Development of China’s spaceborne SAR satellite, processing strategy, and application: take Gaofen-3 series as an example

ABSTRACT

After mastering the key technologies of manufacturing spaceborne Synthetic Aperture Radar (SAR), China’s SAR satellites have been successfully launched into space. As the only civil microwave satellite listed in the “National High-resolution Earth Observation System Major Project,” the Gaofen-3 (GF-3) 01 satellite is the first C-band multi-polarization SAR satellite with a resolution of 1 m. GF-3 series satellites stand out among civil SAR satellites worldwide because of their high resolution, wide swath, high geometric radiation quality, multiple imaging modes, and long operation time. Taking GF-3 series satellites as an example, this study introduces the development of China’s civil SAR satellites, as well as their processing strategies and applications. The success of the GF-3 series satellites shows that China’s SAR remote sensing technology has stepped into a new era of high-quality and high-precision Earth observations.

KEYWORDS:

• High-resolution Synthetic Aperture Radar (SAR)
• GaoFen-3 (GF-3) satellites
• earth observation

1. Introduction

Since the 1970s, China has been conducting Earth observations using Synthetic Aperture Radar (SAR) satellites. In the late 1970s, China’s first airborne SAR system developed by the Institute of Electronics, Chinese Academy of Sciences (IECAS) successfully conducted an Earth observation test in Tengchong City, Yunnan Province (Zhu, Jin, and Huang 1996), making airborne SAR a research hotspot. Compared with the mature development of domestic airborne SAR systems, the research on spaceborne SAR systems in China is scarce.

After thoroughly investigating the principal technologies of spaceborne SAR, China’s SAR remote sensing satellites were successfully launched into space. On 27 April 2006, China’s first SAR satellite, “Yaogan-1 (YG-1),” was successfully launched from the Taiyuan Satellite Launch Center (Zeng et al. 2013). This L-band radar satellite, with a weight of 2.7 t, has played a significant role in scientific experiments, resource censuses, and other fields. Subsequently, on 19 November 2012, the Huanjing-1C (HJ-1C) satellite successfully entered the scheduled orbit. The HJ-1C satellite is the first S-band sun synchronous orbit SAR satellite in China with an orbit altitude of 500 km (Chen, Li, and Wei 2006; Shi et al. 2012; Tian et al. 2014). The radar system carried by HJ-1C has two working modes: strip mode and scanning mode, with image widths of 40 km and 100 km, respectively, and a spatial resolution of 5 m (Zhao et al. 2017). The detailed parameters of HJ-1C are shown in Tables 1 and 2 (He et al. 2013; Liu et al. 2014). The constellation of HJ series satellites (HJ-1A, HJ-1B, HJ-1C) initially met the environmental and disaster monitoring needs of China.

Table 1. Main parameters of HJ-1C radar.

Item	Index
Band	S
Center frequency	3200 MHz
Polarization	VV
Incident angle	25°-47°
Weight	<200 kg
Designed life	3 yr

Table 2. Imaging performance of HJ-1C radar.

Imaging mode	Resolution (m)	Image width (km)	Range ambiguity (dB)	Azimuth ambiguity (dB)
Strip	5	40	> −18 dB	> −14 dB
Scan	25*	100*		> −16 dB

(*Resolution and image width of the scan mode are acquired with a multi-look factor of 1 in azimuth and 4 in the range direction).

On 10 August 2016, the Chinese GF-3 01 satellite, developed by the Institute of Space Technology of the China Aerospace Science and Technology Corporation, was successfully launched from the Taiyuan Satellite Launch Center (Yang, Wang, and Ren 2017). The GF-3 01 satellite is the first civil microwave remote sensing imaging satellite listed in the “National High Resolution Earth Observation System Major Project” and is also the first C-band multi-polarization SAR with a resolution of 1 m (Zhang and Liu 2017). GF-3 features a high resolution, wide swath, high geometric radiation quality, multiple imaging modes, and long operation time (Wang et al. 2017). The detailed imaging mode and parameters of the GF-3 01 satellite are presented in Figure 1 and Table 3. Currently, it is the SAR satellite with the maximum number of imaging modes in the world (Lin, Ye, and Yuan 2017), and can be used to monitor marine and land information globally under all light and weather conditions. Through the left – right attitude maneuver, its observation range is expanded, and the rapid response capability is improved, which further expands its application scenario. The C-band multi-polarization microwave remote sensing information obtained by the GF-3 01 satellite can be used in many fields, such as oceans, disaster reduction, water conservancy, and meteorology, and is considered an important technical support for China’s marine development, land environmental resources monitoring, and disaster prevention and reduction (Fan et al. 2017b; Sun and Shi 2017).

Figure 1. Imaging mode of GF-3 01 satellite (Zhang and Liu 2017).

Table 3. Imaging performance of GF-3 01 satellite.

