Remote sensing science and technology have been widely used in various branches of Earth system science and in many other fields, including ecology, environmental science, agriculture, forestry, transportation, and urban development. Geographic remote sensing science, as defined in this article, only refers to the fundamental scientific questions that arise from the intersection of remote sensing science and technology and geographic science and from the applications of remote sensing to various branches of geographic science (Figure 1).

The rapid development of remote sensing has provided both important information sources and scientific and technological support for the study of physical geography, human geography, sustainable development, and human-nature systems, and has promoted the development of fundamental remote sensing research such as quantitative inversion, remote sensing product validation, and scale transformation. Improvements in remote sensing radiative transfer modelling and quantitative remote sensing inversion theory based on a priori knowledge have facilitated development of the theory and method of synergetic inversion based on multisource remote sensing (Li et al., 2013; Li et al., 2016; Liang and Wang, 2019). The production of quantitative remote sensing products for the global land surface and the development of a synergetic inversion system based on multisource remote sensing data has enabled China to create long-time series global remote sensing products (Liang et al., 2021). The validation of quantitative remote sensing products has made great progress in the construction of relevant theories and methods. In particular, methods have been developed for acquiring ground truth at the remote sensing pixel scale, and the in situ networks have been constructed for validating remote sensing products. ‘Bottom-up inductive’ and ‘top-down deductive’ methods have been proposed for solving remote sensing scale problems (Li and Wang, 2013). Theory and experimental methods for remote sensing scale transformation have been gradually established through conducting the Heihe watershed allied telemetry experimental research (HiWATER) (Li et al., 2013).

Remote sensing science has been integrated into geographic science via the gradual formation of several major application branches, including vegetation remote sensing (Liu, 2014), land use/land cover remote sensing (Chen and Chen, 2018), geomorphological remote sensing, hydrological remote sensing, cryosphere remote sensing (Li et al., 2020a), and remote sensing of humanities and socioeconomic elements (Xu et al., 2016) (Figure 2). Geographic remote sensing science has thereby driven the development of geographic science to an unprecedented depth and breadth. Remote sensing has played an important role in mapping the 1:1,000,000 Chinese vegetation map and the national forest resource map. Chinese scientists have successively released the first global land cover datasets at resolutions of 30 meters and 10 meters (Chen and Chen, 2018; Gong et al., 2019b) and developed annual maps of global artificial impervious areas between 1985 and 2018 at a 30-m resolution and other high-resolution land cover products (Gong et al., 2020). Remote sensing and geographic information systems (GIS) technologies that integrate positioning and quantitative and qualitative methods have been used to compile the Geomorphological Atlas of the People’s Republic of China (Zhou and Cheng, 2010), thereby providing accurate key data for geomorphological research. The two-phase Heihe River remote sensing experimental project was completed over a 10-year span and has realized the deep integration of remote sensing with integrated eco-hydrological study (Li et al., 2013). In addition, remote sensing has also facilitated the quantitative assessment of the United Nations Sustainable Development Goals based on long-period, multiscale, macroscopic and microscopic information from multisource remote sensing observations and the usage of big Earth data (Guo, 2020).

