Introduction

The PlanetScope (PS) satellite constellation, launched by Planet Labs Inc., represents a transformative advancement in Earth observation. Operating at an altitude of about 500 km, significantly lower than that of Landsat-8 and − 9 (705 km) and Sentinel-2 satellites (786 km), the PS constellation comprises a fleet of “Doves” – nano-satellite approximately the size of a loaf of bread (10 cm x 10 cm x 30 cm), each equipped with multi-spectral cameras (Planet Team 2023). These cameras capture data across four bands in the visible and near-infrared (NIR) spectrum, ranging from 455 to 860 nm. Since the deployment of the first flock on June 22, 2016, the constellation has expanded to 149 satellites, capable of monitoring 200 million km2 per day and producing 175.8 million images annually, with a spatial resolution of 3 m and daily revisit capabilities (Roy et al. 2021; Planet Team 2023). This extensive observation capacity has positioned PS as a key player in remote sensing, delivering near real-time data crucial for various applications. However, the geographical referencing of PS images can sometimes be challenging due to the lack of ground control points, a limitation that underscores the necessity for cautious application in precise geographical studies (Schaeffer et al. 2022). In addition, the geometric and radiometric quality of PS images typically does not meet the “analysis ready” standard required for smooth integration into analytical workflows (Frazier and Hemingway 2021). Although Planet Labs provides additional processing services to improve image quality, access to these services is subject to the terms of the licensing agreement (Frazier and Hemingway 2021).

The introduction of commercial satellite platforms like PS has significantly broadened the scope of cloud-free imaging, which is especially valuable for monitoring natural phenomena such as vegetation dynamics, agricultural developments, and environmental changes. The daily imaging capability of PS provides a fresh perspective, particularly in applications requiring frequent monitoring, offering a distinct advantage over traditional optical satellite platforms like Landsat and Sentinel, which have less frequent revisit time, as noted in previous studies (Mansaray et al. 2021). Thanks to these advantages, significant applications of PS data have already been explored across various fields including remote sensing, environmental monitoring, agriculture, pollution tracking, and natural hazards assessment. Studies leveraging PS data have demonstrated its utility in addressing urgent global challenges, such as in forest change detection (Francini et al. 2020), crop health monitoring (Kokhan and Vostokov 2020; Gomez Selvaraj et al. 2020), and evaluating the impacts of natural disasters (Tavakkoli Piralilou et al. 2019; Levin and Phinn 2022). Additionally, PS data has been employed in analyzing land use/land cover (LULC) changes (Mishra et al. 2020; Acharki 2022), mapping plastic-covered greenhouse farming areas (Acharki and Kozhikkodan Veettil 2022; Veettil and Acharki 2023), identifying waste burning plumes (Scott et al. 2023), wildfire burned area mapping (Pham-Duc and Nguyen 2024), and particulate matter air quality (Le Roux et al. 2021). These applications highlight PS’s growing role in monitoring Earth’s surface, land dynamics, and ecology and environmental changes.

Despite the technological advancements of PS satellites, its integration into the broader research community has been relatively gradual compared to other freely accessible platforms like Landsat and Sentinel. This slower adoption can be attributed by the commercial natural of PS data, which imposes access restrictions, particularly impacting researchers from developing countries with limited financial supports. Additionally, the broader spectral range of Landsat and Sentinel satellites enhances their utility for a wider range of scientific missions. Managed by major agencies such as the National Aeronautics and Space Administration (NASA), the United States Geological Survey (USGS) and the European Space Agency (ESA), Landsat and Sentinel platforms benefit from numerous support programs that facilitate large-scale research activities and collaborations with leading researchers and institutions worldwide.

Today, 8 years after the successful launching of the first PS’s flock, PS data is more and more popular, and has been used increasingly within the remote sensing community (Frazier and Hemingway 2021). This increasing trend in the number of scientific publications is also observed with data acquired by other satellites (Yan et al. 2021; Hemati et al. 2021; Pham-Duc and Nguyen 2022). Given the increasing volume of scientific publications, there is an urgent need for a comprehensive analysis on the development of scientific contributions and research trends associated with PS data. Among different methods, bibliometric analysis has become an effective approach for the quantitative investigation of scientific literature (Broadus 1987). This approach employs statistical analysis to assess publishing trends and uncover relationships between published works, playing a crucial role in understanding global dynamics within a research field (Salinas-Ríos and García López 2022). Additionally, results of bibliometric analysis help identify key research themes, evaluate the impact of scientific publications and researchers, and assist organizations in forecasting and planning figure research directions (Ninkov et al. 2021; Salinas-Ríos and García López 2022).

