HyperScout-H Instrument for Hera Mission

• HyperScout-H is a dual-use hyperspectral imager developed for ESA's Hera mission, featuring a Three-Mirror Anastigmat design for high-fidelity asteroid surface mapping.
• It employs a Fabry–Pérot multi-band filter array and advanced calibration techniques to achieve spectral accuracy within 3% and spatial resolutions down to approximately 0.5 m.
• The instrument’s robust processing pipeline—from raw data ingestion to georeferenced hyperspectral cubes—enables comprehensive analysis of surface composition, space weathering, and impact experiment outcomes.

The HyperScout-H (HS-H) instrument is a dual-use hyperspectral imager developed for the European Space Agency's Hera mission. Deployed as one of five core scientific payloads, HS-H plays a leading role in the hyperspectral and morphological characterization of the Didymos–Dimorphos binary near-Earth asteroid system, with a particular focus on quantifying planetary defense experiment outcomes following the NASA DART impact. HS-H uniquely combines a wide instantaneous field of view, robust snapshot imaging and spectral data acquisition in the 0.65–0.95 μm window, and spatial sampling capacities able to resolve surface features down to ≈0.5 m.

1. Instrument Design and Optical Configuration

HS-H is derived from the HyperScout-2 nanosatellite platform, featuring a snapshot acquisition system that utilizes a Three-Mirror Anastigmat (TMA) all-reflective telescope design. This architecture delivers a paraxial field of view of approximately 15.5° × 8.3°, with a ray-traced full field of ~15.9° × 9.9° after distortion. The system suppresses stray light with four metallic baffles and avoids refractive elements to minimize ghosting.

The TMA optical system focuses the scene onto a pixelated, back-illuminated CMOS focal plane, overlaid with a Fabry–Pérot multi-band filter array. Each 5×5 subpixel "macropixel" yields 25 discrete, non-overlapping channels spanning 0.650–0.952 μm. The overall array comprises 2,048 × 1,088 subpixels, or 409×217 full macropixels. The focal length is 41.25 mm, yielding an f-number of approximately 4.0, and the clear aperture is 10.3125 mm. The detector pixel pitch is 5.5 μm; the instrument mass is 5.145 kg, occupying 238×190×277 mm³.

In the paraxial approximation, the half-angle field of view in the $x$-direction is:

$$\theta_x \approx \arctan\left(\frac{0.5\,d_x}{f}\right)$$

Setting $d_x \approx 12\, \mathrm{mm}$ and $f = 41.25\,\mathrm{mm}$ yields $\theta_x \approx 7.75^\circ$, confirming the total FOV.

2. Spectral Capabilities and Calibration

Each macropixel resolves 25 spectral bands with respective central wavelengths $\lambda_i$ and full-width at half-maximum (FWHM) $\Delta\lambda_i$ between 0.0082 and 0.0222 μm. Spectral response is dictated by static filter functions $F^i(\lambda)$; there are no gratings or dispersion elements.

For calibration or simulation against laboratory/reference spectra $S(\lambda)$, the response for channel $i$ is computed via:

$$S^i_{\rm HS-H} = \frac{\displaystyle \int_{\lambda_{\min}}^{\lambda_{\max}} F^i(\lambda)\, S(\lambda)\, \lambda\, d\lambda}{\displaystyle \int_{\lambda_{\min}}^{\lambda_{\max}} F^i(\lambda)\, \lambda\, d\lambda}$$

Radiometric performance was evaluated through laboratory calibration campaigns by both cosine Remote Sensing BV and ESA/ESTEC, with master dark, flat, and distortion products agreeing to within 2%. The dark current median and standard deviation are temperature- and exposure-dependent:

$$m(t_{\mathrm{exp}}, T)[\mathrm{DN}] = 146 + (9.10\, t_{\mathrm{exp}} + 0.5) e^{0.1225 T}$$

$$\sigma_{\rm noise}(t_{\mathrm{exp}}, T)[\mathrm{DN}] = 12.57 + (1.18\, t_{\mathrm{exp}} - 0.0252) e^{0.1225 T}$$

Read noise is typically 12 DN (increasing to ~22 DN at longest $\lambda$); the linear full well is ≈4000 DN, yielding a dynamic range of approximately 330:1. The SNR for macropixel-averaged channels surpasses 200 under flat-field laboratory illumination, with SNR for subpixels on meteorite samples in the range 60–100.

