Astronomy at the Bottom
of the World

Part II: Observing & Living at the South Pole

Cail Daley

Astronomy on Tap

April 16, 2026

Why can’t we do astronomy here?

• Lights
• Clouds
• Humidity

Microwaves!

Illustration: Harney & Sons Master Tea Blends

Microwaves are light — the same electromagnetic radiation as visible light, just with wavelengths a million times longer (centimetres instead of nanometres). Our eyes can’t see them, but they’re everywhere: wifi, bluetooth, cell phones, radar, microwave ovens — and the light Lennart just showed you, the cosmic microwave background, photons left over from 380,000 years after the Big Bang. Same kind of light, broadly the same part of the spectrum.

Why a kitchen microwave heats your food. A magnetron inside the oven emits at 2.45 GHz. Water is a polar molecule — lopsided charge distribution — so the oscillating field yanks it back and forth, and friction between molecules generates heat. (Pedantically: 2.45 GHz isn’t a sharp water resonance; it’s on the rising shoulder of liquid water’s broad Debye relaxation, which peaks around 10–20 GHz. Water’s sharp rotational lines sit higher up — 22 and 183 GHz — and live in the atmosphere, where they matter to us. 2.45 GHz was chosen for ovens because it’s an unregulated industrial band that heats food well and penetrates a few cm deep.)

The Parkes perytons. From 1998, the 64-metre radio dish at Parkes in New South Wales — the same telescope that relayed the Apollo 11 moon-landing broadcast — started seeing bright millisecond-long bursts. They named them “perytons,” after a mythical antlered deer that casts a human-shaped shadow. The timing was suspicious: the bursts clustered around lunchtime, almost always on weekdays. For over a decade no one knew what they were.

It got worse in 2007 when Parkes also detected the first fast radio burst (FRB) — a millisecond flash from cosmological distances. Perytons looked close enough to FRBs to threaten the whole young field. Were FRBs even real?

In 2015, Emily Petroff and collaborators (MNRAS 451, 3933) installed an RFI monitor on-site and cracked it: the culprit was the cafeteria microwave oven. If a hungry astronomer opened the door before the timer hit zero, the magnetron’s shut-down transient leaked a broadband burst that the telescope, pointed just right, caught as a peryton. Two decades of mysterious signals, solved by a kitchen appliance.

The double point. First: cities don’t just pollute with visible light — they scream in microwaves too. Every wifi router, phone, radar, and oven pumps anthropogenic radiation into the microwave sky. Put a telescope like SPT in Paris and it would drown in RFI. Second: even if you shielded all that out, the atmosphere itself absorbs at our actual observation bands (SPT works at 95, 150, 220 GHz) — driven by the water-vapour rotational line at 183 GHz and its pressure-broadened wings. That’s the “humidity” bullet from the last slide, now with teeth. Going up a mountain — and eventually to the driest place on Earth — solves both problems at once. Next slide.

(Strictly: microwave ovens themselves aren’t a direct problem for CMB telescopes — 2.45 GHz is ~40–90× below SPT’s bands. The Parkes story works because their multibeam receiver was at 1.4 GHz, and magnetron shut-down leaks broadband. The real enemy for us is atmospheric water vapour.)

So we go up the mountain:

Pic du Midi de Bigorre (2877 m). Credit: Matthieu Pinau

Observatories have lived on mountain tops since the 1800s — above the lights, above the weather, above as much atmosphere as you can climb over. But clouds aren’t even the real enemy.

The sky isn’t transparent

• Even on a perfectly clear day, the air is full of water vapor
• Water vapor absorbs the cosmic microbrew microwave background (CMB)
  — and, worse, glows back

CMB

Credit: NASA
via Wikimedia Commons

Option 1: go to space

• No atmosphere, no weather — can observe everything

COBE (1992) · NASA

WMAP (2001) · NASA

Planck (2009) · ESA

COBE (1989–1993) — Discovered the cosmic microwave background anisotropies (tiny one-part-in-100,000 fluctuations — the seeds of all structure). FIRAS also measured the spectrum as a perfect blackbody at 2.725 K. Nobel Prize 2006 (Mather & Smoot). Resolution ~7° — too coarse to resolve the acoustic peaks.

Local hook: George Smoot — one of the two COBE Nobelists — has been affiliated with Université Paris Cité since his prize. Worth a mention given the venue.