In recent years, bibliometric analysis has played an increasingly important role in comprehensively assessing research trends and supporting management and decision-making in science and technology (van Raan 2014). It has been used to investigate the development of diverse research fields such as research in sustainable development goals (SDG), deforestation, microplastic pollution in the marine environment, Quantum-Dot-Sensitized solar cells research, and applications of Google Earth Engine in remote sensing and earth science research (Aleixandre-Benavent et al. 2018; Pham-Duc et al. 2023; Mishra et al. 2023, 2024; Dan et al. 2023). In Earth observation, several review and bibliometric studies have accessed the scientific contribution and research trends associated with Landsat, Sentinel, and MODIS satellites (Chaves et al. 2020; Pham-Duc et al. 2020; Yan et al. 2021). These studies have provided quantitative statistics on the visibility of these satellite missions in the scientific literature. Additionally, they have identified trends, gaps, and, to some extent, suggested directions for future studies related to the Landsat, Sentinel, and MODIS satellite data. In contrast, despite the growing body of research output using PS data, there is a notable absence of a comprehensive bibliometric study focused on PS data. This study aims to address this gap by systematically analyzing the scientific contributions and research trends related to PS data. This bibliometric analysis will summarize the general characteristics and trends of scientific publications using PS data, evaluate global collaboration networks, and identify popular research topics. By focusing on scientific articles and conference papers, this study will provide insights into the global utilization of PS data and help forecast potential research directions related to PS data in the field of remote sensing.

After the introduction, Sect. 2 presents main processing steps in our methodology. Results are provided in Sect. 3, and discussed in Sect. 4. Section 5 concludes this study, and highlights its main limitations.

Methodology

The study utilized data extracted from the Scopus database, which was selected for its extensive bibliographic and citation coverage, surpassing that of other databases like PubMed and Web of Science (Eito-Brun 2018; Mishra et al. 2021). The use of the Scopus databases ensures that the bibliometric analysis presented in this paper was conducted from the largest possible number of publications related to PS data (Gavel and Iselid 2008; Meho and Rogers 2008). The search query string in Box 1 was utilized to identify all publications used for this bibliometric analysis, while Fig. 1 illustrates the flowchart of publication selection process. First, the authors selected all publications being published between January 2017 and December 2023, and having either keywords “planetscope” or “planet scope” in their titles, abstracts or keywords. The result returned 647 publications. Although using only these two keywords may result in missing some relevant publications that discuss PS data without explicitly mentioning these terms, developing a comprehensive search string that captures all publications related to the use of PS data is a challenging task. Nevertheless, the authors are confident that the vast majority of PS-related documents were included in our publication collection. As such, we do not anticipate that this limitation will have a significant impact on the overall trends or conclusions drawn from our bibliometric analysis. Second, 33 documents were excluded as they were written in languages other than English. These exclusions were done because articles in other languages may adhere to different publication standards and citation practices, which could introduce inconsistencies when compared to those published in international journals. Additionally, analyzing non-English documents required resources such as translation and specialized language skills, which the authors currently lack. The remaining 614 publications were scanned to exclude those not classified as articles or conference papers. As a result, 28 articles of other types were excluded to focus solely on peer-reviewed documents, ensuring that our analysis is based on high-quality research. The titles and abstracts of the remaining 586 publications were fully scanned, leading to the exclusion of 4 publications due to their content being irrelevant to the research objective. Finally, the metadata of the final publication collection of 582 documents was exported from the Scopus database to a comma-separated values (.csv) file for post-processing using VOSviewer version 1.6.20 (van Eck and Waltman 2010) and Biblioshiny version 4.0 (Aria and Cuccurullo 2017), using techniques presented in recent publications (Pham-Duc and Nguyen 2022; Pham-Duc et al. 2023). Biblioshiny is an open-source tool designed for conducting comprehensive science mapping analyses of scientific literature. Developed in the R environment, the software offers flexibility and facilitates integration with other statistical and graphical packages, allowing for rapid updates and integration (Aria and Cuccurullo 2017). On the other hand, VOSviewer is a free Java-based software designed for visualizing and analyzing citation networks within scientific collections. The software is effective in generating graphical representations of bibliometric maps, particularly for large datasets, making the visualizations easy to interpret (van Eck and Waltman 2010). Note that the search query string and the publication selection process were conducted on 20 January, 2024.

Box 1 The search query string used in this study.

TITLE-ABS-KEY ( “planetscope”  OR  “planet scope” )  AND  PUBYEAR  >  2016  AND  PUBYEAR  <  2024  AND  ( LIMIT-TO ( LANGUAGE ,  “English” ) )  AND  ( LIMIT-TO ( DOCTYPE ,  “ar” )  OR  LIMIT-TO ( DOCTYPE ,  “cp” ) )

Fig. 1
figure 1

The flowchart of publication selection process applied in this study

Full size image

Results

General information

General statistical information of the publication collection is shown in Table 1. During the 2017–2023 period, 438 articles (75%) and 144 conference papers (25%) related to PS have been published in 172 different sources. At the time of this study, the publication collection received a total of 6,063 citations, making the average citations per document being 10.42. As shown in Fig. 2, there are 150 documents which did not receive any citation yet, which accounts for one-fourth of the collection. For the remaining, 280 documents received less than 10 citations (48.11%), 65 received between 11 and 20 citations (11.17%), 61 received between 20 and 50 citations (10.48%), and 26 received more than 50 citations (4.47%). The collection of 582 documents were conducted by 2,299 authors, with 2,985 author appearances, making the average co-authors per document being around five. There were 17 single-authors who published only 22 single-authored documents, which accounts for only 3.8% of the collection.

