Design and Analysis of Spaceborne Hyperspectral Imaging System for Coastal Studies

Abstract

Hyperspectral payloads with high spatial and spectral resolution, combined with a wide field of view, are crucial for tackling the complexity of coastal and estuarine water ecosystems, enabling effective monitoring of water quality and ecological conditions. This study introduces a modular spectrometer design utilizing multiple sub-modules in an extended slit configuration. The system delivers a spectral resolution of 5 nm (400–1000 nm) and 10 nm (1000–2500 nm), a spatial resolution of 20 m, and a swath width of 80 km. Smile and keystone distortions are maintained below 1/5 of a pixel. Using Modran to simulate solar irradiance, the SNR of different targets under typical background conditions is calculated. Compared to conventional designs, the proposed modular approach provides compactness and high fidelity, effectively addressing size and optical aberration challenges. The simulation results confirm the system’s robustness, setting a benchmark for next-generation coast observation missions, particularly in coastal monitoring, underwater exploration, and dynamic environmental change tracking.

1. Introduction

The intensifying impacts of human activities and climate change present substantial challenges for monitoring coast and aquatic ecosystems, such as pronounced spatial heterogeneity, temporal variability, and limitations in current observation methods. Comprehensive, accurate, and real-time monitoring and management are essential for advancing scientific research, guiding environmental policy decisions, and promoting sustainable marine resource management. In recent decades, remote sensing has become a crucial tool for addressing these challenges, providing large-scale and synchronized observations. However, the low spectral resolution of traditional multispectral remote sensing limits its capacity to accurately distinguish chlorophyll, algae, and aquatic vegetation with closely overlapping spectral features. Hyperspectral remote sensing addresses the limitations of traditional methods by providing significantly higher spectral resolution, enabling detailed detection and analysis of aquatic ecosystem dynamics. Its capability to simultaneously capture high-resolution spatial and spectral data revolutionizes coast monitoring, enhancing the identification and quantification of subtle oceanic variations and complex oceanographic phenomena [1,2,3,4,5].

Spectral dispersion technologies are integral to hyperspectral remote sensing, with each offering distinct advantages suited to specific applications. Prism-based dispersion provides simplicity and broad spectral coverage but is limited by nonlinear dispersion and low spectral resolution. In order to realize the detection of large image field (long slit), the spectrometer has a large volume and poor environment adaptation, such as Germany’s EnMAP [6,7,8] and China’s Tiangong-1 [9,10] and EO-1 of NASA [11,12]. The nonlinear dispersion of these loads significantly degrades image quality. Fourier transform spectrometers achieve ultra-high spectral resolution over broad spectral ranges but are mechanically complex and highly sensitive to vibrations. They are suitable for atmospheric monitoring and applications demanding ultra-high spectral resolution, including gas detection and vertical profile measurements. Convex-grating dispersion, in contrast, is distinguished by its compact design and integration of dispersion and focusing functions. By eliminating the need for additional focusing optics, convex gratings greatly reduce the size and complexity of optical systems, making them well-suited for miniaturized spaceborne hyperspectral sensors. Additionally, convex gratings offer high spectral resolution and energy efficiency, extending their applicability across ultraviolet to near-infrared ranges. These advantages have been validated in missions such as HICO for coastal monitoring [13,14], PRISMA for land and ocean observation [15,16], ZY-1-02D for comprehensive resource mapping [17,18], and the AHSI payload aboard the GF-5 satellite [19,20,21]. Convex-grating dispersion represents a revolutionary advancement in hyperspectral remote sensing, addressing the increasing demand for high-resolution, compact, and efficient observation systems. The parameters of these spectrum satellites are shown in

Table 1. Performance comparison of spectral satellites.

