Capturing Artemis SLS at Launch Pad 39B

Two days before launch. I downloaded ESA's Sentinel-2C imagery of Kennedy Space Center at 10 metres per pixel via the STAC API. You can see SLS on Launch Complex 39B (circled), the crawlerway leading south to the Vehicle Assembly Building, and the Atlantic coastline.

Sentinel-2C / MSI · 10 m/pixel · March 30, 2026 · Downloaded from AWS via STAC API (earth-search.aws.element84.com) · Cloud Optimized GeoTIFF

Launch Day Weather Watch

I pulled imagery from three different satellites to watch the weather over Florida. MODIS on Terra sees the Space Coast up close from 705 km. VIIRS on NOAA-20 captures the full peninsula from 824 km. GOES-19 watches from geostationary orbit at 35,786 km, updating every minute. Between them: partly cloudy skies, clearing toward evening. Go for launch.

Left: MODIS / Terra (705 km) · Centre: VIIRS / NOAA-20 (824 km) · Right: GOES-19 / ABI (35,786 km) · NASA Worldview Snapshots API · April 1, 2026

Ignition

6:35 PM. Twin boosters and four RS-25 engines fire. I processed GOES-19's raw infrared data — the satellite watches from 35,786 km above the equator — and found a sudden heat spike at the exact coordinates of Launch Complex 39B. Each pixel covers roughly 2 km — the rocket's exhaust is diluted across that entire area — yet one pixel lights up clearly above the local background. That bright dot is the rocket.

GOES-19 / ABI · Band 7, 3.9 μm shortwave IR · Temporal difference vs 1 min before launch · 22:35 UTC · Raw data: s3://noaa-goes19/ABI-L2-MCMIPM/

The Trail

I built this animation from seven raw NetCDF files at 30-second intervals, from T−30 seconds through T+150 seconds. Two overlapping mesoscale sectors give this cadence. I subtracted the pre-launch image so only new heat remains. A dot of fire appears, brightens, drifts northeast, then vanishes.

GOES-19 / ABI · 3.9 µm temporal difference · T−30s to T+150s · 30-second cadence · 7 raw NetCDF files from NOAA S3

The Plume

I compared the same moment across three different wavelengths. In visible light — it's 6:36 PM, the sun is low — the plume is lost in the clouds. In infrared at 3.9 μm, the exhaust blazes like a lighthouse against the cool ocean. In the fire detection channel — the same algorithm NOAA uses to spot wildfires from space, comparing shortwave IR against the longwave background — the rocket produces a strong, unambiguous signal. Same physics as a wildfire, except this one is accelerating off the pad.

GOES-19 / ABI · Left: Band 2 (0.64 μm visible) · Centre: Band 7 (3.9 μm IR) · Right: Band 7 minus Band 14 (fire detection) · T+60s

Acoustic-coupled seismic signals at near-field stations

I searched for Raspberry Shake stations near the launch site — citizen-science seismographs built on Raspberry Pis, thousands deployed worldwide, all openly available through FDSN. I found stations at 23 km and 52 km from KSC, pulled their data, and ran it through ObsPy. Both showed strong acoustic-coupled seismic signals arriving within minutes of launch. The signal arrives at roughly the speed of sound (~340 m/s), not at seismic body-wave velocity (6 km/s) — confirming I was seeing atmospheric propagation, not an earthquake. The signal lasted over 2 minutes, matching the 126-second Solid Rocket Booster burn — each second of ascent sending a new wavefront from progressively higher altitude.

Raspberry Shake geophones at 23 km and 52 km from KSC. Cyan shading: expected acoustic window (~340 m/s). Red markers: P-wave (~6 km/s). Blue markers: S-wave (~3.5 km/s). Purple markers: Rg surface wave. Data: Raspberry Shake FDSN (AM network). Bandpass: 0.5–10 Hz. Processed with ObsPy. Analysed by Arushi Nath.

Infrasound detections across the network

SLS produces over 200 dB of acoustic energy at the pad — I wanted to see how far that signal travelled. I queried the Raspberry Shake/Boom network and the International Monitoring System (IMS, run by the Comprehensive Nuclear-Test-Ban Treaty Organization) for stations within range of KSC. I found 7 Raspberry Boom infrasound sensors within 320 km — and when I plotted their signal-to-noise ratios, the arrival times lined up perfectly with distance, consistent with atmospheric propagation at ~300 m/s.