Fig. 2
figure 2

Citation distribution of the publication collection

Full size image
Table 1 Main statistical information of the publication collection during the 2017–2023 period
Full size table

The annual number of documents related to PS data published during this 7-year period is shown in Fig. 3a. The first article was published in 2017, evaluating a new class of optical PS observations for detecting horizontal surface displacements caused by the 14 November, 2016 Mw 7.8 earthquake in Kaikoura, New Zealand (Kääb et al. 2017). This was also the only publication being published in 2017. In the next three years, the number of publications increased rapidly, with 19 documents in 2018, 64 documents in 2019, and 95 documents in 2020. 2021 marked the first time the annual number of publications exceeded three digits, being 109. In the recent two years, the growth rate has slowed down, reaching 138 documents in 2022, and 156 documents in 2023.

Fig. 3
figure 3

a Annual number of publications, and (b) Research disciplines in which PS data has been used

Full size image

After being indexed in the Scopus database, each document is automatically classified to the most related Scopus Categories of research. Figure 3b shows the ten research disciplines in which PS data has been utilized the most. The most relevant research area is Earth and Planetary Sciences, which accounts for approximately 35.6% of the publication collection. Computer Science and Environmental Science account for 13.7% and 13.3%, while Agricultural and Biological Sciences and Engineering account for 9.1% and 8.2%, respectively. Other research areas where PS has been used includes Social Sciences (6.1%), Physics (5.1%), Mathematics (2.6%), Materials Science (1.8%), and Energy (1.7%). Other research fields only account for 2.8% of the collection.

Contribution by countries

The publication collection was published by scholars based in 89 different countries; however, the top ten most productive countries accounted for more than 70% of the total number of documents (n = 414), and 83% of the total citations (n = 5,004). Detailed information of these countries is listed in Table 2. USA published the most documents (n = 153) and received the most number of citations (n = 2,111). These numbers account for more than one-fourth of the total documents, and one-third of the total citations, respectively. With more than 50 publications from each country, China (n = 57) and Brazil (n = 52) were in the next two positions, while Italy was in the fourth position with more than 40 publications. Other countries in Table 2 are from Europe (Germany and UK), Asia (Indonesia, India, and Japan), and Australia.

Table 2 The most productive countries based on the number of documents and citations
Full size table

Among these countries, only a half have published more than three documents (n = 48). The international collaboration between these 48 countries is shown in Fig. 4. Each node represents a country, and the size of nodes is proportional to the number of publications. Close partner countries are identified and displayed in the same colors. A total of seven country clusters are grouped, and the main contributors of each cluster are listed in Table 2. The first and most important cluster (green) is formed around the USA, China, Indonesia, and three European and four Asian countries. However, contributions from these seven countries are much limited than the three main contributors. The second cluster (red) includes eleven partner countries with India, Australia, and France are main contributors. The third cluster (blue) includes the UK, the Netherlands, Italy and Spain in Europe, South Africa, Kenya, and Zimbabwe in Africa, and Mexico. The four European are the main contributors of this cluster. Brazil – the third most productive country, Canada, and Malawi form the fourth cluster (orange). Contribution of the remaining three clusters is much less significant compared to other main clusters. In addition, country members of these three clusters are mostly located in Europe and Asia, with Belgium, South Korea, and Saudi Arabia in the yellow cluster, Sweden and the Philippines in the violet cluster, and Russia and Austria in the cyan cluster.

Fig. 4
figure 4

The international collaboration network of 48 countries which published at least three documents related to PS observations

Full size image

Contribution by scholars

The most productive research teams who published the most number of publications using PS satellite observations are shown in Fig. 5. Each node represents a scholar, and the size of nodes is proportional to the number of publications, published by each scholar. After removing all isolated scholars who has no research collaboration with others, Fig. 5 illustrates seventeen research groups formed by sixty different scholars from various countries. The biggest cluster (red) includes eight scholars, mainly from Brazil, who focus on crop mapping and biomass estimation using PS observations. The second cluster (green) also includes eight scholars, centered on Wang J. and Wu J. from University of Hong Kong. This research team uses PS observations to study forestry and leaf phenology. The third research group (blue) is formed around Sadeh Y. from Monash University, Australia, focusing on data fusion between PS and other optical satellite observations for leaf area index (LAI) monitoring. The fourth research group (violet) is formed around Meena S. from University of Twente, Netherlands who uses PS observations to support the research direction related to landslide mapping. Wicaksono P. is the main contributor of the fifth research group (pink), with an outstanding number of publications compared to other groups (n = 17). Many publications from this team focus on using PS observations for sea grass mapping. These five groups are the most productive research teams based on both the number of publications (Fig. 5) and citations (Fig. 6). The number of scholars in the other twelve research groups is limited, varying between 2 and 4, making their contributions being insignificant to the community.

