself. The fog-blanket map with a month slider, the Namib transect and the hour-by-month heatmaps are all live, and the code is open.
Open FOG-WATCH → Code on GitHub →Go deeper
A summary is not a substitute for the papers. Start with the listens if you have a commute, and the mBio paper if you have twenty minutes.
What lives in the fog? Pollution-eating bacteria — Science Friday, 10 July 2026. Lead author Thuong Cao on finding Methylobacterium growing inside fog droplets and eating formaldehyde.
Water from Thin Air — EarthDate, episode 43, 2018. A short episode on the Atacama’s fog-catcher nets, including the village that brews beer from fog. It tells the Stenocara version of the beetle story, so now you know the footnote.
The Microbiome of the Clouds — Science Friday. An earlier segment on the microbes that live in clouds.
Cao, T.T.T. et al. (2026). Growth and formaldehyde degradation of photoheterotrophic Methylobacterium within radiation fogs. mBio. — the “fog is alive” paper, from Garcia-Pichel’s group at ASU.
Malik, D., Andersen, H., Cermak, J., Vogt, R. & Adler, B. (2026). Cloud base height determines fog occurrence patterns in the Namib Desert. Atmospheric Chemistry and Physics 26, 681. Open access. — the ground-truth study my map agrees with.
Evans, S.E., Dueker, M.E., Logan, R.J. & Weathers, K.C. (2019). The biology of fog: results from coastal Maine and Namib Desert reveal common drivers of fog microbial composition. Science of the Total Environment 647, 1547–1556. — fog microbes at Gobabeb and on the Maine coast.
Mitchell, D. et al. (2020). Fog and fauna of the Namib Desert: past and future. Ecosphere 11(1), e02996. Open access. — the big review of which animals depend on fog, and what happens if the fog changes.
Hamilton, W.J. & Seely, M.K. (1976). Fog basking by the Namib Desert beetle, Onymacris unguicularis. Nature 262, 284–285. — the original headstand paper.
Nørgaard, T. & Dacke, M. (2010). Fog-basking behaviour and water collection efficiency in Namib Desert Darkling beetles. Frontiers in Zoology 7, 23. Open access. — the fog-chamber study that sorted out which beetle actually does it.
Open-Meteo Historical Weather API (ERA5, CC BY 4.0) · FOG-WATCH source code
If you only read one, make it Malik et al. It’s open access, it has the best fog maps of the Namib anyone has made, and it’s the reason I believe mine.