Sliding-window SNR at 7 Raspberry Boom sensors within 320 km of KSC. Peak times reflect sliding-window centres (5-min windows), not signal onset. Detection band: 0.7–2.0 Hz. Data: Raspberry Shake FDSN. Analysed by Arushi Nath.

Listening from Toronto

I built my own infrasound station on my apartment balcony in Toronto: a Raspberry Boom sensor (sensitive below 20 Hz) inside a handmade wind-noise enclosure — a bucket with holes patched with steel wool to reduce wind turbulence while allowing pressure waves through. My Raspberry Boom (R3635) experienced a hardware failure before launch day. My Raspberry Shake seismometer (RC893) recorded continuously but did not detect the launch. Over 1,670 km, the acoustic energy spreads and dissipates through the atmosphere, and any remaining signal coupling into the ground was possibly below the sensor noise floor.

Raspberry Boom + bucket with steel-wool baffles (uncovered)

Deployed on my balcony · Toronto · CN Tower in background

I located Orion from Chile

This wasn't as simple as pointing a telescope. Precise ephemeris data for Orion only became available once the spacecraft's orbit was established after ICPS separation — before that, there was no reliable public prediction of where to look. Even with coordinates, Orion was racing across the sky at over 500 arcsec/min (~8.5″/sec) and still manoeuvring with uncertain position. I started with T70 — the widest field available (10° × 6.7°) — to maximize my chances of finding it. Four hours after launch, I pointed T70 in Chile and found it. My measured position matched JPL Horizons to within 2 pixels. I captured 6 frames, then switched to T75 — a 250mm telescope with finer resolution (1.72″/px vs 5.81″/px) — for the main photometry run.

T70 · Chile · Wide field (10° × 6.7°)

Can you spot Orion in this? · 8 frames, T70 Chile, Apr 2 02:48-02:53 UTC

Here is Orion moving across the field at 510″/min

First observation · April 2, 02:48 UTC · 1s exposure · 6 frames · Angular velocity 510″/min (Orion at ~35,000 km, ~25 min after TLI burn)

Position offset from JPL Horizons prediction: 11.9 arcsec (2 pixels)

T70 · April 2, 02:50 UTC · Orion and ICPS appear as a single blob — not yet resolved at this plate scale

R60 · My observatory · Nerpio, Spain · 0.305m Ritchey-Chrétien

I tried my own telescope first. Orion's orbit was carrying it south — from Dec −14° at launch to −26° by mid-mission. From Spain at 38°N latitude, the spacecraft was low on the southern horizon. My R60 in Nerpio couldn't reach it — below the telescope's pointing limit. The observing window from Europe was narrow and closing fast, so I turned to the iTelescope network in Chile and Australia, where Orion was high in the sky.

T75 · Chile · Deep field (2.3° × 1.5°) · 68 frames

Orion in the crosshairs · April 2, 07:01 UTC · SNR 47

That dot being tracked is this — NASA's Orion spacecraft, carrying four astronauts to the Moon. Image: NASA.

Measuring Orion's brightness

By hour 8, T75 in Chile had captured 68 frames of Orion and the ICPS separately. I ran photometry on both. Orion held steady around mag 11.5 in V-band — except for one frame where it flared to mag ~10, possibly a specular glint from the solar arrays or a thruster firing. After correcting for changing distance, the fading trend disappeared — confirming it was geometry, not the spacecraft dimming.

Raw light curve. One anomalous bright point at mag ~10.

Distance-corrected to 1 AU. Fading trend gone.

ICPS rotation period — measured from the light curve

The Interim Cryogenic Propulsion Stage — 13.7 m long, 5 m diameter, powered by a single RL-10 engine. After pushing Orion toward the Moon, it separates and tumbles on its own. Image: NASA.

In my data the ICPS faded from mag ~10 to ~12.75 in 30 minutes — its brightness varying as the tumbling cylinder presents different faces to the sun. I built a periodogram and phase-folded the data. The result: a rotation period of 0.376 hours (22.6 minutes), with a clean double-peaked curve consistent with a tumbling cylindrical body.

Raw light curve. ~2.75-magnitude swing in 30 minutes.

Phased to 0.376 h. Double-peaked = tumbling cylinder.

Space debris: a 1989 rocket body crossing the same line of sight

In the same field of view as Orion, a streaking object appeared — not physically close, but passing through the same patch of sky as seen from Earth, 22,605 km away in its own unrelated orbit. I measured its position across 11 1-second CCD exposures over 5 minutes using Tycho 13.3, plate-solved against ATLAS2, and formatted a standards-compliant ADES astrometric report. I submitted it to Project Pluto's satellite identification service.