Fig. 5
figure 5

The most productive research groups working with PS satellite observations. Each node represents a scholar, and the size of nodes is proportional to the number of publications

Full size image
Fig. 6
figure 6

Similar to Fig. 5, but the size of nodes is proportional to the number of citations

Full size image

Table 3 shows the list of top ten most productive scholars, along with other information about the working institutions, number of citations, and the main research directions using PS observations. Except for Dalagnol R. and Zhu X. who are in the last two positions, all other eight scholars are from four over five most productive research groups shown in Fig. 5. Wicaksono P. ranked first with 17 articles and 164 citations. There are three scholars from Brazil who published 7 articles, three scholars from Brazil and Hong Kong published six, and three scholars from the Netherlands, the USA and Germany published five articles. Based on the number of citations, Meena S. has 237 citations, followed by Wicaksono P. with 164 citations, and Wang J. and Wu J. with both 138 citations. The number of citations of other scholars were much less, varying from 48 to 85 citations, as counted at the time of this study.

Table 3 Top ten scholars published the most number of articles using PS observations
Full size table

Contribution by scientific journals

Contribution of different scientific journals is shown in Fig. 7 (based on the number of articles), and Fig. 8 (based on the number of citations). Note that the authors excluded conference articles from this analysis.

  • Remote Sensing (RS) stands out as the most popular journal for studies related to the use of PS data as it published a total of 144 articles and received 1950 citations. This numbers account for 26% and 32% of articles and citations of the collection, respectively.

  • Remote Sensing of Environment (RSE) is in the second position, with 31 articles (7%) and 739 citations (12%).

  • ISPRS Journal of Photogrammetry and Remote Sensing (ISPRS) also shows a strong presence, with 14 articles and 416 citations, respectively.

  • Journals like International Journal of Applied Earth Observation and Geoinformation and International Journal of Remote Sensing demonstrate a balance between the number of published articles and citations, indicating consistent quality in the research they disseminate.

  • Other journals collectively account for a significant portion of the articles (n = 157) and citations (n = 1287). This reflects a wide interest in the subject matter across various publications, pointing to a diverse and multidisciplinary research landscape of PS data.

As accounting for nearly one-third of articles and 45% of citations, RS and RSE are specialized forums for scholars to exchange knowledge related to the use of PS observations. The high citation numbers in the top three journals – RS, RSE, and ISPRS – suggest that articles published in these journals are influential and widely referenced within the academic community, highlighting their prestige and the relevance of their published works to the field. Journals that appear in both Figs. 7 and 8 tend to focus on environmental science, remote sensing, and geo-information, which are directly related to the applications of PS imagery. High citation counts in these specialized journals can reflect the relevance and applicability of PS data to pressing global issues such as climate change, agricultural productivity, and environmental monitoring. These journals serve as crucial platforms for disseminating research findings, facilitating the exchange of knowledge, and driving innovation in the application of satellite imagery for Earth observation.

Fig. 7
figure 7

Number of articles used PS observations by scientific journals

Full size image
Fig. 8
figure 8

Number of citations received by scientific journals

Full size image

Contribution by funding sponsors

Table 4 indicates the contribution of funding sponsors, which provides valuable insight into the sources of financial support for research projects utilizing PS imagery. With NASA, NSF, and USGS combined, the USA leads in the total number of funded projects related to PS imagery. This is indicative of the country’s significant investment in space, Earth sciences, and technological development. NASA is a leader in space exploration and Earth sciences. With 49 projects funded, NASA’s investment indicates a strong commitment to research utilizing Earth observation data, including PS imagery. These projects are likely to be diverse, covering areas such as climate research, agricultural monitoring, disaster response, and environmental protection.

Table 4 The top ten funding sponsors for scientific projects investigating on the use of PS observations
Full size table

Brazil also shows strong support with the National Council for Scientific and Technological Development (CNPq), CAPES Foundation, and Sao Paulo Research Foundation contributing to a significant number of projects. This reflects Brazil’s focus on scientific research and development, likely driven by the country’s rich biodiversity and the importance of agricultural and environmental monitoring. CNPq, from Brazil, with 25 projects funded, is a governmental agency dedicated to the advancement of scientific research and technology development in Brazil. The focus here is likely to include biodiversity conservation, Amazon rainforest studies, and agricultural efficiency.

