China's first quantitative hyperspectral intelligent computing satellite has successfully entered orbit, with CAS Xi'an Optics & Technology Co., Ltd. showcasing its hardware and software solutions in the race for "space computing power."

Leveraging the advantages of a novel interferometric imaging system, this satellite has achieved the highest spatial resolution and widest swath in the visible-near-infrared spectral band currently available in China, reaching 10m spatial resolution and 7nm spectral resolution—the highest commercial hyperspectral performance indicators—significantly enhancing my country's spaceborne hyperspectral data acquisition capabilities. With a swath width of 14km, it also possesses long-strip continuous pushbroom imaging capabilities, greatly improving data acquisition efficiency and enabling the acquisition of up to 1 million km² of 10m hyperspectral remote sensing images daily. The use of the new interferometric spectral imaging technology allows for freely defined center wavelengths, providing a solid data foundation for quantitative and refined remote sensing applications.

As the first hyperspectral satellite in China's commercial aerospace field to adopt an interferometric spectral imaging system, this satellite breaks through the traditional "sky-sensing and ground-computing" data processing mode, directly completing the acquisition, processing, and decision-making of remote sensing data in orbit. It takes the lead in achieving a historic leap in the field of hyperspectral remote sensing, possessing the dual breakthrough capability of "quantitative remote sensing + real-time response" for the first time in China.

Going forward, the satellite will primarily serve mineral resource exploration in Yunnan and globally, ecological monitoring of mining areas, and the entire lifecycle management of minerals, injecting space-based momentum into the green transformation and digital governance of the mining industry. The successful launch of the Caiyun Hyperspectral 01 satellite marks a crucial step forward in my country's fully independent and controllable quantitative hyperspectral intelligent computing system, and also establishes a significant milestone for my country.Commercial SpaceflightA new benchmark in the field of remote sensing with higher technical indicators, more representative applications, and more intelligent processing.

Figure 1.The Xiguang-2 01 satellite (Caiyun Hyperspectral 01 satellite) was launched into space aboard the Gravity-1 Y4 carrier rocket.

With the successful launch and orbit insertion of the Caiyun Hyperspectral 01 satellite, the total number of satellites in orbit of CAS Xiguang Aerospace has exceeded 12. 

According to Hu Xin, the technical head of the Caiyun Hyperspectral 01 satellite,The satellite carries a hyperspectral optical payload, a panchromatic optical payload, and an on-orbit intelligent processing payload, overcoming the pain points of traditional remote sensing "sky-sensing, ground-based computing" such as time lag and insufficient recognition accuracy, and achieving three technological breakthroughs at the hardware level.This laid the foundation for the "Xiguang Series" quantitative hyperspectral imaging.Smart AstrologyStandalone general technical standards.

Figure 2.Researchers are debugging the Xiguang-2 01 satellite (Caiyun Hyperspectral 01 satellite).

This satellite employs an interferometric spectral imaging system and carries a novel static interferometric spectral payload, eliminating the moving parts of traditional spectrometers and significantly improving on-orbit stability and reliability. Based on advanced...Fourier transform interferometryThe satellite's technology enables it to acquire continuous multi-band information in a single operation, achieving high-precision and high-efficiency hyperspectral data acquisition. Its spectral performance remains stable over the long term, requiring no frequent on-orbit calibration, and it can continuously output high-quality spatial and spectral data. Based on this novel system, the satellite achieves a 10m hyperspectral spatial resolution in the visible-near-infrared band, representing a significant breakthrough in China's commercial aerospace hyperspectral remote sensing field and providing unprecedented data accuracy for quantitative remote sensing applications.

The satellite is equipped with the Zhijia NX4 onboard intelligent computing unit, jointly developed with the Zhijiang Laboratory, with a computing power of 248 TOPS. It is equipped with an ultra-high-speed data interface, which can complete the preprocessing of remote sensing data, target identification and information extraction in real time on orbit, realizing data acquisition, processing and decision-making in space. This will push remote sensing monitoring from passive post-event review to a new stage of real-time proactive early warning, greatly shortening the emergency response time and significantly reducing the pressure of satellite-to-ground data transmission.