Imaging mode		Incident angle (°)	Resolution (m)	Image width (km)	Polarization
Spotlight		20-50	1	10undefined	Single
Ultra-fine strip		20-50	3	30	Single
Fine strip I		19-50	5	50	Dual
Fine strip II		19-50	10	100	Dual
Standard strip		17-50	25	130	Dual
Narrow scan		17-50	50	300	Dual
Wide scan		17-50	100	500	Dual
Global observation		17-53	500	650	Dual
Quad-polarization strip I		20-41	8	30	Quad
Quad-polarization strip II		20-38	25	40	Quad
Extended	Wave	20-41	10	5undefined	Quad
	Low incident angle	10-20	25	130	Dual
	High incident angle	50-60	25	80	Dual

Over the past five years, the GF-3 01 satellite has provided data supporting emergency response and disaster relief more than 600 times with over 1300 SAR images. Its operation has saved China at least 44 billion yuan in economic losses (Zhao et al. 2021). Based on the success of the GF-3 01 satellite, China developed the GF-3 02 satellite, which was successfully launched from the Jiuquan Satellite Launch Center on 23 November 2021. Although its performance is similar to that of the GF-3 01 satellite, it is more developed in many aspects (Liu et al. 2021). It is networked with the orbiting GF-3 01 satellite to form a constellation of sea and land radar satellites, further enhancing China’s satellite observation capabilities. The GF-3 02 satellite further meets the demand for global ocean and land observation and applications in the fields of ocean, disaster mitigation, land, geology, environmental protection, water conservancy, agriculture, and meteorology, providing users with timely, reliable, and stable high-resolution, wide-swath, multi-polarization, and quantified image products.

Presently, there are more than 10 spaceborne SAR satellites in China, covering the L-, S-, C-, and X-bands, with multiple polarization and imaging modes and sub-meter resolution. Meanwhile, distributed L-band SAR constellation and dual-antenna L- and X-band SAR satellites are also under development (Li et al. 2019b). China is entering a period of rapid development of SAR satellites (Zhang et al. 2021).

Taking GF-3 series satellites as an example, this article introduces the development of China’s civil high-resolution SAR satellite along with its payload technology, data processing strategy, and application.

2. Development of satellite and payload technology

2.1. Satellite design

The GF-3 01 satellite is the first C-band multi-polarization SAR satellite independently developed by China (Li, Shao, and Zhang 2020). At the initial stage of satellite design, based on the experience of YG-1 and HJ-1C, researchers analyze the impact of error sources in various links such as SAR payload, satellite platform, observation environment, and ground processing on geometric and radiometric quality, and the influence law of various errors on imaging quality is identified. Subsequently, the errors are analyzed and characterized, and an integrated geometric and radiometric simulation platform of high-resolution SAR satellites is developed, which supports the optimal design of China’s SAR satellite, and lays a solid theoretical foundation for the design and high-precision manufacturing of SAR satellites. On the basis of defining the system and basic configuration of the SAR payload, researchers analyzed the adaptability of the satellite platform and the match between the payload and platform, forming a series of satellite characteristics and technical innovations, and the main technical indicators matched or exceeded similar satellites worldwide (Zhang 2017).

Researchers have proposed a design method for a SAR satellite system based on hybrid multiplexing of space, time, and channel, which breaks through key technologies such as mission-oriented integration system, prolonged satellite life span, incident angle and attitude control based on the optimal path, and highly integrated smart and configurable phased array antenna (Sun, Yu, and Deng 2017). The system achieves the highest resolution of 0.5 m and the satellite load ratio reaches 54%. A design procedure of orbit, long life span, and high-performance SAR satellite system is also proposed, which further derives technical breakthroughs such as multi-source-data-based component life span prediction, chip consistency control method based on PM graphics and wafer-level test data, and satellite full-time intelligent monitoring and management. The annual attenuation rate of the battery is reduced to 0.05%, which can support up to 100 min of continuous imaging, thereby improving the efficiency of SAR satellites in China (Zhang 2017).

2.2. Satellite platform

The GF-3 01 satellite is a three-axis stable Earth observation satellite with a launch weight of 2779 kg and a designed life of eight years, which is more than twice of HJ-1C (Zhang 2017). The satellite operates in a solar synchronous orbit at an altitude of 755 km and adopts a side-view imaging approach. GF-3 01 is comprises a payload and service system. The payload mainly includes the SAR payload, data transmission, and antenna subsystem; the service system is mainly composed of power supply, overall circuit, control, propulsion, data tube, structural and thermal control subsystem that provides installation, power supply, pointing, temperature maintenance, measurement and control, and other supporting services for the payload (Sun, Yu, and Deng 2017; Han et al. 2018; Zhang et al. 2018).

GF-3 01 satellite’s platform buildup allows a pointing accuracy better than 0.03° and stability better than 5 × 10⁻⁴ °/s, with ±31.5° side-swing capacity. The satellite adopts a dual independent bus power supply and is equipped with a triple-junction gallium arsenide solar cell array, a 100 Ah cadmium nickel battery, and a 225 Ah lithium-ion battery, which can meet the imaging demand of approximately 10,000 watts of short-term power consumption (Zhang 2017). Additionally, when the output power of the platform bus is insufficient, the high-voltage power supply can be converted into low-voltage through the grid-connected controller for onboard equipment. The satellite also has an independent health management system that can realize continuous monitoring of key events, evaluate the product status, and preemptively take a variety of effective measures to ensure the safe operation of the satellite (Han et al. 2018).