Satellite	Launch	Spectral Range/μm	Spatial Resolution/m	Spectral Resolution/nm	Width/km	Type
PACE	2024	0.34–0.89	70	5	2700	hyperspectral
HySIS	2018	0.40–2.50	30	10	30	hyperspectral
PRISM	2019	0.40–2.50	30	10	30	hyperspectral
EnMap	2022	0.42–2.45	30	6.5–10	30	hyperspectral
HICO	2009	0.35–1.08	90	10	42	hyperspectral
GF-5	2018	0.39–2.50	30	5	30	hyperspectral
DESIS	2018	0.40–1.00	30	2.55	30	hyperspectral
GCOM-C	2017	0.40–12.00	250/500/1000	-	1000	multispectral
Oceansat-3	2022	0.40–1.01	360	-	1400	multispectral
VIIRS	2011	0.40–12.40	375/750 m	-	3060	multispectral
Sentinel-3	2016	0.40–1.02	300	-	1270	multispectral
This design	-	0.40–2.50	20	5–10	80	hyperspectral

.

Future satellites planned for ocean monitoring include PACE, SBG, and GLIMR. However, the highest performance index of these loads at 500 km orbit is 36 spectral segments, a spectral resolution of 10 nm, and a spatial resolution of 80 m [22]. Coastal studies require high spatial resolution (~20 m) to accurately capture fine-scale variations, along with hyperspectral imaging to effectively analyze the complex optical properties of coastal waters from space. Additionally, hyperspectral data facilitate bathymetric measurements, as demonstrated by Lee and Carder [23,24].

This paper introduces an innovative oceanic hyperspectral satellite imaging spectrometer featuring high-resolution and wide-swath capabilities. The payload operates across a wavelength range of 0.4–2.5 μm, achieving a spectral resolution of 5 nm in the visible-near-infrared range and 10 nm in the shortwave infrared, a spatial resolution of 20 m, a swath width of 80 km, and a high SNR (>500), which is needed for coastal and estuarine water studies. Compared to satellites such as HySIS, PRISMA, EnMAP, and GF-5, it delivers a 2.7-fold increase in swath width and a 1.5-fold improvement in spatial resolution. These advancements are driven by cutting-edge optical innovations, such as an off-axis three-mirror anastigmatic telescope with freeform surfaces and a modular segmented spectrometer design, ensuring high performance within a compact form factor. This transformative system resolves key trade-offs in oceanic monitoring, enhancing capabilities in marine ecology, oceanography, and climate science, while enabling precise and comprehensive monitoring for research and policymaking.

2. System Components

In this paper, a system simulation analysis is carried out on the basis of the existing detectors, illustrated in Figure 1, which operates in a line-scan pushbroom mode, covering a spectral range of 0.4–2.5 µm. The system achieves a cross-track field of view of 9.16° and a spatial resolution of 20 m at an orbital altitude of 500 km. It collects hyperspectral data over an 80 km swath, with an off-axis F/2.75 three-mirror anastigmatic telescope featuring freeform surfaces and an effective diameter of 245 mm. The optical path is divided into visible-near-infrared (VNIR) and shortwave infrared (SWIR) channels through a field-of-view separator. Two channels are integrated into long-slit Offner convex-grating spectrometer modules and coupled with CMOS and HgCdTe detectors. The VNIR band consists of four 1024 × 256-pixel, 27 µm detectors, and is the same as the SWIR bands. The payload is compact, with dimensions of 1427 × 963 × 620 mm³.

Table 2. Key parameters of the proposed system.

Parameter	VNIR	SWIR
Orbit altitude	500 km
Spectral range	0.4–1.0 µm	1.0–2.5 µm
Spatial resolution	20 m	20 m
Spectral resolution	5 nm	10 nm
Swath width	80 km	80 km
Detector size	1024 × 256 pixels (27 µm)	1024 × 256 pixels (27 µm)
Detector type	CMOS	HgCdTe
Focal length	675 mm
F-number	2.75
Radiometric resolution	14 bit
Integration time	2.5 ms

summarizes the key characteristics of this advanced hyperspectral imaging system.