WMAP (2001–2010) — Measured the acoustic peaks and solidified our current model of the universe, ΛCDM. Age, composition, flatness — all pinned down precisely.

Planck (2009–2013) — Perhaps the most successful cosmology experiment of all time. Ushered in precision cosmology: gold-standard CMB temperature and polarization maps, CMB lensing, galaxy-cluster catalogues via the SZ effect.

Correction to a common misconception: COBE did NOT detect the first acoustic peak — that came from balloon experiments around 2000 (BOOMERanG, MAXIMA), then nailed by WMAP.

From detection to precision

Bigger telescope → sharper map. But rockets are expensive: there’s a lighter way up.

Credit: NASA

The iconic progression: CMB sky as measured by COBE (blurry), WMAP (sharper), Planck (razor-sharp). One part in 100,000 — colour scale pushed to the limit. In ~20 years we went from seeing structure to measuring it precisely enough to pin down the age, geometry, composition, and expansion of the universe.

Option 2: fly a balloon

• Rise above ~99% of the atmosphere; almost as good as space

ANITA-IV flight path. Gorham et al. 2019

Cheap, but you can only fly for weeks and the balloon has to land somewhere.

This is ANITA-IV — a neutrino balloon experiment, not CMB, but the flight path is the perfect illustration: three full loops around Antarctica over 28 days before landing near the South Pole. Cheap, but this is why balloons aren’t a permanent solution. Fig. 3 from Gorham et al., arXiv:1902.04005.

Option 3: stay on the ground

• Ground-based telescopes can be enormous, run for years, be upgraded

Where on Earth is dry enough?

Atacama Desert, Chile
5000 m · −20 °C to +10 °C
< 20 mm precipitation / year

Antarctic Plateau
2835 m · −60 °C to −25 °C
< 3 mm precipitation / year

Two mountainous deserts. Atacama is the obvious one; Antarctica is the surprise — by area it’s actually the largest desert on Earth. The Antarctic Plateau is lower than Chajnantor but far drier.

Credits: Atacama — Deborah Kellner (Simons Observatory). South Pole — SPT collaboration.

Getting to the bottom of the world

• The South Pole is only reachable by plane, October through February
• Christchurch
  → McMurdo
  → South Pole

About 40 people spend the winter there.

McMurdo: landing on sea ice

Photo: Alex Polak

That plane is an LC-130 Hercules — the ski-equipped C-130, operated by the US 109th Airlift Wing. It’s the only aircraft that can land at the South Pole proper. Here it’s on the McMurdo sea-ice runway with a marshaller in red ECW gear.

Big planes.

C-17 Globemaster III
Photo: Melanie Archipley

inside the C-17
Photo: Melanie Archipley

The South Pole Telescope

continuous night, for months
the coldest, driest air on Earth

• since 2007
• 16,000 pixels
• 10 m — 6× the size of space CMB telescopes

Photo: Aman Chokshi

Months of continuous night means stable observing: no day-night thermal cycles, no sun glare, a sky that just sits still. And the cold — about −60 °C in winter — means the air holds almost no water vapor, which is the real atmospheric enemy. High and dry we already established; here we add dark and cold.

The telescope has carried three cameras over about 20 years, each one more sensitive. SPT-3G has 16× more pixels than the original SPT-SZ and three times the frequency coverage. It’s still running.

It’s big!

Photo: Alex Polak

The SPT-3G camera

camera & commissioning team, 2017

a SPT-3G pixel
Credit: Adam Anderson

SPT observes by scanning

Video: Aman Chokshi

The telescope rocks back and forth across the same patch of southern sky, hour after hour. Scanning modulates the sky signal above the atmosphere’s and detectors’ low-frequency drift noise. Thousands of hours of data combine into a single deep map.

Event Horizon Telescope calibration

Video: Aman Chokshi

SPT participates in the Event Horizon Telescope — the global VLBI network that imaged the black holes at M87 and the Galactic Center. This clip shows a pointing calibration sequence.

SPT sees its forerunner

Starlink constellation
Credit: Allen Foster

COBE (detection)
Credit: Allen Foster

COBE (spacecraft)
Credit: NASA

We’ve come a long way from the first blurry COBE maps in 1992.

Thank you!

Photo: Aman Chokshi

Shrines

New Year’s

Telescope maintenance

Bonus slides — all photos from Melanie Archipley’s winter-over deck.