The European Union, represented by ESA and Horizon 2020, demonstrates a commitment to space and scientific research through substantial funding. This illustrates the EU’s approach to fostering innovation and addressing environmental challenges at a continental scale. The ESA, representing the EU, supports 39 projects and has a broad mandate similar to NASA’s but with a European focus. ESA’s support for PS-related research likely encompasses climate change, environmental monitoring, and agricultural studies, and promotes collaborative research across member countries.

With 26 projects funded by the National Natural Science Foundation of China, China’s investment underscores its growing role in global scientific research, particularly in the fields of Earth observation and environmental monitoring. The NSFC is a key agency within China that sponsors basic and applied research in all branches of natural science. Funding 26 projects indicates China’s growing interest in utilizing satellite data for various research initiatives, such as urban planning, natural resource management, and environmental monitoring.

Lastly, Natural Sciences and Engineering Research Council of Canada (NSERC) is the primary funding agency supporting scientific and engineering research in Canada. With 10 projects funded, NSERC’s investment in PS imagery use might cover a wide array of research areas, including environmental science, natural resource management, and technological development in remote sensing.

Analysis of most cited articles

The most cited articles in Table 5 indicate a growing interest in research utilizing PS data, with several highly cited papers published between 2019 and 2021. This trend highlights the increasing application of PS data in remote sensing research. The paper with the highest citation count, published in 2020 in the Egyptian Journal of Remote Sensing and Space Science (Mishra et al. 2020), has received 153 citations, averaging 30.6 citations per year. Following closely, the paper by Tassi A. (2020), published in Remote Sensing, has received 135 citations with an annual average of 27 citations per year. Another significant contribution is from Piralilou S.T. (2019), also published in Remote Sensing, which has gained 110 citations, averaging 18.33 citations per year. At the time of this study, these are the three articles which received more than 100 citations, while the citations count for other articles in Table 5 ranged from 73 to 97, underscoring their substantial influence within the scientific community. These highly cited papers were published in various high-quality journals in the field of remote sensing and geoscience, including International Journal of Remote Sensing, ISPRS Journal of Photogrammetry and Remote Sensing, Remote Sensing of Environment, Remote Sensing, The Egyptian Journal of Remote Sensing and Space Science, and Landslides. All these journals are classified in the first quartile (Q1) category of the Scimago journal ranking, suggesting that research using PS data is well regarded in the scientific community.

Table 5 The top ten articles related to the use of PS observations that received the most number of citations
Full size table

Furthermore, Table 5 reveals that among the top ten most cited articles, four exclusively use PS images as their primary dataset, three combine PS images with Landsat and Sentinel images, and the remaining three employ PS as reference dataset for accuracy assessment. This distribution highlights the adaptability of PS data for various scientific purposes in advancing research in the field of remote sensing. However, PS imagery is not the first choice of data within the research community. Notably, five of the top ten articles focus on LULC classification and landslide detection, reflecting the remote sensing community’s high demand for accurate datasets at the local scale in these research directions. Moreover, the leading authors of these ten articles are from ten different countries, indicating the global popularity of PS data within the remote sensing community.

Analysis of most popular keywords

The word cloud highlights the diverse range of research areas utilizing PS data, spanning from environmental monitoring and agriculture to machine learning and remote sensing techniques. Figure 9 presents the 100 most frequently used keywords in the publication collection, with larger words indicating higher frequency or importance, while smaller words represent less commonly used terms. It is evident that keyword “planetscope”, appearing at the center of the word cloud (n = 103), is the most important, while terms like “satellite data” (n = 78) and “satellites” (n = 75) are closely related, emphasizing that most studies focus on satellite imagery and data analysis. The prominence of “machine learning” (n = 83) among the top 100 keywords suggests that these methods are widely employed in PS data-based studies, such as for image classification, mapping, and prediction. PS’s notable advantage of high spatial resolution makes it attractive for research requiring detailed spatial data, which explains the importance of keywords like “spatial resolution” (n = 74) and “image resolution” (n = 66) in Fig. 9. Keywords like “vegetation”, “NDVI”, “forestry”, “land use”, “climate change”, and “crops” further illustrate the increasing importance of PS data in environmental and agricultural applications. Additionally, the presence of terms like “landsat”, “sentinel”, and “sentinel-2” indicates the relationship between PS data and other satellite platforms. Lastly, terms such as “tropics”, “united states”, “china”, “brazil”, and “indonesia” reflect the broad geographic distribution of research utilizing PS data.

