The instrument behind the study is not an ordinary radar. The Millstone Hill facility pairs a megawatt-class transmitter with a fully steerable 46-metre antenna, the Magnetosphere-Ionosphere Steerable Antenna, originally installed at a US Air Force site in 1963 and moved to Millstone Hill in 1978. The technique it exploits, incoherent scatter, detects the faint Thomson scattering of radio waves from free electrons in the ionosphere. Because the scattered signal’s spectrum encodes the temperature and density of the plasma, the radar can directly measure three quantities at once: electron density, electron temperature and ion temperature. No other ground-based method produces plasma temperatures as a direct diagnostic, which is precisely what made this eclipse an unmissable natural experiment.

What set this campaign apart was geometry. Rather than staring vertically at a single patch of sky, as most incoherent scatter studies have done, the team ran the antenna through wide 6-degree-elevation scans, sweeping 21 azimuth positions every 15 minutes. From its hilltop vantage point, the radar’s field of view stretched across a huge swathe of the continent, from roughly 25 to 55 degrees north latitude and 75 to 95 degrees west longitude, covering the eastern and central United States at altitudes from about 200 to 450 kilometres. The eclipse’s path of totality cut straight through the middle of that footprint, allowing the researchers to compare the ionosphere north and south of the shadow in near real time.

Raw radar returns, however, are only half the story. To isolate what the eclipse actually did, the team needed to know what the ionosphere would have done without it. They built an empirical background model using a fourth-degree polynomial regression, trained on more than two decades of Millstone Hill scanning experiments from 2003 to 2024, restricted to geomagnetically quiet conditions between the vernal equinox and summer solstice and excluding eclipse days. The model ingests seasonal and diurnal harmonics, lagged solar EUV flux and geomagnetic indices, and predicts the climatological state of each altitude-azimuth bin. Subtracting those predictions from the eclipse-day measurements yielded the eclipse’s fingerprint. The model’s errors were modest: typically 15 percent or less for electron density, under 10 percent for electron temperature and under 5 percent for ion temperature in the 200 to 400 kilometre range.

The headline result is a dramatic thinning of the plasma. In the F1 region, between 200 and 250 kilometres altitude north of the path of totality, electron density plunged by up to 50 percent, a drop of roughly 2 to 3 times ten-to-the-eleventh electrons per cubic metre, lasting about an hour from 19:00 to 20:00 Universal Time. The response began almost immediately: as soon as the penumbral shadow, the partial eclipse, touched the scanning region and obscuration climbed above zero percent, densities at the lowest altitudes started to fall. Higher up, the story unfolded in slow motion. In the topside ionosphere above 350 kilometres, the depletion arrived 30 minutes to an hour late, sank 40 to 50 percent below background levels, and did not fully recover even by the end of the day, still sitting 20 to 30 percent low at 23:00 UT.

This altitude-dependent lag is physics, not noise. The F1 region is governed by photochemistry: when solar extreme-ultraviolet radiation returns, ionisation resumes within minutes and new plasma floods the layer. The topside, by contrast, is fed from below, as freshly produced electrons diffuse upward along magnetic field lines, a much slower process. The afternoon timing compounded the delay, since the declining sun angle meant weaker photoionisation and elevated plasma temperatures that resisted the downward diffusion needed to refill the layer. The result was a two-speed ionosphere: a lower atmosphere that snapped back within roughly two hours and an upper one that stayed depleted for the rest of the day.

The temperatures told an equally striking story. Electron temperature, which tracks the obscuration level almost in lockstep, fell by up to 40 percent, about 850 Kelvin, at altitudes between 325 and 400 kilometres. The drop began when obscuration exceeded anywhere from 0 to 30 percent depending on altitude, and recovered almost as promptly once the shadow passed. Then came a counterintuitive twist: after the eclipse, electron temperatures rose 10 to 40 percent above background, up to 700 or 800 Kelvin, because the lingering scarcity of electrons left the survivors carrying an outsized share of the remaining solar energy. Ion temperature behaved differently again, dropping by up to 20 percent, or 225 Kelvin, in a narrow band parallel to and just south of the totality path, concentrated at two altitude ranges around 200 to 225 and 300 to 325 kilometres. That 225-Kelvin plunge is among the largest ion-temperature decreases ever reported during a solar eclipse, and it appeared and vanished in near symmetry with the peak of the obscuration.

Perhaps the most puzzling finding was asymmetry. Densities and temperatures north of the eclipse path fell harder than those to the south, an imbalance the researchers attribute to eclipse-driven disturbance winds. As the Moon’s shadow races across the atmosphere, it cools the neutral air beneath it, generating pressure gradients and winds that trail the totality by roughly an hour, a pattern suggested by simulations of the 2017 eclipse. Because the magnetic field over the scanning region tilts only slightly from vertical, the meridional component of these winds can push plasma up or down the field lines depending on which side of the path it sits. North of totality, the winds appear to have dragged plasma downward, deepening the depletion; south of it, upward winds partially cushioned the loss. As the disturbance winds faded and normal poleward circulation resumed about an hour and a half later, the asymmetry dissolved into a uniform depletion across all latitudes.

The observations also exposed gaps in the models. Simulations such as GITM predict the strongest ion-temperature cooling at high altitudes, around 200 Kelvin at 400 kilometres, yet the radar saw the deepest cooling at 200 to 325 kilometres and comparatively little above 350. Simulations also forecast a broad cooling zone spanning at least 20 degrees in latitude and longitude, whereas the measurements showed a narrow depletion band alongside regions of mild warming, a complex pattern the authors suggest may arise from acoustic-gravity waves rippling outward from the cooled shadow, superimposing quasi-periodic ripples of 20 to 120 minutes on the overall cooling. Reconciling these discrepancies, the team notes, will require better energetics in coupled ionosphere-thermosphere models, including more realistic light ion species and upper boundary conditions.

There is a practical coda, too. The radar band increasingly collides with a modern nuisance: the Starlink satellite belt at 550 kilometres altitude, whose hard metallic surfaces can swamp the delicate incoherent scatter signal. The team found that at the low-elevation, off-vertical geometry used here, the contamination does not corrupt ionospheric measurements at the altitudes that matter, a reassuring result for a technique whose future depends on coexisting with megaconstellations. For over a century, since the first ionospheric eclipse studies of the 1920s, eclipses have served as controlled experiments for near-Earth space. The 2024 event, scanned across a continent rather than through a single beam, showed that the ionosphere is not one layer responding to one switch, but a vertically stratified system in which chemistry, diffusion and wind each answer the Moon’s shadow on their own clock, and some answers are still arriving hours after the Sun comes back.

Subject of Research: Ionospheric response to the 8 April 2024 total solar eclipse observed by the Millstone Hill incoherent scatter radar

Article Title: Incoherent Scatter Radar Observations of the F Region of the Ionosphere above the Eastern US during the April 2024 Solar Eclipse

Article References: Derghazarian, S., Goncharenko, L. P., Serrano, M. C., Zhang, S.-R., Erickson, P. J., Coster, A. J., Singh, D., & Rideout, W. (2026). Incoherent Scatter Radar Observations of the F Region of the Ionosphere above the Eastern US during the April 2024 Solar Eclipse. Annales Geophysicae, 44(2), 1109-1130. https://doi.org/10.5194/angeo-44-1109-2026

Image Credits: AI Generated

DOI: 10.5194/angeo-44-1109-2026

Keywords: ionosphere, solar eclipse, incoherent scatter radar, Millstone Hill, electron density, electron temperature, ion temperature, thermosphere, MIT Haystack Observatory, F region, totality path, space weather