2.3. Satellite payload

Compared with earlier domestic SAR satellites, GF-3 01 satellite’s payload system has the advantages of multi-polarization, multiple working modes, high resolution, large image width and antenna size, low power consumption, high radiometric resolution (the ability of the sensor to differentiate among subtle variations in brightness, i.e. the energy measured by the instrument), and long imaging time, as well as internal and external calibration functions. A total of 12 imaging modes, such as spotlight, strip, scan, etc., are integrated into the system, making it the in-orbit civil SAR satellite with the most operation modes in the world; it has a maximum continuous working duration of 50 min. It can obtain C-band multi-polarization SAR images with a resolution of 1–500 m and an imaging width of 10–650 km. Its radiometric resolution is better than 2 dB, and the radiation accuracy can reach 1 dB, which is significantly improved compared to that of HJ-1C (Sun et al. 2019). The GF-3 01 satellite is equipped with a 15 m × 1.232 m four-polarization waveguide slot phased-array SAR antenna (Sun, Yu, and Deng 2017). The antenna comprises a waveguide slot antenna, four-channel T/R module, delay amplification module, wave control unit, RF transceiver and calibration feed network, secondary power supply, high- and low-frequency cable networks, active mounting plate, structural frame, deployment mechanism, and thermal control. These configurations give the GF-3 01 satellite the ability to flexibly shape, scan, and broaden a two-dimensional wave beam with high polarization isolation and power efficiency (Zhang 2017).

This satellite adopts high-precision and high-quality imaging approaches. It also features several technical innovations, such as high-precision expansion and maintenance of large phased-array antenna, intelligent follow-up temperature control of phased-array antenna, and polarization distance fuzzy suppression based on waveform coding. Its relative radiometric accuracy is better than 0.8 db, the absolute radiometric accuracy is better than 1.4 db, and the radiometric resolution is better than 3.4 db (in imaging modes with spatial resolution 1–10 m) and 1.7 db (in imaging modes with spatial resolution 25–500 m) (Sun, Yu, and Deng 2017; Zhang 2017).

In addition to the functionality of Earth observation using its SAR payload, the GF-3 02 satellite is equipped with a ship positioning module. This module enables it to locate ships at sea and identify and image certain ships, as well as obtain ocean meteorology and navigation information in a timely manner, which will improve the safety of navigation at sea (Hu 2021). This is an important development in China’s SAR satellite payloads.

2.4. Imaging technology

The observation performance of SAR satellites is mainly reflected in two aspects: spatial resolution and Earth observation width. Since the application of spaceborne SAR in remote sensing, researchers have endeavored to improve the resolution and image width to increase the information contained in SAR images (Deng et al. 2019). The application scenarios, such as land use and vegetation cover surveys, ocean monitoring, and glacier observation, all prefer SAR to have wide-swath imaging ability, so that the information captured by the sensors can be swiftly interpreted. From HJ-1C to GF-3, the spatial resolution of China’s civil SAR satellite has increased from 5 m to 1 m, and the image width has increased from 100 km to 650 km, which significantly improves satellite performance.

As the first C-band multi-polarization civil SAR imaging satellite in China, the GF-3 satellite has attracted extensive attention at home and abroad because of its high-quality imaging results. In the sliding spotlight mode, the azimuthal imaging time can be increased by maneuvering the speed of the radar beam moving on the ground; thus, azimuthal width and resolution are improved. Consequently, images captured in the sliding spotlight mode have the highest resolution (Belcher and Baker 1996; Zhang et al. 2018). The spotlight and strip modes can be considered special cases of the sliding spotlight mode. When the radar beam travels with zero velocity on the ground, it is in spotlight mode. Although its azimuth resolution is high, this mode can only image small areas (Lanari et al. 2001; Zamparelli, Agram, and Fornaro 2014). When the moving speed of the radar beam on the ground is equal to the flight speed, it is in strip mode. In this mode, the sensor is capable of continuous imaging over a large area, yet its azimuthal resolution is limited owing to the size of the antenna (Han et al. 2018).

At present, GF-3 SAR images are mainly processed using a classical filtering algorithm. To further improve the image quality, a sparse SAR imaging method based on complex image data was introduced and implemented on existing GF-3 SAR data. Compared with the existing SAR data, the obtained imaging results have a lower sidelobe, higher signal-to-noise ratio, and better resolution (Bi et al. 2020).

Extended imaging time is an important development of SAR satellites in China. Compared with the GF-3 01 satellite, the longer imaging time of GF-3 02 is of immense significance for marine applications. A general remote sensing satellite can only obtain signals from the ocean surface, whereas the GF-3 can directly observe internal changes in the ocean (Yang, Wang, and Ren 2017; Li et al. 2018). For each satellite imaging operation, a longer imaging time enables the GF-3 02 satellite to obtain twice as much information as the GF-3 01 satellite, which is especially conducive for monitoring ocean internal waves to provide early warnings for ships and submarines (Hu 2021).

Improvement of the imaging mechanism is another important development of China’s SAR satellite, which has a significant impact on the clarity of satellite images. Compared with the GF-3 01 satellite, the biggest development in the imaging mechanism of GF-3 02 is the significant reduction of the “scallop effect” (Li et al. 2019c). The image of the GF-3 02 satellite is more stereoscopic and clearer with finer detail, and the image range is wider, which is of immense significance for applications such as natural disaster monitoring.

