How Do You Measure Earth's Hydrogen Exosphere When Everything Else Glows Too?

From a halo orbit around the Earth-Sun L1 point, a gravitational balance point sunward of Earth, NASA's Carruthers Geocorona Observatory, the agency's first mission dedicated to Earth's exosphere, looks back at a tenuous cloud of hydrogen atoms stretching far beyond the atmosphere. That cloud, the exosphere, glows brightest in ultraviolet light at 1216 angstroms, called Lyman-alpha. Carruthers launched in September 2025 and began science operations in its halo orbit around the Earth-Sun Lagrange 1 point in January 2026. Its GeoCoronal Imager has two coaligned photometric imagers: a narrow-field imager for high spatial resolution near Earth's limb and a wide-field imager for the entire exosphere. Seeing the hydrogen glow, however, is not the same as isolating it.

The exosphere is faint, and the sky is full of other photons

Carruthers images contain more than exospheric hydrogen. Stars appear in the field. The Moon and outer planets add light. Interplanetary hydrogen, or IPH, scatters solar Lyman-alpha at the same 1216-angstrom wavelength, making it in-band contamination that mimics the signal. Near Earth's limb, oxygen emissions at 1304 and 1356 angstroms add out-of-band light, meaning light outside the hydrogen wavelength that still reaches the detector. Alex Zhang and colleagues at the University of Illinois Urbana-Champaign and Boston University describe algorithms to retrieve and remove these backgrounds on orbit.

A calibration pipeline built to subtract before it measures

The work follows the mission's data products from L0 raw images to L1C absolutely calibrated images in photons per square centimeter per second. L1B images already have instrument effects removed. The new pipeline then finds stars without a catalog, using a dynamic detection scheme that looks for bright outliers above the local background. The Moon and outer planets are masked using known orbits, and small gaps are filled by interpolation from nearby pixels.

IPH is harder because it shares the hydrogen wavelength. The team builds a reference map from off-nadir observations and from the outer annuli of wide-field images pointed at Earth's nadir. Those annuli lie beyond 25 Earth radii, where the exosphere contributes only about 1 percent of the IPH signal. The map is normalized by solar Lyman-alpha flux and scaled to each image using relative responsivity between channels and filters. Validation of the relative Ly-alpha responsivity retrieval found a negligible mean and a standard deviation of 0.25 percent.

Out-of-band oxygen light requires a filter sequence. The narrow-field imager cycles through LyaN (the narrowband Lyman-alpha filter), Open (unfiltered), CaF2, and SrF2 filters, with a second LyaN exposure in the operational sequence. CaF2 and SrF2 are long-pass filters that suppress Lyman-alpha to expose out-of-band light. The paper's validation used 30-minute exposures for Open, CaF2, and SrF2 and a 60-minute LyaN exposure, plus a 30-minute storm-time LyaN image. Each filter transmits a known mix of wavelengths. By comparing the images, the algorithm separates hydrogen Lyman-alpha, oxygen 1304, oxygen 1356, and a smooth background. Because sharp gradients near Earth's disk are sensitive to blur, the team also uses a deblurring method with wavelet denoising. The final retrieval solves for amplitudes and exponential decay rates with altitude.

What the synthetic sky revealed

The validation used a numerical image simulator, not the real sky. It generated synthetic raw exposures for both channels and processed them through the full calibration chain. The results give a clear sense of expected performance.

First, IPH retrieval recovered the simulated interplanetary hydrogen with a mean error of -0.13 percent for the narrow-field imager and 0.4 percent for the wide-field imager. The standard deviation stayed below 1.4 percent for both, well inside NASA's 30 percent accuracy requirement.

Second, out-of-band oxygen retrieval, measured relative to the target Lyman-alpha signal, had a mean error of 0.03 percent and a standard deviation of 0.88 percent, meeting the mission's 20 percent requirement.

Third, the full pipeline from L0 to L1C produced a percent error on the ground-truth exospheric radiance centered at -2.9 percent. The scatter was larger: standard deviations of about 12.4 percent for the narrow-field channel and 9.05 percent for the wide-field channel. The paper reports an approximately 3 percent mean bias, with both mean and scatter still within NASA's 30 percent threshold.

The caveats are built into the validation

These numbers come from simulations, not final flight measurements. The paper is an arXiv preprint, and the algorithms are now processing flight data, but the reported performance estimates depend on the simulator's assumptions about instrument behavior, noise, and the scene. On-orbit point-spread function, the blur signature of the optics, will be refined using stellar observations. The out-of-band retrieval assumes the scene stays stable across the filter sequence. The IPH validation map shows annular artifacts from daily downsampling of spacecraft geometry; operational maps should be smoother. Deblurring also adds noise variance at the L1C stage. And IPH removal relies on solar Lyman-alpha flux estimates from a previous processing cycle, with 30-minute to 24-hour latency, which the paper argues is negligible because solar Lyman-alpha flux is stable over about 14 days.

Why a faint hydrogen exosphere is worth this much trouble

Earth's exosphere is where the atmosphere meets the magnetosphere, radiation belts, and solar wind. Charge exchange with exospheric hydrogen can influence how long and severe geomagnetic storms become. Hydrogen escape is also a primary mechanism for permanent water loss on Earth, Venus, and Mars, though the role of magnetic fields and trapped plasma remains poorly understood. Carruthers is designed to map the exosphere's global structure and watch it change during storms.

To read that story, scientists first have to subtract the rest of the universe. The new calibration pipeline is a set of instructions for doing that carefully, with error bars that are measured rather than assumed. If it works as simulations suggest, the reward is a clearer view of how Earth leaks its atmosphere into space.