As crewed missions beyond low Earth orbit become more frequent, documenting and reporting debris encounters like this — even line-of-sight coincidences — contributes to the growing catalogue that keeps future flights safe.

T75 · April 2, 06:52:52 UTC · Orion–ICPS separation: 12.9 arcmin · Orion–debris separation: 23.0 arcmin

Second night — 134 more frames from Spain and Chile

T80 in Spain caught Orion for the first time — 10 fast exposures at 0.1 seconds. Then T75 in Chile ran the longest session yet: 124 frames over nearly an hour.

Clouds across all three observatory sites. No data. Orion keeps flying — I wait.

Halfway to the Moon — still tracking

Orion is coasting toward the Moon at roughly 160,000 km from Earth. It's fainter now — I increased exposures from 1 second to 30 seconds to keep the signal. 60 images over nearly an hour from T75 in Chile.

T75 · Chile · Deep field (2.3° × 1.5°)

April 5, 05:02 UTC · 30s exposure · Angular velocity 11.4″/min — 45× slower than first acquisition · Bright object at bottom is Antares

Twenty hours before the flyby

Orion is deep in cislunar space, closing on the Moon. 20 images from T30, Siding Spring, Australia, at 60-second exposures. Exposure time has doubled again — the spacecraft is now twice as far as last night.

T30 · Australia · Narrow field (42′ × 28′)

April 6, 16:06 UTC · 60s exposure · Angular velocity 11.0″/min — speeding up as the Moon bends its path

Two hours after the lunar far-side flyby

The crew has just flown around the far side of the Moon and broken Apollo 13's distance record at 406,771 km. Orion emerged from behind the Moon and I caught it again — 25 images from the same T30 in Siding Spring, Australia.

T30 · Australia · Narrow field (42′ × 28′)

April 7, 15:12 UTC · 60s exposure · Washed out by moonlight — Orion at its farthest, the Moon at its closest

Heading home — still tracking from Chile

Orion is on the return leg, falling back toward Earth. 32 images at 60-second exposures from T75 in Chile.

T75 · Chile · Deep field (2.3° × 1.5°)

The last look — 17 hours before splashdown

The final telescope session of the mission. 150 frames at 10-second exposures over one hour from T75 in Chile — the largest single session, and the last time I would see Orion through a telescope before it re-entered the atmosphere. The Moon was only 28° away, washing the field, but the spacecraft was still there.

T75 · Chile · Deep field (2.3° × 1.5°) · 150 frames

Return, re-entry, and splashdown

On April 10 at 8:07 PM EDT, Orion splashed down in the Pacific Ocean ~50 miles off San Diego — the fastest crewed re-entry since Apollo, at Mach 35 and 3,000°F. I monitored re-entry and splashdown with 25+ seismic, infrasound, satellite IR, weather radar, and hydroacoustic sensors across Southern California.

Artemis II recovery zone · U.S. Coast Guard SOLAR chart · Source: USCG Navigation Center (navcen.uscg.gov)

Instruments and software

🔭

Telescopes — all operated remotely from Toronto via iTelescope

T70 · Chile

Samyang 135mm f/3.5 · FOV 9.97° × 6.66° · 5.81″/px

T75 · Chile

250mm f/3.8 · FOV 2.29° × 1.53° · 1.72″/px

T80 · Spain

65mm f/2.8 · FOV 10.37° × 6.93° · 5.81″/px

T30 · Australia

508mm f/4.4 · FOV 41.6′ × 27.8′ · 0.81″/px

R60 · Spain (mine)

0.305m RC · Nerpio · Too low for Orion

📡 Raspberry Shake + Boom (AM network), IMS/CTBTO, DWPF broadband, CI + AZ seismic (SCEDC) · 30+ stations
🛰 Sentinel-2C (ESA), GOES-19 + GOES-18 (NOAA), MODIS (NASA), VIIRS/NOAA-20, GLM lightning mapper
📡 NEXRAD KNKX/KSOX/KVTX weather radar, MBARI MARS hydrophone, GNSS TEC
💻 AstroPy, Photutils, ObsPy, Astrometry.net, Tycho 13.3, netCDF4, GDAL, Project Pluto
📊 Periodograms, differential photometry, ADES reports, STA/LTA triggers, envelope z-scores
📝 arXiv preprint forthcoming · Full data + code to be released · CC-BY-4.0