2.5. Revisiting time

From HJ-1C to GF-3 01, the revisit time is shortened from four days to less than three days. Additionally, the dual side-view imaging strategy allows the GF-3 01 satellite to revisit identical area in less than 1.5 days (Zhang 2017). With the successful launch of the GF-3 02 satellite, an even shorter revisit time can be realized, as a constellation has been formed, implying more and faster data acquisition under the same conditions. Taking China’s Antarctic scientific research as an example, the GF-3 constellation provides high-resolution images along the route of the “Xuelong” research vessel for sea ice density and thickness analysis and the safe navigation of the “Xuelong” (Zeng et al. 2017). However, the GF-3 01 satellite can only monitor the “Xuelong” navigation once or twice daily, whereas the GF-3 constellation can achieve more than three times a day’s imaging frequency, which can better guarantee the accuracy and safety of the “Xuelong” on its way to the south pole (Hu 2021).

3. Development of SAR data processing

3.1. Data geometric calibration method

With the successive launch of SAR satellites in China, high-resolution SAR data worldwide can be quickly obtained through satellite networks. While the resolution and radiation quality of domestic spaceborne SAR images are constantly improving, ground objects can now be clearly seen from space (Deng et al. 2019). However, SAR images have not been effectively utilized, resulting in serious data wastage. Indicators such as resolution and radiation quality are still far from meeting the needs of remote sensing applications, thus improving the geometric quality of domestic SAR images has become an important issue (Yin and Yang 2018). Owing to the lack of fundamental research and the limitations of satellite hardware, a large gap exists between domestic and foreign SAR satellite image positioning accuracies. The designed geometric positioning accuracy of most SAR satellites in China is approximately hundreds of meters. After geometric calibration, the geometric positioning accuracy of certain SAR satellites, without using ground control points, can reach up to 100 m (Liu et al. 2014). To address these problems, researchers have improved the geometric quality of domestic SAR satellites through software advancement to further broaden their application scenarios.

Geometric positioning accuracy is an important technical indicator of SAR satellites and a key factor affecting the processing and application of SAR images. The error sources affecting the geometric positioning accuracy of spaceborne SAR mainly include: orbit error, SAR system time error, atmospheric propagation effect, Doppler center frequency estimation, and terrain height error (Oliveira, Paradella, and Silva 2011). Presently, the geometric positioning accuracy of advanced spaceborne SAR systems has significantly improved worldwide. The plane positioning accuracy of the European Remote Sensing Satellite (ERS) reaches 20 m (Shimada et al. 2009), that of the COSMO-SkyMed satellite of Italy can reach 15 m (Zhang et al. 2012), while the geometric positioning accuracy of the ALOS of Japan (Shimada 2010), Radarsat-2 of Canada (Zhou et al. 2013), TerraSAR-X of Germany (Raggam et al. 2010), and Sentinel-1 of the European Space Agency (Schmidt et al. 2018) are all better than 10 m.

As for the parameters used for calibration, Zhou (2013) used 22 corner reflectors to study the geometric calibration of Radarsat-2 and TerraSAR-X data in Inner Mongolia, northeast China. Based on research by foreign scholars, three geometric calibration parameters for spaceborne SAR were proposed: initial range distance (range time), azimuth time, and range pixel interval. However, the geometric calibration model and calculation method for spaceborne SAR were not considered. In 2011, Jiang and Zhang (2011) proposed a geometric calibration model and calculation method for spaceborne SAR with four geometric calibration parameters: range-direction time, azimuth-direction time, pulse repetition frequency, and sampling frequency, which laid a theoretical foundation for relevant studies on the geometric calibration of spaceborne SAR in China.

To improve the data quality of China’s first SAR satellite “YG-1,” Zhang, Fei, and Li (2010) selected the range initial time delay measurement error, range sampling rate error, direction initial time error, and azimuth pulse repetition frequency error as calibration parameters; thereafter, through calibration processing, the data positioning accuracy was improved. HJ-1C, the first civil spaceborne S-band SAR in China, initially had a positioning accuracy of 100 m, but with the efforts of researchers, the accuracy has been improved by several meters (Liu et al. 2014). Subsequently, with the advancement of technology, the initial positioning accuracy of GF-3 has significantly improved. To verify the geometric accuracy achieved by the different modes of GF-3 images, Wang (2017) analyzed the SAR geometric error source and performed geometric correction tests based on the Rational Polynomial Coefficient (RPC) model with and without Ground Control Points (GCPs) for five imaging modes: the SL, UFS, FSI, QPSI, and SS modes. The experimental results show that the checkpoint residuals are large and consistent without GCPs, but the root mean square error of the independent checkpoints for the case of four corner control points is better than 1.5 pixels, achieving a similar level of geometric positioning accuracy to that of international satellites, and GF-3 satellite can be used for high-accuracy geometric processing and related industry applications.