Bad-pixel detection employs a ±5$\sigma$ threshold; masked pixels are interpolated using a local mean over a $5\times5$ kernel. Geometric distortion is calibrated via checkerboard targets, with polynomial fits for converting distorted to ideal focal-plane coordinates. In-flight calibration will be refined with star fields.

3. Data Processing and Algorithmic Pipeline

HS-H data are processed by the MOGI toolbox (Python/C++), structured in four stages:

1. Raw data ingestion and housekeeping
2. Calibration (application of dark/bias, flat-field, bad-pixel, and distortion corrections)
3. Demosaicing for hyperspectral cube assembly at subpixel scale
4. Georeferencing and map projection with SPICE kernels and classification

Demosaicing addresses the underdetermination of $\hat{f}(i, j, k)$ (normalized spectrum) and $b(i, j)$ (brightness):

• Macro-averaged: single spectrum per macropixel, spatial resolution limited to the macropixel grid.
• Spatial-ratio interpolation: under the assumption of spatially slowly-varying brightness, uses the ratio of neighboring subpixel measurements for improved subpixel reconstructions.
• Regularized inversion: utilizes a graph Laplacian prior for local spectral smoothness to recover full 3D cubes at subpixel scale.

Spectral parameter mapping algorithms include:

• Taxonomic classification: pixelwise clustering in the 25-dimensional vector space, employing Random Forest or Convolutional Neural Network classifiers trained minimizing cross-entropy loss (following taxonomies such as Mahlke et al. 2022).
• Spectral slope:

$$S = \frac{R(\lambda_2) - R(\lambda_1)}{\lambda_2 - \lambda_1}\quad[\%/\mu\mathrm{m}]$$

• Band depth at 0.9 μm:

$$\mathrm{BD} = 1 - \frac{R_\mathrm{band\,min}}{R_\mathrm{cont}}$$

• Band center ($\lambda_c$): quadratic fit in the continuum-removed spectrum to the local minimum.

4. Ground Validation and Performance Verification

Laboratory campaigns used two polished ordinary-chondrite meteorites (H5 El Hammami and L5 SaU 001) atop a 99% Spectralon reference under halogen illumination at 30° incidence, with a camera-target distance of approximately 1.03 m. Frames at $t_\mathrm{exp} = 1$ s (avoiding saturation) provided typical meteorite signals of 1000 DN (SNR 60–80).

Data processing encompassed dark/bias/flat corrections, extraction of Spectralon reflectance, and application of filter transfer functions. Across 30×30 macropixels, HS-H reflectance curves matched the laboratory SHADOWS spectro-goniometer data to within 2–3% in both band shape and depth.

Spatial resolution in laboratory configuration was approximately 0.68 mm/macropixel; flight scenarios yield ~13 cm/subpixel at 1 km range, ~5 m/macropixel at 10 km approach, and below 1 m/macropixel in the 4.5 km EXP phase. Spectral accuracy is $\Delta R/R \lesssim 3\%$ over all 25 channels, with band center errors under 10 nm.

5. Scientific Objectives and Investigation Strategies

HS-H is tasked with several primary science objectives:

1. Global compositional mapping: Extraction of surface taxonomic classifications and mafic mineral abundance via band center tracking over Didymos and Dimorphos.
2. Space weathering mapping: Analysis of spectral slope and band depth to distinguish fresh from weathered regions, including characterizing the DART impact crater and ejecta material.
3. Exogenous material identification: Detection of taxonomic outliers and albedo anomalies to identify possible foreign boulders or remnants of the impactor.

The large FOV (approximately 157 deg²) allows orbital dynamics and dust environment monitoring, supporting detection and temporal evolution of ejected plumes and mutual asteroid motion across different encounter phases (ECP 20–40 km, to COP 4.5 km).

6. Data Products and Expected Outputs

HS-H supports a hierarchical data product framework:

Level	Description
1B	Radiometrically and geometrically calibrated 25-band data cubes
2	Demosaiced reflectance cubes at subpixel sampling, georeferenced with SPICE kernels
3	Spectral parameter maps (taxonomy, slope, band depth, band center, albedo) and principal component/weathering indices
4	Science deliverables: compositional unit maps, weathering models, crater ejecta maps, global phase curves, and 3D surface-composition models

These outputs are intended to synergize with the other Hera payloads' acquisitions, providing comprehensive geomorphological unit characterization and supporting detailed planetary defense response analyses. The instrumental workflow, from optical design through data calibration to spectral mapping, has been validated in representative laboratory settings and is designed to facilitate robust asteroid spectral surface mapping throughout the Hera mission trajectory.