Fig. 9
figure 9

Word cloud of the 100 most popular keywords

Full size image

A world cloud is useful for the visualization of important research topics; however, it is impossible to draw the connection between these keywords. A deeper analysis, focusing on the co-occurrence of the 100 most popular keywords is shown in Fig. 10. Each node represents a keyword, and the size of nodes is proportional to the importance of keywords. Relevant keywords, determined by the frequency they appeared together in other publications, are identified and grouped into five different clusters, coded by different colors:

Fig. 10
figure 10

The co-occurrence network of the 100 most popular keywords shown in Fig. 9

Full size image

  • The first cluster (red) includes articles focused on the use of PS and other satellite-derived data for agricultural research based on applications of vegetation indices.

  • The second cluster (yellow) includes articles focusing on the applications of radar remote sensing on floods, landslides, and deforestation. For articles in this cluster, PS observations were used mainly for accuracy assessment to validate results obtained from SAR data.

  • Articles in the third cluster (blue) mainly used PS observations as the reference dataset for accuracy assessment with results derived from other optical satellites, such as Sentinel-2 and Landsat.

  • The fourth cluster (green) includes articles that used PS as the primary dataset to study several topics such as forestry, land use land cover mapping, and ecosystems.

  • Finally, the fifth cluster (violet) suggests the most important machine learning methods that have been utilized with PS data, such as artificial neural networks and deep learning networks.

Discussions

PS satellite imagery is increasingly gaining attention from the scientific community and is being widely used in scientific research across various fields. The increasing number of publications and citations, particularly in Earth and Planetary Sciences, highlights PS’s significance in high-resolution environmental monitoring and assessment. The trend towards more frequent use of PS data indicates its growing importance within the research community, fulfilling critical needs for high spatial and temporal resolution imagery. Additionally, the multidisciplinary nature of PS data usage is also indicated through the interconnectedness of clusters in Fig. 10. For instance, studies on agriculture (red cluster) are frequently linked to machine learning techniques (violet cluster), reflecting the use of AI to improve agricultural monitoring. Similarly, environmental monitoring (green cluster) and disaster management (yellow cluster) are interconnected through the application of both optical and radar remote sensing data, showing a trend toward the integration of multiple datasets and methods in solving complex environmental challenges. The significant presence of machine learning, deep learning, and artificial neural networks in the violet cluster indicates that PS data is increasingly being used in conjunction with advanced computational methods (Fig. 10). This suggests that researchers are moving beyond traditional classification techniques and are exploring more sophisticated approaches to analyze the rich detail provided by PS imagery.

Nevertheless, in terms of quantity, the articles and research utilizing PS images in the field of earth observation cannot be compared with those using Landsat or Sentinel data due to the latter two’s availability and long-term archival (Wulder et al. 2022). Disregarding technical aspects, providing these images for free is one of the most crucial factors that enable the remote sensing research community to access this valuable data. Commercial remote sensing images are primarily hindered by their high cost, which constitutes a major obstacle in their access and utilization for scientific research, especially in the low income countries. Nonetheless, PS, a commercial remote sensing entity, has shown growth in the number of scientific publications that utilize its data. This result is evident in Fig. 3, which shows the growth rate and the diversity of research fields in which PS images are used. This growth is also partly due to the support policy of Planet Labs through Planet’s Education and Research Program. This program provides limited, non-commercial access to PS imagery for any university-affiliated student, faculty member, or researcher (https://www.planet.com/markets/education-and-research/).

PS satellites provide high spatial-temporal resolution imagery, making them particularly suitable for detailed environmental and ecological monitoring, natural disaster assessment, land surface dynamics, and earth observation. This capability is essential for accurately mapping small-scale features such as individual crop types, crop growth, plastic-covered greenhouse farming areas, small forest patches, and even specific stages of vegetation growth or health (Kokhan and Vostokov 2020; Cheng et al. 2020; Gomez Selvaraj et al. 2020; Acharki and Kozhikkodan Veettil 2022). High-resolution data allow researchers to conduct fine-scale analyses that are crucial for understanding ecosystem dynamics, land use/land cover change, and environmental pollution (Tran and Tran 2021; Schaeffer et al. 2022; Bueno et al. 2023). The constellation of PS satellites offers frequent revisit times, ensuring up-to-date imagery is available for continuous monitoring (Roy et al. 2021). This is particularly advantageous for applications like crop monitoring and forestry, where changes can occur rapidly due to seasonal variations, agricultural practices, or environmental factors (Kokhan and Vostokov 2020; Francini et al. 2020; Bhavana et al. 2023). In addition, the high resolution and timely availability of data make it invaluable for assessing damage, planning recovery efforts, and improving resilience against future events. This application explains the interest in using PS data for landslide mapping, flood impact assessment, wildfire assessment, and oil spill monitoring (Tavakkoli Piralilou et al. 2019; Schaeffer et al. 2022; Levin and Phinn 2022; Pham-Duc and Nguyen 2024). The top ten of most cited articles also shows the diversity of research area related to the use of PS dataset including LULC classification, landslide detection and mapping, crop monitoring, sea grass mapping, and vegetation dynamics (refer to Table 5). These trends reflect PS’s utility in providing detailed, timely data that supports a wide range of research areas.