To achieve better geometric calibration accuracy, based on the positioning consistency constraint of the conjugate points, Deng et al. (2017) presented a geometric cross-calibration method for the rapid and accurate calibration of GF-3, which could accurately calibrate the geolocation parameters without using any high-accuracy control data of the calibration field. The results showed that this method can achieve a calibration accuracy as high as that of the conventional field calibration method. For a certain level of systematic error after geometric calibration, Zhao (2017) classified the errors that affect geometric positioning accuracy into three categories: fixed system error, time-varying system error, and random error. The system error was effectively compensated by using ground control points and global atmospheric parameters; thus, the geometric positioning accuracy of GF-3 was improved to the pixel level.

Researchers have successively proposed geometric calibration methods for SAR satellites considering atmospheric propagation delay, forming a complete system that is applied to GF-3, and its accuracy is better than 3 m without ground control points, which meets advanced world standards (Deng et al. 2019). Additionally, to address the problems of high deployment cost and low efficiency of artificial corner reflectors, Zhao (2017) developed high-precision automatic corner reflectors that can be remotely controlled. These reflectors were deployed in the China (Songshan) satellite remote sensing calibration field – the first normalized calibration field of SAR satellites in China.

3.2. SAR interferometry

The 13th Five-Year Development Plan of National Satellite Surveying and Mapping considers the advancement of interferometric radar satellites as a key task in constructing the Earth observation system. China has accelerated the application research of SAR satellites to realize interferometry capability and image acquisition under cloudy and rainy weather and supports the generation of a global Digital Elevation Model (DEM) and regional surface deformation monitoring (Zhang 2017; Marbouti et al. 2021). Currently, among China’s SAR satellites in orbit, only Tianhui-2 (TH-2) is designed and manufactured for interferometry applications. In summary, China’s civil satellites for InSAR measurements are still in the initial stages of development. However, a few in-orbit SAR satellites still exhibit interferometry capability, particularly GF-3 (Zheng, Chen, and Zhang 2020). The original purpose of GF-3 is not SAR interferometry, but analysis of its data proves that it has the potential for interferometric imaging. In 2016, the first interferometric image and DEM results were released at the third advanced symposium on imaging radar for Earth observation, which marked the successful start of interferometry of China’s civil SAR satellites (Ma et al. 2018).

For repeated-pass interferometry, given the limitations of satellite platform in-orbit control, China’s SAR satellites are still unable to stably obtain revisit image pairs that meet interferometry standards, and this poses a challenge to images’ interferometric processing. Traditional InSAR processing methods, including registration, baseline estimation, phase filtering, and unwrapping, are not applicable to China’s SAR satellites. Orbit drift, phase noise, unsatisfactory baseline, and low coherence of domestic SAR satellites cannot fully satisfy the requirements for stable and reliable interferometry (Sun et al. 2019). Additionally, with the continuous development of SAR satellites, improvement of resolution, and diversification of working modes, InSAR data processing strategies will continue to encounter new problems and challenges, with substantial room for improvement.

In view of the various problems faced by China’s SAR satellites interferometry, Chen, Zhang, and Zhang (2016) deduced a direct relationship between registration and satellite state parameters and observed target characteristics, proposing a registration method considering large deformations and serious decorrelation, with a registration accuracy better than 1% pixels. Based on the mathematical modeling of InSAR phase noise, Chen, Zhang, and Zhang (2016) also proposed a wavelet domain multi-scale InSAR phase noise filtering method suitable for domestic SAR satellite interferometry. It has better noise filtering and phase edge and detail retention abilities, which supports the realization of domestic SAR satellite interferometry. Figure 2 shows a demonstration of interferogram by GF-3 of Dengfeng, China.

Figure 2. A demonstration of interferogram by GF-3 of Dengfeng, China.

3.3. Polarimetric SAR

Polarimetric SAR can obtain the scattering matrix of each pixel and synthesize a variety of polarization scattering information, including linear, circular, and elliptical polarizations. Therefore, compared with conventional SAR, it has considerable advantages in target detection, recognition, texture feature extraction, etc. (Wu, Hong, and Wang 2007). Polarimetric SAR is sensitive to the shape and direction of vegetation scatterers. By measuring the polarization scattering matrix of each pixel, it is possible to decompose the complex ground object scattering process into several single scattering processes, and use the polarization scattering information of ground objects in different polarization states to detect the target characteristics more accurately. Polarization data have enormous application value and potential for quantitative remote sensing, and have, therefore, become a research hotspot (Greatbatch 2012).

The feature information of the ground object is mainly reflected in the amplitude, phase, frequency, and polarization response of the SAR image. Amplitude, phase, and frequency response information have been successfully used in applications. Owing to the limitations of technology and theoretical development, full polarization information extraction has not been achieved in the last two decades (Kim and Zyl 2000). Polarimetric SAR satellites transmit and receive electromagnetic waves with different polarization modes through multiple channels to obtain more comprehensive electromagnetic wave-scattering characteristics of ground objects. It is widely used in crop classification and yield estimation, forest investigation, biomass estimation, oceans, geology, hydrology, resources, the environment, disaster monitoring, and other fields (Yang et al. 2015).