2.1. Telescope Design

To achieve high-resolution imaging over a wide wavelength range, the telescope system adopts an all-reflective design. A triaxial structure was selected as the initial design, as shown in Figure 2. The object distance is considered to be equivalent at infinity, the vertex curvature radii of the primary mirror, the secondary mirror, and the third mirror are R₁, R₂, R₃, respectively, the center interval of the primary mirror and the secondary mirror is d₁, the center interval of the secondary mirror and the third mirror is d₂, the center interval of the third mirror to the image surface is l₃, and the focal length of the system is f’. The formulas for calculating the size of the structure from the dimension coefficient of the profile are as follows [25]:

$$R_1={\displaystyle \frac{2}{{\beta }_1{\beta }_2}}{f}^{\prime }$$

(1)

$$R_2={\displaystyle \frac{2{\alpha }_1}{{\beta }_2\left(1+{\beta }_1\right)}}{f}^{\prime }$$

(2)

$$R_3={\displaystyle \frac{2{\alpha }_1{\alpha }_2}{1+{\beta }_2}}{f}^{\prime }$$

(3)

$$d_1={\displaystyle \frac{1-{\alpha }_1}{{\beta }_1{\beta }_2}}{f}^{\prime }$$

(4)

$$d_2={\displaystyle \frac{{\alpha }_1\left(1-{\alpha }_2\right)}{{\beta }_2}}{f}^{\prime }$$

(5)

$$l_3={\alpha }_1{\alpha }_2{f}^{\prime }$$

(6)

In the above formulas, α₁ is the masking ratio of the secondary mirror to the primary mirror, α₂ is the masking ratio of the three mirrors to the secondary mirror, β₁ is the magnification of the secondary mirror, and β₂ is the magnification of the three mirrors. To make the system compact and small in size, take |d₁| = |d₂| to obtain α₁ = 0.5. In this case, the primary mirror and the three-mirror coplanar and have the most compact structure [26].

The initial structural parameters can be effectively calculated using third-order aberration theory. However, in a super-telephoto system, the large separation between the primary, secondary, and tertiary mirrors poses challenges in maintaining system stability and practical engineering implementation. To address this, freeform surface is introduced to increase the number of free variables, reduce system volume, and correct aberrations. Additionally, an off-axis design is implemented to expand the field of view and eliminate obstruction. The mathematical expression for an even aspherical surface is given by

$$z={\displaystyle \frac{c{r}^2}{1+\sqrt{1-(1+k)c^2{r}^2}}}+{\displaystyle \sum _{i=1}^N{A}_i}{E}_i(x,y)$$

(7)

Here, z is the radial height, c is the vertex curvature, k is the conic constant, r is the radial ray coordinate, N is the total number of polynomial coefficients in the series, A_i is the coefficient of extension polynomial, and E_i is Zernike polynomial.

Table 3. Parameters of the telescope.

Parameter	Primary	Secondary	Tertiary
R	−6140.43 mm	−1437.44 mm	−1476.51 mm
−e2	−2.77	−1.59	−0.18
d	−969.80 mm	998.92 mm	1381.49 mm
Optical structure	Three-Mirror Anastigmat (TMA)
Entrance pupil diameter	245 mm
Field of view	9.16°
Spatial resolution	20 m
Swath width	80 km@500 km
Focal length	675 mm
F-number	2.75
Spectral range	0.4–2.5

summarizes the telescope parameters, and the telescope structure is shown in Figure 3. The use of freeform optical surfaces significantly reduces the system’s size and weight compared to traditional optical systems. This compact design improves performance by enabling higher spatial resolution and a wider field of view, which are critical for high precision coast monitoring. Freeform optics also offer greater flexibility in optimizing the trade-off between spectral coverage and spatial resolution, thus meeting the stringent requirements for a wide field of view, high resolution, and high signal-to-noise ratio imaging [27,28]. This feature is especially beneficial for applications in dynamic coast monitoring and underwater target detection.

The modulation transfer function (MTF) indicates the system’s ability to retain spatial detail at different wavelengths. The MTF is defined as the ratio of relative image contrast to relative object contrast, and it quantifies the ability of an optical system to retain detail and the response of the system to different spatial frequencies. Achieving >0.85@18.52 lp/mm ensures exceptional imaging quality for both spectral and spatial dimensions, as shown in Figure 4. The root-mean-square (RMS) spot radius is consistently below 6.0 µm, corresponding to less than half a pixel size, as shown in Figure 5. The spot size was quantified using the RMS spot radius, which calculates the RMS radius of the intersection of all the rays to the chief ray. This value is used to quantify the effect of system aberrations. The expression is as follows:

$$r_{RMS}=\sqrt{{\displaystyle \frac{1}{N}}{\displaystyle \sum _{i=1}^N(x_i^2+y_i^2)}}$$

(8)

2.2. Spectrometer Design

Traditional Offner spectrometers encounter significant challenges as slit length increases, including greater system size, increased complexity, and difficulty in aberration control. For example, an Offner spectrometer with a 48 mm slit requires an instrument length of approximately 320 mm. Increasing the slit to 108 mm extends the spectrometer length to over 1500 mm, rendering it impractical for spaceborne applications [29,30]. Additionally, the reliance on large-scale infrared array detectors increases significantly, inevitably driving up the instrument’s development cost.