Geographic information systems (GIS) emerged in the 1960s, and Goodchild proposed the concept of geographic information science (GIScience) in 1992 (Goodchild, 1992). GIS was introduced to China in the late 1970s, and pioneering Chinese scientists, such as the academician Shupeng Chen, advocated for and introduced geographic information science research to China. These scientists proposed a series of concepts and methods, such as the ‘geographic information graph’, that facilitated the methodological development of geographic information science (Figure 3). From 1990 to 2000, Chinese scientists mainly conducted theoretical and methodological studies in the fields of GIS construction models, digital elevation analysis, vector-raster integrated spatiotemporal data models (Gong and Xia, 1999), object-oriented GIS (Gong, 1997), GIS-based spatiotemporal data models, and spatial statistics and analysis. A series of domestic GIS softwares with independent intellectual property rights, such as Super Map, MapGIS, and GeoStar, were developed. Since 2000, the in-depth study of geographic information science and demand for developing Digital Earth and Digital China has resulted in the increasingly intensive application of GIS, accompanied by methodological and technological innovations by Chinese scientists in the virtual geographic environment (Lü, 2011), spatial topological relations (Chen and Guo, 1998), geospatial statistics (Wang and Xu, 2017), geographic data mining, geospatial uncertainty, cellular automata and geographic simulation systems (Li et al., 2017), and geospatial prediction. Since 2010, the rapid development of smart city construction (Li et al., 2014; Gong et al., 2019a) has made GIS the fundamental support platform for urban management, land use, public health, disaster monitoring, and other applications. GIS platforms and related softwares in China have developed rapidly and independently. The market share of domestic GIS software exceeded 50% for the first time in 2015. Since 2018, outstanding progress has been made in the geographic information science in China in the fields of new information and communications technology (ICT), big geographic data (Pei et al., 2019), artificial intelligence, and ubiquitous geographic information services. A series of internationally prominent fundamental theories and key technologies have been developed, including full-space information systems (Zhou, 2015), panoramic maps (Zhou et al., 2011), geoscenography (Lü et al., 2018), and social sensing (Liu et al., 2015). China’s geographic information science efforts have led to remarkable progress in the fields of big Earth data integration (Guo, 2018) and behavioural trajectories and space. The development and application of geographic information science and technology have gone far beyond the scope of geographic science research, gradually moving from scientific research and industrial applications to public and social services.

The main research fields in geographic data science (Singleton and Arribas-Bel, 2021) are mathematical geography, big geographic data, machine learning, and artificial intelligence geography (Figure 1). Big data and artificial intelligence have driven the formation and development of geographic data science. Chinese scientists have developed mathematical geography methods that go beyond the scope of statistical methods, while incorporating machine learning, evolutionary methods, and complex network theories (Yuan et al., 2019). A variety of geographic data, such as Earth observations, sensor networks, volunteered geographic information (VGI), and human behavioural data, have rapidly converged into a series of massive geographic data resource pools. Consequently, a series of standards and specifications, data models, aggregation methods, transmission models, sharing models, and infrastructures have been established for big geographic data. This progress has enabled the seamless integration of GIS, remote sensing, global positioning systems (GPS), and spatial decision support systems (SDSS) along with improved scientific data integration and the sharing network for the Earth’s surface system (Chen et al., 2020; Li et al., 2020b). Chinese scientists have used massive Earth observation data and multisource fused data to build a large Earth data platform (Guo, 2018) that provides new methodologies for Earth science research and has provided extremely significant support for the United Nations Sustainable Development Goals (Guo, 2020) and the ‘Digital Belt and Road Initiative’ (Guo, 2018). Big data for human behaviour have been used to establish methods for analysing human spatiotemporal behaviour characteristics. A research framework for social sensing has thus been developed (Liu et al., 2015) that has triggered the development and breakthroughs of big social data. Traditional statistical methods are limited utility for exploring complex and nonlinear relationships, multivariate collaborative and spatiotemporal coupling features, and ultralarge-scale computations embedded in geographic data. Along with rapid developments in big geographic data, breakthroughs in artificial intelligence methods represented by deep learning have infused vitality into research on processing big data, whereby computational methods have been quickly developed. Chinese scientists have applied deep learning neural networks to process big data, such as remote sensing images, street view images, text data, and social media data, resulting in important breakthroughs. For example, deep learning has significantly outperformed traditional methods in the fusion of multisource information such as time series, space, spectrum, and semantics (Zhang and Luo, 2020).

Tasks that must be accomplished for geographic remote sensing science include vigorous development of fundamental theories and methods, wide application of remote sensing to various branches of geographic science, development of basic geographic remote sensing products, and improving the capability of supporting geographic science and the United Nations Sustainable Development Goals. Future tasks include performing more studies on the mechanism of remote sensing, Earth surface system models, big Earth data methods, and the assimilation of remote sensing information into Earth surface system models. In particular, efforts should be focused on solving three major scientific challenges in geographic remote sensing science, i.e., quantitative remote sensing inversion, scale transformation, and remote sensing product validation, to realize major breakthroughs and discoveries in geographic science.