This study’s findings not only demonstrate the wide range of applications for PS imagery but also spotlight the major contributors to research based on PS. The varied geographical representation of contributing institutions from countries like Brazil, Indonesia, Hong Kong, India, Canada, the USA, the Netherlands, and Germany, underscores a global interest in leveraging high-resolution satellite imagery for a broad spectrum of scientific inquiries. This wide-reaching involvement highlights PS’s global applicability and continued role in tackling local and international environmental and geographic challenges. However, despite these positive trends, our analysis also identified several challenges associated with PS data. The commercial nature of PS imagery limits access for researchers, particularly in low-income countries, leading to regional disparities in scientific output and collaboration. This restricted accessibility can skew research trends, concentrating contributions in wealthier nations with the resources to purchase data. Table 4 shows that most of the top 10 funding sponsors for PS-related scientific projects are from developed countries, including high-income nations like the USA, Canada, and EU, and upper-middle-income countries like Brazil and China, according to the World Bank classification in 2024 (https://data.worldbank.org/country). To address these disparities, initiatives such as open-access data programs, like the Planet’s Education and Research Program, and collaborative agreements between institutions are critical in democratizing access to PS data. Additionally, implementing image reuse policies that support scientific research communities is strongly encouraged (Roy et al. 2021). Furthermore, expanding data policies, including government-funded data purchases like Norway’s International Climate & Forests Initiative (https://www.nicfi.no/), is recommended to enhance data accessibility (Roy et al. 2021).

The international research collaboration network is influenced by a range of complex factors, including scientific, economic, geopolitical, and cultural considerations (Hou et al. 2021). For PS data, these partnerships are likely driven by international development goals, shared research priorities, and funding from international organizations. As shown in the analysis, PS data is attracting global interest, particularly for addressing local and international environmental challenges through high-resolution satellite imagery. Consequently, the network of collaborations using PS data is expanding. Countries such as the USA, China, Germany, and the UK serve as central nodes in this network (Fig. 4), reflecting their strong research infrastructures and technological capabilities in satellite data processing and analysis. These countries often act as hubs for collaboration, providing access to advanced technologies and satellite imagery while benefiting from the unique environmental and agricultural contexts of their partner countries. Furthermore, many collaborations are driven by shared environmental, agricultural, or disaster-related challenges. For instance, tropical countries like Brazil, Indonesia, Vietnam, and India collaborate with global leaders in satellite data processing to address issues such as LULC change, deforestation tracking, biodiversity monitoring, and climate change. Similarly, African countries (e.g., Kenya, Zimbabwe, Malawi, Ethiopia) work with European nations to tackle food security, land degradation, and disaster management, leveraging their partners’ technological expertise. Moreover, collaborations involving PS data are often influenced by funding mechanisms and policy frameworks. International funding programs, such as Horizon 2020 and Horizon Europe (https://research-and-innovation.ec.europa.eu/funding_en), play a key role in promoting collaboration by encouraging large-scale projects that address global environmental challenges. Bilateral agreements between space agencies (e.g., NASA and ESA) also support data-sharing initiatives, allowing countries to access critical satellite data for research purposes. Additionally, funding agencies are crucial in supporting scientists’ research projects and facilitating the expansion of international collaboration networks (see Sect. 3.5 and Table 4). These funding mechanisms often support capacity building and knowledge transfer, ensuring that developing nations benefit from the expertise of more technologically advanced partners.

The research community using PS data appears to be highly fragmented over the study period. The first authors of the 10 most cited articles came from 10 different research institutes/universities located in 10 different countries (Table 5). This result supports the findings in Fig. 5, which indicates that the most effective research groups working with PS satellite data seem to be isolated. Unlike the applications of Landsat or Sentinel imagery, which many leading scholars/research groups from many countries around the world have collaborative research connections with each other. This finding can partly be explained by the fact that PS imagery is commercial satellite products; therefore, access to this dataset for research is still limited for scholars, especially those from developing countries. Furthermore, PS imagery has only been released recently (2016). Meanwhile, both Landsat and Sentinel imagery are being provided freely by the USGS and the ESA since many years. Specifically, Landsat data has a long archival history, from the time the first Landsat image (Landsat 1) was made available in 1972, Landsat data has been archived for more than 50 years (Crawford et al. 2023; Wulder et al. 2022). More importantly, the data management agencies of the Landsat and Sentinel-2 programs also have many efforts related to providing a model for future collaboration and cooperation (Wulder et al. 2022). For PS data, it is encouraging that the user community is growing and increasingly sharing algorithms, codes, and software that help to correct and improve the geometric and radiometric quality of PS imagery (Frazier and Hemingway 2021). Moreover, creating platforms for data sharing and collaboration, organizing international workshops, or encouraging joint projects between countries could be potential solutions to enhance collaboration of this research community.