With the development of technology, China’s spaceborne SAR satellites’ polarization capability has also developed from single polarization (HJ-1C) to multi-polarization (GF-3). Through multiple polarization channels, GF-3 can obtain the electromagnetic scattering characteristics of the target under different polarization modes, as well as its polarimetric scattering matrix, thus capturing more comprehensive scattering features of the target (Jin, Qiu, and Huang 2019). The polarization scattering matrix contains more information, which can identify more comprehensive physical features of the target, such as direction, shape, roughness, and dielectric constant and provide more useful information for large-area ground object classification, target detection, and recognition (Wu, Hong, and Wang 2007).

3.4. Onboard processing

Onboard processing technology has also been a breakthrough in the development of China’s SAR satellite technology. Compared with the GF-3 01 satellite, GF-3 02 is equipped with an onboard real-time processor, which allows instant data processing in specific observation modes. Previously, after GF-3 01 finished imaging a certain area, it still took several hours of onboard processing before the user could access the images (Han et al. 2018). For the GF-3 02 satellite, this time has been shortened to within 20 min, which is a significant progress in terms of emergency rescue applications (Hu 2021).

4. Development of SAR applications

In China, although a series of national plans and programs have supported the application research of spaceborne SAR in agriculture, forestry, ocean and water resources, minerals, and other aspects, it started late and the overall application is not very sophisticated (Sun, Tang, and Zhai 2009; Shao, Wu, and Li 2021). Therefore, keeping in step with the edging SAR technology, developing the application research of China’s self-developed spaceborne SAR satellites is an arduous task for microwave remote sensing scientists and technicians (Li, Wang, and Jiang 2021).

The main task of the GF-3 project is to realize all-day all-weather ocean and land observation through its high-resolution and multi-polarization SAR data and improve the efficiency of ocean monitoring and disaster management, as well as other aspects including agriculture, forestry, land, environmental protection, public and national security, housing and construction, transportation, statistics, and other industries (Zhang 2017). GF-3 also improves the rapid response ability to emergencies, and plays an important demonstration role in leading the application of civil high-resolution microwave remote sensing satellites in China. Next, this article briefly summarizes the development of the main applications of GF-3 satellites.

4.1. Marine applications

China’s SAR satellites have continuously monitored global marine resources and provided high-quality and stable observation data, which offered vital data support for marine disaster prevention and mitigation, marine dynamic environment monitoring, and other aspects. GF-3 is especially popular among oceanographers because of its diverse working modes and polarization features.

Shao (2021) collected more than 10,000 VV and HH-polarized GF-3 images acquired in quad-polarization strip and wave modes during 2017–2021, and observed wind patterns over open seas with incidence angles ranging from 18° to 52°. Previous studies have demonstrated GF-3‘s capability of wave height and typhoon inversion (Zhou et al. 2018; Zhu et al. 2020; Xia, Yokoya, and Pham 2020). Similar studies using GF-3 have also been conducted by other researchers (Zhang et al. 2019; Chen et al. 2020; Yang et al. 2021; Sheng et al. 2018). Using different approaches, they all have demonstrated the sea surface parameter retrieval ability of GF-3, as shown in figure 3.

Figure 3. Wind speed and wave height derived from GF-3 data (Sheng et al. 2018).

Recently, target recognition and classification based on SAR images have rapidly developed – they are considered a powerful supplement to traditional optical-image-based methods. Since the launch of GF-3, substantial data have been archived, providing significant data support for target recognition and classification. The data have been widely used in the field of marine target recognition and classification, as shown in figure 4, including:

1. Ship detection (Liu et al. 2017; Wang et al. 2017; Zhang et al. 2017; An, Pan, and You 2018; Ma et al. 2019; Wang et al. 2019; Zhang et al. 2021a)

2. Dynamic monitoring of the sea (Fan et al. 2017a; An et al. 2018; Fang et al. 2021; Wan et al. 2021; Shao et al. 2022)

3. Water body identification (Wang et al. 2018; Gu et al. 2019; Li et al. 2019a; Qin et al. 2019; Xu et al. 2021; Zhang et al. 2021b)

4. Sea-ice monitoring (Zhang et al. 2017; Han et al. 2019; Zhang et al. 2021c)

Figure 4. Ship and floating raft detection results (Fan et al. 2017a; Liu et al. 2017).

With the continuous advancement of deep learning, target detection based on deep learning has become mainstream in this field and is increasingly applied for target recognition and classification in SAR images (Li et al. 2017; Wang et al. 2017; Wang et al. 2017a; An, Pan, and You 2018; Ma et al. 2019; Wang et al. 2019; Zhang et al. 2021; Zhang et al. 2021c). Traditional target recognition and classification algorithms are mostly designed for specific scenes, and entail cumbersome image preprocessing, segmentation, etc. The deep-learning target detection method can process SAR data in complex scenes and realize end-to-end target detection. Compared with traditional approaches, it not only effectively reduces the impact of the external environment but also significantly improves the detection efficiency and accuracy.