Emerging information technologies can be embedded in geographic information science to develop geospatial cognition, expression, analysis, simulation, prediction, and optimization methods and to explore methods for mapping natural geographic spaces and human and social spaces to geographic information space. Future tasks include geographic scene modelling, solving fundamental scientific problems for the realization and application of geographic information systems, and establishing fundamental theories and key technologies to solve stranglehold problems in China’s geographic information industry.

Geographic data science should deeply integrate observations, data, and model simulations facilitated by rapid developments in the areas of digital Earth, big data, cloud computing, artificial intelligence, and other emerging technologies in combination with remote sensing information analysis and application, geographic information, and geographic analysis methods. In addition, automatic extraction and discovery of hidden geographic knowledge and patterns are used to determine the spatial-temporal relationship between multiscale geographic events and geographic elements and characterize the occurrences of these elements. The main problem of ‘geographic data is exploding while geographic knowledge is still lack’ can thus be effectively addressed.

(1) Key directions for developing a fundamental theoretical framework of information geography include systematically analysing the spatial-temporal distribution, evolution process, and elemental interaction laws of information in information space and formulating theories and methods for analysing the spatial-temporal distribution and pattern characteristics of information. (2) Key directions in the area of fundamental theories and methods of intelligent analysis and simulation of multifaceted information are to combine the integrated expression theory of multifaceted information with the fusion of ‘geometric representation, algebraic calculus, mechanistic processes, and cyber systems’ and to explore unified organization and integration methods for multisource data. Other key directions in this area are performing spatial analysis, inference, decision-making, and service modelling on structured and unstructured geographic data, along with building geo-intelligence and geo-simulation systems in combination with physically based process models and machine learning methods. (3) Key directions in the area of the theory and methods of adaptive computing of multifaceted information include the development of geographic law-driven spatial data structuring and indexing methods in addition to data models, data structures, collaborative computing, and system modelling methods for the high-dimensional spatial description and analysis of big geographic data and to break through performance bottlenecks in spatial computing methods in mobile GIS, three-dimensional (3D) GIS, big data, and smart cities. (4) Key directions in the area of geospatial information visualization and holographic mapping are advancing key theoretical methods of high-dimensional spatial description, real-time dynamic visualization, and virtual-real fusion display of big geographic data and developing data-driven virtual geographic environment construction methods. Holographic visualization of geographic information with full perspective, full factors, full information, and full contents would thereby be realized.

(1) Key directions in quantitative remote sensing inversion involve achieving breakthroughs in remote sensing mechanisms for dense vegetation, glacier tomography, and urban structures. Other key directions in this area include studying synergistic inversion theory and model-data assimilation for land surface process parameters based on multisource remote sensing and developing novel theories and methods of quantitative remote sensing inversion based on big data, machine learning, artificial intelligence, and other cutting-edge technologies. (2) Key directions in remote sensing product validation include developing novel observation technologies that go beyond remote sensing pixel-scale observations and novel theories of pixel-scale truth estimation that integrate spatial-temporal changes in geographic elements. (3) Key directions in scale transformation are to develop more effective remote sensing observation experiments to collect multiscale data and develop the scale transformation theory and methodology for processing spatiotemporal geographic data. (4) Key directions in vegetation remote sensing involve developing methods for high-resolution dynamic monitoring of vegetation, including the 3D structure and functional factors, expanding fluorescent remote sensing applications, and developing long-time series products of vegetation parameters at global and regional scales. (5) Key directions in land use/land cover remote sensing include building a land cover classification system for big geographic data, developing deep learning-oriented intelligent land cover mapping methods, and generating a time series for medium- and high-resolution land cover products at global and national scales. (6) Key directions in geomorphological remote sensing involve performing more in-depth analyses of geomorphology-oriented multisource remote sensing data and developing comprehensive utilization capabilities and methods that can automatically extract and identify landform information based on artificial intelligence algorithms and high-resolution remote sensing monitoring systems for landform changes. (7) Key directions in hydrological remote sensing are to develop observation systems for key watershed hydrological variables based on the integrated space-air-ground observing system and build a big hydrological data platform by integrating multisource space-air-ground remote sensing observations, models, data assimilation, artificial intelligence, and other methods. Other directions include developing long-term/real-time, high-precision, and high spatial-temporal resolution hydrological remote sensing products with hydrological cycle closure and improving hydrological prediction and water resource management capabilities. (8) Key directions in cryosphere remote sensing involve improving monitoring systems and developing satellite observation plans specifically for monitoring rapid changes in the cryosphere and performing satellite constellation observations in addition to long-term and high-quality cryosphere data products. (9) Key directions in remote sensing for human geography and sustainable development are to carry out innovative researches on big data platforms, Digital Earth, intelligent information processing algorithms, and active service models and to integrate remote sensing products into geoscience modelling and analysis, along with national strategies and needs such as the United Nations Sustainable Development Goals, ‘Belt and Road Initiative’, ‘Beautiful China’, and ‘New Urbanization’.