To address these challenges, this study introduces a modular spectrometer design that integrates multiple smaller spectrometers into an extended slit configuration. Building on the concept of field-of-view segmentation, the telescope’s wide swath is partitioned into multiple narrow sub-fields. Each sub-FOV is mapped to a dedicated spectral imaging system, and multiple independent systems are seamlessly integrated to reconstruct the full imaging swath. This approach minimizes system size, simplifies manufacturing, and enhances alignment accuracy while preserving high spectral fidelity. The slit system consists of four staggered sub-modules, each measuring 27 mm in length, yielding a total slit length of 108 mm. Figure 6 presents the slit splicing diagram.

However, implementing multiple systems with stringent requirements for precise installation and registration poses significant challenges, including the following:

• Arranging the pixels from four slits in two rows into a single straight line;
• The spectral coregistration of the VNIR and SWIR spectrometers;
• Each spectrometer and slit installation problems;
• The stray light problem in the imaging spectrometer.

In response to the above key technical challenges, we propose relevant technical approaches and application cases to support technical feasibility, including the following:

✓

Due to position deviations, time delays, and varying viewing angles among multiple detectors, geometric and radiometric corrections are required to ensure that the final image is continuous and seamless.

(1)

Geometric Correction: the relative position of each detector is determined and adjusted using onboard GPS and an imaging geometry model.

(2)

Radiometric Correction: Laboratory pre-calibration or onboard calibration ensures radiometric consistency across detectors. Overlapping regions of adjacent detectors are used to smooth brightness transitions, minimize stitching artifacts, and ensure seamless alignment.

(3)

Time synchronization and dynamic compensation: utilizing satellite attitude adjustments, time synchronization is applied to ensure all detectors capture images with a unified time reference.

(4)

Data fusion and seamless stitching: following geometric and radiometric corrections, as well as time synchronization, multi-detector imagery is fused to produce a seamless, large-scale remote sensing image.

(5)

Application case: In the SWIR channel of the GF-5 satellite’s AHSI payload, four 512 × 512 mercury cadmium telluride (HgCdTe) focal plane detectors are precisely spliced together, resulting in a 2048 × 512 SWIR focal plane detector [19]. This detector splicing method is similarly applied in the long-wave infrared (LWIR) channel of TIRI, a small satellite developed by CASEarth [31].

✓

Spectral registration is a crucial step in aligning multi-band imagery across different spectral bands, such as VNIR and SWIR. By applying image registration via control points or automatic feature matching, along with geometric transformation (e.g., resampling) and radiometric correction, images from different bands can be seamlessly integrated to produce a high-quality hyperspectral image. Satellites such as AHSI, PRISMA, and EnMAP achieve image registration across different spectral bands, including VNIR and SWIR, through accurate geometric and radiometric correction.

✓

In hyperspectral imaging systems, accurate remote sensing measurements and high image quality rely on precise optical and mechanical alignment. This is particularly important when imaging involves multiple spectrometers and detectors, as installation accuracy is crucial for geometric registration, spectral registration, and radiometric consistency in the final image. The alignment of optical elements, high-precision support structures, fine-tuning screws, and precision positioning devices (e.g., laser rangefinders) are key to ensuring high-quality imaging. Furthermore, since each spectrometer can be independently calibrated and tested prior to mounting on a fixed support bracket, installation complexity is significantly minimized.