4.2. Agriculture and forestry

China needs to feed a quarter of the world’s population and deal with their carbon emissions. Therefore, agriculture and forestry are two essential aspects (Torre, Gao, and Macinnis-Ng 2021; Zhou and Ismaeel 2021). GF-3 contributes to these fields through its polarized images with large coverage. Han et al. (2019) presented the first result of GF-3 quantitative application in crop monitoring using a new polarized water cloud model to improve the accuracy of biomass extraction without soil moisture data. Zhu et al. (2020) and Xia, Yokoya, and Pham (2020) extracted mangrove biomass using GF-3 data; the latter study also demonstrated that full-polarimetric GF-3 data are superior to dual-polarimetric Sentinel-1 data for mapping mangrove tree species in the tropics. Zhou et al. (2018) accurately classified forest species in northeast China using GF-3 and ALOS-1 PALSAR data, as shown in figure 5.

Figure 5. Distribution of mangrove biomass (Zhu et al. 2020) and forest classification results using GF-3 (Zhou et al. 2018).

4.3. Topography and mapping

Although the GF-3 satellite was not initially designed for interferometry, its interferometric imaging potential has been confirmed. Ma et al. (2018) proved that GF-3 repeat-pass data possess certain advantages such as high imaging quality, stable spatial baseline, and high temporal and spatial coherence, which are conducive to interferometry applications. GF-3‘s interferometry capability is further validated by Sun et al. (2019) and Zheng, Chen, and Zhang (2020); the latter study also established the superiority of GF-3 over YG-29 in DEM generation, as shown in figure 6.

Figure 6. DEM of Dengfeng City of China derived from YG-29 (a) and GF-3 (b) (Zheng, Chen, and Zhang 2020).

Regarding regional SAR processing, to eliminate the multiplicative speckle noise in SAR images, a Euclidean distance-weighted anisotropic diffusion speckle suppression method is proposed (Wang et al. 2017), which not only ensures the optimal edge texture but also improves the equivalent number of looks. To satisfy the requirements of high geometric radiation quality in the production of regional SAR orthophoto images, researchers have proposed a block adjustment method for SAR images to ensure the relative and absolute positioning accuracy of the adjusted images. Aiming at the radiation difference between adjacent orthophotos of spaceborne SAR images, researchers have proposed a color correction method based on random cross-observation. All these methods have been applied to the generation of the GF-3 SAR orthophoto map of China (Wang et al. 2022), as shown in figure 7. The absolute geometric accuracy of the map was better than 8 m. After mosaicing, the relative radiation intensity distribution was consistent with the scattering features of actual ground objects, and the color transition along the edge was smooth, which solved the problem of geometric and radiation consistency in orthophoto processing without control points.

Figure 7. GF-3 SAR orthophoto map of China.

4.4. Disaster alleviation

Currently, disaster prevention and reduction using remote sensing approaches mainly include disaster element monitoring, disaster risk assessment, disaster emergency monitoring, and post-disaster recovery (Shan et al. 2020; Trinder and Liu 2020). Satellite remote sensing technology has the advantages of large coverage, high resolution, and repeated observations (Zhong et al. 2021). In the case of natural disasters such as droughts, floods, typhoons, earthquakes, landslides, debris flows, and forest and grassland fires, GF-3 has played an important role in terms of data support for monitoring. This has been possible by virtue of its imaging advantages such as multi-mode, multi-polarization, and wide-swath.

Kang et al. (2018) realized fast flood detection using GF-3 data and a fully convolutional network. Ding et al. (2019) extracted landslides in mountainous areas using GF-3 polarized data. In July 2018, floods caused by a dam collapse hit southern Laos. Pre- and post-event images from the GF-1, 2, and 3 satellites were used to evaluate the situation and provide data support for China’s emergency material assistance to Laos. GF-3 data have also been successfully applied for deformation extraction, as shown in figure 8. Chen et al. (2020) published the first application of GF-3 to monitor co-seismic deformation of an earthquake, the results of which have a high Pearson’s correlation coefficient with those derived from radarsat-2, Sentinel-1A, and Alos-2 PALSAR-2 data. Wang et al. (2020) used GF-3 data to detect the surface motion parameters of the Yiga Glacier based on offset tracking technology. During the July 2020 flooding in Poyang Lake, the GF-3 01 satellite continuously monitored the flooded areas, providing crucial data to frontline disaster relief departments.

Figure 8. Deformation phase extracted from GF-3 (a) and Sentinel-1 (b).

5. Conclusion and outlook

From YG-1 to HJ-1C and then to the GF-3 series, the development of spaceborne SAR satellites in China has come a long way, and is currently among the best in the world. In particular, the successful launch and operation of the GF-3 constellation has changed the situation that spaceborne high-resolution SAR images rely on imports, and provided high-quality, high-precision Earth observation data for users in various sectors worldwide. Taking GF-3 as an example, this study systematically overviews the development of China’s civil SAR satellite payload technology, data processing, and applications, and demonstrates the progress made by China in recent years in radar remote sensing technology. GF-3 satellites have provided data support for China’s marine development, land environmental resources monitoring, disaster prevention and reduction, etc., which will further promote social and economic development. It has also laid a solid foundation for the development of China’s spaceborne SAR satellite.

In the future, China will launch a series of new SAR satellites, such as GF-3 03 interferometric satellites and dual antenna L-band and X-band SAR satellites. Among them, LuTan-1, an L-band SAR system containing two satellites, has the primary objective of acquiring the DEM and surface deformation. It has many other promising uses such as single-channel polarization InSAR for forestry applications and mixed-polarization InSAR for land classification. LT-1 will be a powerful supplement to the existing SAR network to obtain multi-dimensional information of geodynamic processes and monitor the global environment.