Operating in a 535km sun-synchronous orbit, the satellite can conduct high-frequency, routine monitoring of key areas. No longer limited to "single-domain applications," it can independently support routine observation needs in multiple fields, including mining area ecological monitoring, ecological protection, urban governance, and agricultural and forestry monitoring.

If the "intelligent computing constellation" is the hard foundation for efficiently acquiring data, then the self-developed "hyperspectral remote sensing AI model" is the soft engine for activating the value of data, solving the pain points of low utilization rate of massive hyperspectral data in the industry, reliance on human experience for interpretation, and difficulty in large-scale implementation.

CAS Xi'an Optics & Aerospace, in collaboration with Zhejiang Laboratory, has developed a hyperspectral remote sensing AI model. Relying on its own high-quality, massive hyperspectral data, the model deeply integrates spectroscopy, artificial intelligence, and industry knowledge to train a dedicated hyperspectral AI model. In fields such as precision agriculture and environmental protection, this model enables a leap from "observing growth" to "analyzing nutrition," from "detecting pollution" to "identifying types," and from "wide-area search" to "targeted positioning." This makes spectral "fingerprints" more accurate, broader, and easier to use, and establishes a front-end link between "on-board acquisition and on-orbit interpretation," allowing satellites to autonomously output standardized analysis results.

Figure 3. CAS Xiguang Aerospace and its customers pose for a group photo at the launch site to celebrate the successful launch of Xiguang-2 01 satellite (Caiyun Hyperspectral 01 satellite).

Water monitoring is a typical application scenario. According to reports, the Caiyun Hyperspectral 01 satellite carries a hyperspectral water and sediment classification model jointly developed by both parties. Employing a lightweight on-orbit inference scheme, it can directly determine water turbidity levels and mark potentially polluted areas from space, transmitting only the classification conclusions and spectral data for key areas. Ground-based systems then conduct further refined analysis. This "on-orbit preprocessing + ground-based fine processing" satellite-ground collaborative mode is suitable for scenarios such as river management, flood control scheduling, and watershed ecological protection. The entire model supports on-orbit iterative updates to continuously optimize recognition accuracy.

Against the backdrop of emphasizing the value of data during the 15th Five-Year Plan period, the use of massive, high-precision spectral data to train large industry models and achieve the leap from "seeing" to "understanding" is becoming a core amplifier for aerospace data to transform from "resources" to "assets."

While commercial remote sensing is rapidly becoming widespread globally, traditional hyperspectral satellites generally suffer from drawbacks such as limited onboard storage, high satellite-to-ground transmission costs, and long data processing cycles, making it difficult to meet the rapid response needs of scenarios such as emergency monitoring, precision agriculture, and ecological environmental protection.

Xi'an Institute of Optics and Precision Mechanics CAS Aerospace Science and Technology Group Co.,LTD has precisely broken through the industry's bottleneck with a dual-track approach of "hardware and software": using "intelligent computing constellations" as hardware support and "hyperspectral AI models" as software engines, achieving synergy and mutual empowerment to connect the entire chain of aerospace data from collection to application. This approach not only aligns with the development requirements of new-type productivity but also precisely addresses the core contradiction in the industry: "massive data, scarce applications."

Figure 4. Xiguang-2 01 satellite (Caiyun Hyperspectral 01 satellite) was launched into space aboard the Gravity-1 Y4 carrier rocket.

Qin Jing, Chairman of CAS Xiguang Aerospace, stated that looking towards the "15th Five-Year Plan," CAS Xiguang Aerospace has identified two main development lines: the "hardware line" involves continuously iterating on-board intelligent hardware to strengthen space-based real-time computing capabilities; the "software line" involves iterating industry-specific hyperspectral large-scale models to deepen the application of remote sensing data in the industry. CAS Xiguang Aerospace will work with industry partners to continuously integrate the "communication, remote sensing, and computing" industrial chain, constructing a fully independent space-based intelligent computing system that integrates "space sensing, data collection, computing, and application," empowering digital monitoring and refined governance across various industries with domestically developed quantitative hyperspectral intelligent computing capabilities.