✓

Stray light’s impact on imaging spectrometers is a critical concern, particularly in hyperspectral imaging systems, where it can significantly degrade image quality and data accuracy. The presence of a slit in linear array push-scan spectrometers effectively minimizes the entry of external stray light by restricting off-axis light paths. To mitigate internal stray light, various optical techniques are employed. Baffles and apertures strategically placed within the optical system confine light propagation, preventing unwanted reflections from reaching the sensor. Advanced optical coatings, such as anti-reflective and high-absorption coatings, further suppress stray light by reducing undesired reflections and scattering. Blackening treatments, including anodized aluminum coatings and carbon-based materials, effectively absorb stray light, enhancing system contrast. These stray light suppression techniques have been successfully validated in operational hyperspectral missions, such as AMMIS [32], demonstrating their effectiveness in improving radiometric accuracy.

Furthermore, the optimization of the Offner convex-grating spectrometer has led to notable improvements in spectral performance, as illustrated in Figure 7 and Figure 8, with detailed system specifications presented in

Table 4. Design specifications of imaging spectrometer.

Parameter	VNIR	SWIR
Spectral range	0.4–1.0 µm	1.0–2.5 µm
Spectral resolution	5 nm	10 nm
Detector pixel size	27 µm	27 µm
Detector array size	1024 × 256 × 4	1024 × 256 × 4
Band number	120	150
NA	0.182	0.182
Slit length	27.648 mm ± 10 μm	27.648 mm ± 10 μm
Slit width	27 ± 0.5 μm	27 ± 0.5 μm
Slit material	Fused Silica
Straightness of the slit	≤2 μm
Edge smoothness of the slit	Ra ≤ 0.2 μm
Dispersion width	3.24 mm	4.05 mm
Smile distortion	≤1/5 pixel	≤1/10 pixel
Keystone distortion/pixel	≤1/10 pixel	≤1/10 pixel

. The refined design enhances spectral resolution, minimizes optical aberrations, and ensures a more uniform signal-to-noise ratio (SNR) across the spectral range. These enhancements are crucial for ensuring high-fidelity spectral data, particularly in applications requiring precise material identification and atmospheric correction.

The spectrometer demonstrates exceptional spectral imaging performance. As illustrated in Figure 9, within the 0.4–1.0 μm spectral range, the MTF exceeds 0.70 at 18.52 lp/mm at the central wavelength of 0.7 μm. Similarly, as depicted in Figure 10, within the 1.0–2.5 μm range, the MTF surpasses 0.69 at 18.52 lp/mm at 1.75 μm. These results confirm the system’s high spatial resolution and optical precision across both spectral ranges.

Smile and keystone distortions are common optical aberrations in spectral imagers that significantly affect image geometry, especially during spectral scanning. Smile distortion manifests as a spectral shift of the sensor across the entire field of view (FOV), causing spectral misalignment and potentially degrading spectral calibration accuracy. This distortion often results in a curved shape when spectral lines are viewed along the detector array. Keystone distortion appears as a spatial misalignment between different spectral bands, leading to geometric inconsistencies across the spectral axis. This misalignment may cause difficulties in accurately registering multi-band images and disrupt spatial continuity.

The distortions, which are visualized in Figure 11, must be minimized through precise optical design, detector alignment, and image post-processing techniques to ensure high-quality spectral data with minimal geometric error. Advanced spectral imagers incorporate calibration and correction algorithms to reduce the impact of smile and keystone image quality.

As shown in Figure 12 and Figure 13, the maximum spectral smile distortion was measured at 3.27 μm, while the maximum keystone distortion was observed at 0.98 μm in the VNIR band. Similarly, as depicted in Figure 14 and Figure 15, the maximum spectral smile distortion in the SWIR band was measured at 1.6 μm, with the maximum keystone distortion occurring at 0.76 μm. These results underscore the system’s ability to maintain high-fidelity spectral imaging across both the VNIR and SWIR bands, demonstrating effective optical design and distortion control. The precise characterization of these distortions provides valuable insights for calibration strategies, ensuring improved spectral accuracy and geometric integrity in hyperspectral imaging applications.

Additionally, we designed a single-slit spectrometer and compared its performance with a multi-slit spliced spectrometer array, as illustrated in Figure 16. The results indicate that the multi-slit spliced long-slit spectrometer achieves a 10-fold size reduction compared to a conventional single-slit spectrometer, while maintaining high spectral and spatial resolution.

