Thermospheric response to solar EUV variability probed with orbital-energy decay from NinjaSat CubeSat GNSS telemetry

Solar extreme ultraviolet (EUV) radiation heats and expands the thermosphere, modulating atmospheric density at altitudes in low Earth orbit and thereby increasing satellite drag and accelerating orbital decay. Previous studies have relied on empirical solar proxies, such as the F10.7 cm radio flux and Two-Line Element (TLE)-derived orbit solutions, which lack sufficient spectral resolution and temporal cadence to resolve wavelength-dependent thermospheric responses, and on orbital-decay metrics—most commonly the satellite altitude itself—that are not conserved quantities along the orbit and have rarely been validated against independent-density measurements. Here, we combine spectrally resolved EUV irradiance from the GOES-18 Extreme Ultraviolet and X-ray Irradiance Sensors at seven wavelengths (25.6–140.5 nm) with high-cadence GNSS position and velocity data from NinjaSat, an X-ray-astronomy 6U CubeSat orbiting at approximately 500–530 km altitude, over an 8-month period (April–November 2024). We derive the orbital-decay signal from the J₂ -corrected specific mechanical energy computed directly from the GNSS position and speed telemetry—a conserved quantity whose within-orbit scatter is about 50 times smaller than that of an altitude-based (point-mass energy) decay metric and which, unlike TLE/SGP4-based approaches, requires no propagated reference orbit. This energy-based decay rate correlates with independent thermospheric-density retrievals from the Swarm and GRACE-FO missions with Spearman coefficients of 0.85–0.92 (daily means), and its systematic-error budget (radiation pressure, unmodeled gravitational perturbations) remains below the statistical uncertainty of the daily decay rate. Cross-correlation analysis shows that the decay rate is strongly related to EUV and far-ultraviolet (FUV) irradiance (Spearman coefficients up to ~ 0.8), with a positive lag of order 1–3 days. Using a block-bootstrap procedure that accounts for the strong autocorrelation of both series, we find that the 95% confidence intervals of the estimated lag overlap across all seven wavelengths, extending from the 6-h search floor to approximately 100 h, while the central lag estimates lie at roughly 30–70 h depending on wavelength and analysis choice; a wavelength-dependent separation of the response times is not statistically resolved with the present dataset and remains a plausible but unconfirmed hypothesis pending independent constraints from physics-based thermosphere models. These results demonstrate that standard GNSS telemetry from a CubeSat, combined with a conserved orbital-energy metric and independent-mission validation, can quantitatively characterize the thermospheric drag response to solar EUV forcing, offering a scalable observational approach for space-weather monitoring and satellite drag modeling.