In the future, SAR satellites will continue to play an irreplaceable role in Earth observation, and China will continue to explore and innovate in the following aspects:

1. Improved satellite payload technology. After approximately 30 years of development, spaceborne SAR technology has transformed from a single imaging mode (strip) to a variety of imaging modes (strip, spotlight or sliding spotlight, scanning, and mosaicing), the polarization mode has evolved from single polarization to multi-polarization, and the resolution has been improved from 100 m to sub-meter level, which undoubtedly drives up the technical indexes of the satellite. In the future, SAR technology will continue to develop toward high-resolution, wide-range, multi-dimensional, and multi-angle, to further improve the quality of Earth observation.

2. High temporal resolution of the SAR data. With the expansion of application fields, existing spaceborne SAR systems are being challenged by diversified observation tasks, which require large-scale, high-frequency, repeated observation of certain areas while ensuring high resolution. To achieve this goal, China is currently working on the following: 1). A geosynchronous orbit SAR satellite, which works at a higher geosynchronous orbit to obtain a low revisit time and large coverage; 2). Agile small SAR satellites, which with their highly reliable and high-precision orbit maneuvers can realize the rapid revisit of hotspots; 3). Video SAR satellites, which can observe stationary targets all day under all weather conditions; and 4). Satellite constellations, which through the network of multiple satellites improve the revisit time of satellites, such as the current Sentinel- 1A and 1 B constellations.

Additionally, China will continue to explore and innovate in terms of multi-frequency and multi-baseline SAR, passive SAR, cross-orbit interferometry, differential frequency interferometry, and other novel aspects.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data of the Chinese Satellite Gaofen-3 used in this study have not been released to the public for policy reasons, so there is no data available for free.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China (NSFC) [grant number 41801397], National Key Research and Development Program of China [grant number 2018YFC0825803], the Key Research and Development Program of Hubei Province [grant number 2021BID009], the Natural Science Foundation of Liaoning Province [grant number 2020-BS-259], and the Key Laboratory of Land Satellite Remote Sensing Application, Ministry of Natural Resources of the People’s Republic of China [grant number KLSMNR-202107]

Notes on contributors

Guo Zhang

Zhenwei Chen obtained his PhD degree from Wuhan University in 2017. He is primarily engaged in research on InSAR data processing and applications. He is in charge of a project funded by the National Natural Science Foundation of China and has published six SCI papers and three EI papers.

Shunyao Wang

Guo Zhang received a PhD degree in photogrammetry and remote sensing from the School of Remote Sensing and Information Engineering, Wuhan University, in 2005. and have been working at the State Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing, Wuhan University, where he became a professor in 2011. His research interests include space photogrammetry (including processing and applications for optical, SAR, video, and laser loads).

Zhenwei Chen

Shunyao Wang is currently pursuing a PhD degree in photogrammetry and remote sensing in the State Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing, Wuhan University. His research interests include InSAR and geohazard monitoring.

Yuzhi Zheng

Yuzhi Zheng obtained a PhD degree from Wuhan University in 2020; she currently works at the Wuhan University. Her research interests include radar remote sensing, the intersection of remote sensing, and other disciplines. She has published more than 10 papers in international journals.

Ruishan Zhao

Ruishan Zhao received a PhD degree in surveying and mapping from the School of Geomatics, Liaoning Technical University, Fuxin, China in 2018 and became a lecturer. His core research interest lies in the geometric processing of spaceborne SAR imagery. He is the director of the IEEE PES Substation Satellite Committee in China.

Taoyang Wang

Taoyang Wang received a PhD degree in photogrammetry and remote sensing from the School of Remote Sensing and Information Engineering, Wuhan University, in 2012, where he became an Associate Research Fellow in 2015. His research interests include space photogrammetry, geometry processing of spaceborne optical/SAR/InSAR imagery, target detection, and recognition based on satellite video.

Yu Zhu

Yu Zhu, PhD, Engineer at Beijing Space Vehicle Overall Design Department, China Academy of Space Technology. His research interests include satellite payload design and development, telecommunication technology, aerospace science, and engineering.

Xinzhe Yuan

Xinzhe Yuan obtained a PhD degree in communication and information systems from the Graduate University of Chinese Academy of Science, Beijing, China, in 2007. He is currently Associate Research Fellow at the National Satellite Ocean Application Service, Beijing, China. His research interests include SAR systems, SAR ocean applications, and SAR sea surface current field inversion.

Wei Wu

Wei Wu, PhD, Deputy Director of the Disaster Assessment Department, National Disaster Reduction Center, Ministry of Emergency Management. His research interests include disaster monitoring via remote sensing, spatial data management and visualization, disaster risk, and loss assessment.

Weitao Chen

Weitao Chen obtained a PhD degree in environmental science and engineering from the China University of Geosciences (CUG), Wuhan, China, in 2012. He is now a professor at the School of Computer Science at CUG. He has authored and co-authored more than 30 papers. His main research interests include machine learning and the remote sensing of the environment.