2.3. Integration of Telescope and Spectrometer

The integrated telescope–spectrometer system ensures seamless imaging and high-performance coast observation. Incoming light is directed through the primary mirror and split into two optical paths by a field splitter: one for the visible-near infrared (VNIR) band and the other for the shortwave infrared (SWIR) band.

As illustrated in Figure 17, the high-resolution coast observation payload has been successfully simulated. Figure 18 presents the modulation transfer function (MTF) curves across different wavelengths. At the Nyquist frequency of 18.52 lp/mm, the MTF exceeds 0.68 for all observed wavelengths, demonstrating exceptional image quality and spectral fidelity.

2.4. Tolerance Analysis

Tolerance analysis is a critical step in optical system design, ensuring that the final system meets the desired performance specifications. Due to manufacturing and alignment limitations, optical components and assemblies inevitably exhibit deviations, including surface figure errors, misalignment tolerances, and coating thickness variations. These imperfections can degrade optical performance by introducing aberrations, reducing the MTF, and compromising overall image quality.

In this study, Zemax 2017 version and the Monte Carlo algorithm were employed to perform tolerance analysis. The specifications and tolerance ranges for each optical component are detailed in

Table 5. System tolerance distribution.

Parameter	Mirror	Radius/mm	Thickness to Next Mirror/mm	Decenter/mm	Tilt/°	Size/mm2
Telescope	Primary	6140.43 ± 5.00	969.80 ± 0.20	±1.00	±0.02	500 × 370
	Secondary	1437.44 ± 2.00	998.92 ± 0.20	±0.50	±0.01	280 × 236
	Tertiary	1476.51 ± 2.00	1381.49 ± 0.20	±0.50	±0.01	620 × 560
Spectroscopy of VNIR	Primary	69.32 ± 0.50	33.71 ± 0.10	±0.02	±0.01	52 × 28
	Secondary	35.57 ± 0.50	31.53 ± 0.10	±0.02	±0.01	7 × 7
	Tertiary	68.35 ± 0.50	66.67 ± 0.10	±0.02	±0.01	50 × 28
Spectroscopy of SWIR	Primary	70.34 ± 0.50	32.57 ± 0.10	±0.02	±0.01	54 × 30
	Secondary	33.36 ± 0.50	33.24 ± 0.10	±0.02	±0.01	7 × 7
	Tertiary	71.23 ± 0.50	70.19 ± 0.10	±0.02	±0.01	50 × 28

. Using a random sampling-based statistical approach, the Monte Carlo method generates numerous parameter variations within the specified tolerance range, evaluates system performance for each configuration, and computes the resulting performance distribution.

The tolerance assignment of the system is presented in Table 5. After 500 Monte Carlo simulations, the results indicate that the MTF exceeds 0.63 with a 90% probability in the VNIR band and greater than 0.59 in the SWIR band, demonstrating that the assigned tolerances are well-optimized. The results are summarized in

Table 6. Results of the system tolerance analysis.

Monte Carlo Probability/%	VNIR MTF @18.52 lp/mm	SWIR MTF @18.52 lp/mm
90	0.63	0.59
70	0.68	0.61
50	0.69	0.65
30	0.72	0.67

.

3. Signal-to-Noise Ratio Analysis

The SNR is influenced by factors such as detector sensitivity, optical transmittance, and background radiance. Comprehensive considerations, including optical design, manufacturing, alignment, electronic circuitry, and detector performance, indicate that the static transfer function of the high-resolution, ultra-wide-swath imaging spectrometer is predicted to exceed 0.35.

In this section, we will derive the relevant calculation formulas. The signal-to-noise ratio of the imaging spectrometer is determined by orbital altitude, spatial resolution, spectral resolution, and the performance of the detector, optical system, and electronic system. The SNR is calculated as follows [33,34,35,36,37]:

$$SNR={\displaystyle \frac{S}{\sigma }}$$

(9)

$$\sigma =\sqrt{S+{\sigma }_{dark}^2+{\sigma }_{read}^2+S_{back}}$$

(10)

where S is the number of signal electrons received on the imaging detector; σ is the noise electron number; σ_dark is the dark current noise and σ_read is the readout noise; and S_back is the background stray light. The radiance at the entrance pupil and the irradiance at the focal plane of the imaging spectrometer can be expressed by Equation (11):

$$E(\lambda )={\displaystyle \frac{\pi }{4F^2}}L(\lambda )\tau$$

(11)

where L(λ) is the radiance of the target’s reflected light, F is the optics F-number, E(λ) is the irradiance at the focal plane, and τ is related to the grating diffraction efficiency and the optical system transmission. The radiation travels through the optical system and is captured by the imaging detectors. The signal output can be given by the following equation:

$$S={\displaystyle {\int }_{{\lambda }_1}^{{\lambda }_2}\frac{E(\lambda )}{hc/\lambda }}\cdot \varphi \cdot A\cdot \eta \cdot t\cdot \tau d\lambda$$

