Direct Imaging and Gradient-Based Analysis of the 12 August 2026 Partial Solar Eclipse from a Freely Rotating High-Altitude Balloon

Technical Summary: Direct Imaging and Gradient-Based Analysis of the 12 August 2026 Partial Solar Eclipse from a Freely Rotating High-Altitude Balloon

Problem Statement
High-altitude balloons offer an inexpensive platform for atmospheric and astronomical observations, including solar eclipses. However, direct solar imaging from a balloon payload is challenging due to rapid, uncontrolled rotation of the gondola. Previous missions have relied on dedicated, active solar-pointing or stabilization systems (e.g., Peters [2]; Bowman et al. [3]) to maintain a solar bearing. This study addresses whether quantitative eclipse information can be recovered from a mechanically simple, passive system lacking active stabilization or solar pointing, utilizing the natural rotation and oscillation of a freely suspended payload to scan the sky.

Methodology
The experiment was conducted on 12 August 2026, launching from Oldenburg, Germany (53.126° N, 8.231° E). The payload consisted of a Stratoflights expanded-polystyrene probe box housing two Insta360 ONE RS cameras equipped with 4K Boost wide-angle lenses (16-mm full-frame equivalent, f/2.4).

• Filtering: Both cameras were covered with filter material sourced from BRESSER eclipse viewing glasses (compliant with EN ISO 12312-2:2015).
• Acquisition: The system operated in interval-photography mode (ISO 800, 1/1000 s, 10 s interval) without active pointing. Approximately 1,300 images were captured during the ascent, reaching a burst altitude of ~35 km.
• Image Selection: Only 21 frames containing a visible solar image were identified. These frames spanned from 19:19:15 to 20:38:21 CEST, covering both the increasing and decreasing partial phases, with the closest pre-maximum frame at 20:05:07 CEST.
• Analysis: To extract quantitative data, 200 × 200 pixel crops centered on the solar image were analyzed. No sharpening, interpolation, or synthetic reconstruction was applied. A two-dimensional gradient-based solar edge analysis was employed:
  1. Images were converted to grayscale and Gaussian-smoothed.
  2. The intensity gradient magnitude $G(x,y)$ was calculated to locate the solar edge (region of strongest intensity transition).
  3. The visible solar area ($A_i$) was determined by delineating the largest connected region at the maximum gradient.
  4. Image-derived obscuration ($O_i^{img}$) was calculated by normalizing $A_i$ against the first frame ($A_1$), which was assigned a known geometrical obscuration ($O_1^{geo} = 2.71\%$).
• Ground Truth: Geometrical obscuration was calculated topocentrically for the launch location, yielding a local maximum of 86.32% at 20:08:53 CEST.

Key Results
The study successfully recovered the eclipse progression from non-stabilized imagery:

• Correlation: The image-derived obscuration showed a strong correlation with the independently calculated geometrical obscuration (Pearson $r = 0.959$).
• Error Metrics: The mean absolute difference was 9.5 percentage points, and the root-mean-square error (RMSE) was 11.5 percentage points.
• Specific Agreement: Several frames showed close agreement (e.g., 19:29:14: 15.49% geometric vs. 16.48% image-derived; 20:05:07: 83.95% vs. 80.03%).
• Systematic Deviations: The method tended to underestimate obscuration, particularly in the later sequence. The authors attribute these deviations to uncorrected factors including the camera point-spread function, field-dependent lens response (vignetting/aberrations), filter geometry, camera-to-camera differences, and atmospheric scattering at low solar elevations.

Significance and Claims
The paper positions this work strictly as a proof-of-concept.

• Passive Viability: It demonstrates that passive payload rotation, combined with wide-field imaging and frequent interval photography, can yield both visually useful and semi-quantitative direct eclipse observations without the complexity of active solar pointing.
• Methodological Simplicity: The gradient-based edge analysis provides a robust method for estimating changing solar crescent areas without requiring absolute photometry or calibrated unocculted solar references.
• Future Calibration: The authors explicitly state that the current systematic errors preclude the method from being used as calibrated solar-limb metrology. Instead, the results define a calibration strategy for future experiments, which should include ground-based characterization of lens vignetting and point-spread functions, simultaneous cross-calibration of cameras, and the use of RAW data and inertial attitude sensors.
• Context: The study complements other 2026 balloon campaigns (e.g., in Spain) by providing data from a location outside the path of totality (86.3% obscuration) using low-cost, commercially available hardware accessible to students and the public.

In conclusion, the authors assert that while the current data contains substantial systematic uncertainties, the approach successfully validates the extraction of quantitative eclipse information from mechanically simple, non-stabilized balloon imagery, laying the groundwork for a calibrated, passive wide-field measurement method in future missions.