(1) Key directions in Earth information modelling and scene GIS include the development of conceptual and information models to create abstract descriptions of the Earth system; fundamental theories and methods for scene GIS and geographic information location aggregation; novel theories and methods for material/event-based geographic information modelling and analysis; information geography for the evolution of ternary space; and virtual-real fusion visualization methods for geographic scenes. (2) Key directions in information modelling and simulation of geographic systems are to study acquisition methods for full-azimuth, full-perspective, and full-content geographic information and develop a spatiotemporal integration framework and data fusion method for multisource geographic data. Other key directions are studying comprehensive modelling, simulation, and analysis methods for geographic systems and performing collaborative modelling of geographic scenes to improve analysis of process simulations and spatial optimization of geographic systems. (3) Key directions in vital technologies of geographic information systems include developing automated and efficient construction methods for geographic scene models and studying GIS modelling and analysis methods in addition to intelligent geographic analysis methods for the Earth’s sphere structure and integrated multiple geographic elements. Other key directions include developing mobile GIS, 3D GIS, big data, smart cities, and other GIS application platforms and innovating GIS full-media visualization and interaction methods to support applications for national strategies, regional development and other fields.

(1) Key directions in the theory and application of big geographic data are to study theories and methods for analysing, aggregating, and sharing big geographic data and ubiquitous geographic data; advance key technologies for processing and analysing big geographic data; and investigate data models, data structures, collaborative computing, and system modelling methods for the integrated analysis of spatiotemporal big data. Other key directions include further development of vital methods and technologies such as high-dimensional spatial description, real-time dynamic visualization, virtual-real integration presentation, and deep learning of spatial information of big geographic data, in addition to improving big geographic data standards and formulating a national development strategy for the analysis of big geographic data. (2) Key directions in social computing and sensing involve coordinating the popularization and application of ubiquitous sensors to integrate big spatial data in areas such as social media, human trajectories, and crowdsourced tags with remote sensing data to help ‘understand geography through human activities’ and ‘understand human activities through geography’. (3) Key directions in artificial intelligence geography include developing novel geographic deep neural networks that integrate visualization, analysis, and computing dimensions based on geographic laws and interactions and establish deep intelligent networks exclusively for geoscience scenarios. Other key directions involve studying integrated modelling, dynamic sensing, and causal network analysis methods for complex multisource big geographic data based on artificial intelligence and breaking through bottlenecks in artificial intelligence technology for geographic information organization and management, spatiotemporal analysis, modelling and simulation, and interactive visualization to improve the intelligence level of digital reconstruction, analysis, and prediction of the past, present, and future of the physical geographic world. (4) Key directions in mathematical geography are to investigate the geographic space-time framework for space-time fusion based on non-Euclidean geometry and construct multimeasure, multiscale, formalized, and computable mathematical models for unified visualization of geographic space and time. Other key directions include establishing mathematical expressions and feature measurement models for multiscale and multilevel geographic phenomena and constructing methodological systems for modelling and simulating mathematical geography.