(12)

where $\varphi$ is the detector filling factor, A is the pixel area, η is the average quantum efficiency of the detector, t is the integration time of the detector, h is the Planck constant, and c is the speed of light in a vacuum.

Considering that at 500 km orbit, the speed of the satellite is about 7.6 km/s, the detector integral time is calculated by the following formula:

$$t_{\mathrm{int}}={\displaystyle \frac{GSD}{v}}$$

(13)

where t_int is the integration time, GSD is the ground resolution, and v is the satellite speed. The calculated result is 2.63 ms, and the integration time of 2.5 ms is finally taken.

To verify the capability of the system in coastal observation, we selected several typical coastal materials as test subjects, including an abyssal water body, eutrophic water (e.g., algal blooms), vegetation, and buildings. Their surface albedos were 0.05, 0.15, 0.3, and 0.5, respectively. The radiance received by the imaging spectrometer at orbital altitude under typical atmospheric conditions was simulated using Modran 5.0 version. The simulation parameters are provided in

Table 7. Modran simulation parameters.

Parameters	Value
Model atmospheric	Mid-latitude summer
Type of atmospheric path	Slant path
Mode of execution	Radiance with scattering
CO2 maxing ratio	400 ppmv
Surface albedo	0.05\0.15\0.3\0.5
Temperature at first boundary	300 K
Visibility	23 km
Cloud/rain aerosol	No clouds or rain
Initial wavelength	0.4 μm
Final wavelength	2.5 μm
Zenith angle	180°
Solar altitude angle	30°

and final simulation results are shown in Figure 19.

We utilized two advanced detectors developed by the Shanghai Institute of Technical Physics, Chinese Academy of Sciences, specifically designed for remote sensing detection. The parameters of the spectrometers are shown in

Table 8. Main parameters of the spectrometer detectors.

Parameters	VNIR	SWIR
Response wavelength	0.4–1.0 μm	1.0–2.5 μm
Pixel size	1024 × 256 (27 μm)	1024 × 256 (27 μm)
Type	CMOS	HgCdTe
Filling factor	75%	85%
Readout noise	30e− rms/pixel	105e− rms/pixel(@233k)
Dark current	48e−/pixel/s	69e−/pixel/s(@233k)
Quantum efficiency	0.55	0.53
Integrating capacitance	30 fF	15 fF/30 fF
Saturation electron number	4.68 × 105e−	0.9 × 105/1.8 × 105e−
Integration time	2.5 ms

, and the spectral response curve of the detectors is shown in Figure 20.

The HgCdTe detector used in this study operates in a dual-gain mode, incorporating both low-gain and high-gain settings to optimize the hyperspectral imaging system’s performance across targets with varying reflectance, while effectively preventing detector saturation in high-reflectance scenarios. In high-reflectance environments (e.g., buildings or snow-covered surfaces), the detector is configured to operate in low-gain mode with a short integration time, thereby reducing the accumulated photon signal to prevent pixel saturation and preserve radiometric accuracy. Conversely, in low-light conditions (e.g., nighttime remote sensing or weak signal targets), the detector switches to high-gain mode with an extended integration time, enhancing the signal amplitude, improving the signal-to-noise ratio, and increasing detection sensitivity.

The gain adjustment strategy optimizes the dynamic range under complex illumination conditions, enhancing both the accuracy and applicability of radiometric measurements in hyperspectral imaging. By dynamically regulating gain levels, the system effectively prevents detector saturation, ensuring reliable data acquisition across varying reflectance scenarios.

In this study, the optical parameters of the spectrometer were input into grating simulation software to analyze the diffraction efficiency. Simulations were conducted at incident wavelengths of 600 nm and 1500 nm, using metallic aluminum as the base material. The diffraction efficiency results for the VNIR and SWIR gratings are presented in Figure 21 and Figure 22, showing that the efficiency for both spectral bands exceeds 0.44.

Finally, the SNR is calculated according to Formulas (9)–(12). The simulation calculation result is shown in Figure 23.

The exceptional performance of AHSI in detecting chlorophyll concentration in water bodies has been well demonstrated in previous studies [38]. Therefore, we conducted a comparative analysis of the signal-to-noise ratio (SNR) differences between the two systems. As can be seen from the

Table 9. Comparison of the SNRs with GF-5 AHSI.

Parameters		AHSI	Design
Solar zenith angle		30°
Surface albedo		0.5
Wavelength/nm	600	686	668
	900	369	583
	1200	452	734
	1500	460	625
	1700	405	579
	2400	194	258

, the proposed system can provide more than 500 of SNR and is significantly improved compared to AHSI in both the near-infrared and shortwave bands.

This robust performance ensures that the system can maintain a high SNR across diverse observational scenarios, meeting the stringent requirements for its applications, such as in coastal environmental monitoring, coastal water color detection, and underwater exploration. These results confirm the system’s readiness for practical deployment in demanding marine hyperspectral sensing missions.

This system is currently facing significant challenges in massive data transmission and processing, constrained by onboard computing capacity, storage limitations, and power consumption. To address the challenge of massive data volumes generated by hyperspectral remote sensing satellites, we are actively researching and overcoming key technical barriers in onboard data optimization and compressed transmission. Our ongoing work focuses on developing an integrated approach, combining lossless compression (CCSDS 123.0-B-2) and lossy compression (e.g., JPEG2000, SPIHT) to achieve an optimal balance between data reduction and information preservation. Additionally, we are investigating onboard spectral band selection, leveraging spectral feature extraction and target recognition algorithms to prioritize the transmission of critical spectral bands. However, implementing this method efficiently on a computationally constrained satellite platform remains a significant challenge.

Moreover, this study proposes an onboard radiometric calibration strategy combining a radiometer comparator and a solar diffuser, ensuring the long-term radiometric stability and data consistency of the hyperspectral remote sensing system. The radiometer comparator performs relative radiometric calibration by measuring the differential response between a known standard radiation source and the sensor, and the solar diffuser utilizes the Sun as a stable, high signal-to-noise reference source, providing radiometric calibration for visible to near-infrared bands and effectively compensating for long-term detector drift. In the future, this approach can be further enhanced by incorporating Lunar Calibration and Deep Space Calibration, establishing a multi-source calibration framework to improve the quantitative accuracy of remote sensing data, thereby meeting both scientific observation and engineering application requirements.

4. Conclusions

This study presents a novel high-resolution, wide-swath hyperspectral imaging payload, specifically designed for coast observation, addressing critical limitations in current coast remote sensing technologies. By leveraging an off-axis three-mirror telescope with freeform surfaces and a modular spliced spectrometer design, the system achieves significant advancements in imaging performance and compactness. The key results include a spectral resolution of 5 nm (VNIR) and 10 nm (SWIR), a spatial resolution of 20 m, and an 80 km swath width, with validation confirming an MTF exceeding 0.75 at the Nyquist frequency. Thanks to the realization of wide-band imaging, from the visible to the shortwave on the same system, the complementarity and correction between visible and shortwave images provide support for high-quality spectral imaging.

Compared to existing systems, such as HySIS, PRISMA, and EnMAP, the proposed design demonstrates up to 2.7 times the swath width and 1.5 times the spatial resolution, offering robust capabilities for coastal target detection, water color monitoring, and dynamic environmental monitoring. The modular spectrometer architecture simplifies manufacturing and alignment, while ensuring scalability and robustness for future hyperspectral missions.

This innovative payload bridges the gap between high performance and practical implementation, providing a transformative tool for advancing ocean science, supporting sustainable resource management, and addressing pressing environmental challenges. By setting a benchmark for next-generation coast observation technologies, this work contributes significantly to the field of hyperspectral remote sensing and its applications in marine ecology, climate science, and environmental monitoring.