In bibliometric analyses, the purpose and context in which a data source is used can significantly affect the interpretation of citation metrics (Haustein and Larivière 2015). While PS imagery’s potential is well-recognized, the discussion must acknowledge that some of the most cited studies did not exclusively rely on PS data as their primary dataset. At the time the data was extracted for this study, while being among the most highly cited papers related to PS data according to the provided list (Table 5), the Mishra P.K. (2020) study did not actually employ PS imagery as the primary data source for LULC change analysis. In the Mishra’s study, Landsat and Sentinel datasets formed the core remote sensing data, with PS used only for the specific purpose of accuracy assessment of the 2017 LULC classification results. This indicates that the citations it has received are not solely for work central to PS imagery. Thus, for a refined analysis, it would be ideal to understand the role PS data played in each of the most cited articles. In the context of our study on PS data, this would mean that the citations for the Mishra P.K. (2020) paper might be less reflective of its influence specifically within the PS research community, compared to other articles that use PS data as their main focus, such as Piralilou S.T. (2019), Wicaksono and Lazuardi (2018), and Cheng Y. (2020) (Table 5). In addition, only four over ten most cited articles in Table 5 uses PS as the primary dataset, while in the other six articles PS data is either used as a complementary nor a reference dataset. This observation suggests that while PS data is valuable, it often plays a complementary role alongside other remote sensing platforms. It indicates that certain research themes, particularly those that rely heavily on multi-temporal or long-term time-series data (e.g., long-term climate studies) or themes that require broader geographical coverage (e.g., global-scale studies), may not be fully represented in the current body of research using PS imagery. Additionally, this suggests that PS data may be more suitable for certain types of research, and the remote sensing community is still exploring its full potential.

PS’s role as a complementary dataset is frequently highlighted in keyword analyses. While PS data is crucial for certain applications (e.g., LULC mapping and vegetation monitoring), it is often used in conjunction with other datasets such as Sentinel-2, Landsat, and SAR to improve accuracy. This trend is evident in the strong connections between keywords like “accuracy assessment”, “Sentinel”, and “Landsat” in the blue and yellow clusters of Fig. 10. Furthermore, the keywords “accuracy assessment” has maintained its high relevance from 2019 to 2022 (Fig. 9), demonstrating sustained interest in using PS data to enhance classification accuracy. This is exemplified by Mishra’s paper, which used PS imagery solely for accuracy assessment and became the most cited in our dataset (Table 5). This nuance regarding the limited use of PS data in this highly cited work should be kept in mind while analyzing the bibliometric trends related to utilization of PS satellite imagery across different studies.

As showed in the keyword analysis, research using PS data appears to be increasingly integrated with advanced processing techniques like machine learning (Fig. 9). PS satellites capture images at a high resolution, providing a wealth of detailed data. Machine learning algorithms excel at extracting insights from complex, high-dimensional datasets, making them an ideal tool for analyzing the rich data provided by PS (Tavakkoli Piralilou et al. 2019; Gomez Selvaraj et al. 2020; Tassi and Vizzari 2020; Bhavana et al. 2023). The frequent revisits of PS satellites generate large volumes of data over time (Roy et al. 2021). Traditional data processing techniques can be insufficient for such big data sets, whereas machine learning algorithms can efficiently handle, process, and analyze this data to uncover trends and patterns that might not be evident through manual analysis. In addition, machine learning models, once trained, can analyze data consistently and objectively, reducing the potential for human error and bias. This can lead to more accurate results in applications such as LULC classification, change detection, and environmental monitoring (Levin and Phinn 2022; Acharki and Kozhikkodan Veettil 2022; Duan et al. 2023; Bhavana et al. 2023).

Despite the promising advancements, the study also highlights challenges and limitations associated with the use of PS imagery. First, the commercial nature of PS data poses accessibility issues, particularly for researchers in low-income countries where financial support from public and private sectors for science is still limited. Although users can take benefits of the Planet’s Education and Research Program, which provides non-commercial access to PS imagery for educational and research purposes, the company only provides 5000 km2 per month, which is not suitable for large-scale monitoring. Second, PS sensors capture imagery in only four spectral bands. While it is useful for general purposes, the lack of other broader spectral bands, including shortwave infrared (SWIR), limits its applications in several important research directions of Earth science such as wildfire detection and monitoring, and water resource management. Third, the small swath of PS imagery (typically covers an area of about 20 km x 20 km) presents many challenges in terms of area coverage, data management, processing complexity, and costs, particularly for large-scale applications (NASA 2020). Fourth, the relatively recent introduction of PS satellites compared to long-established platforms like Landsat and Sentinel has resulted in a smaller body of literature and fewer collaborative research networks.

To overcome the limitations associated with PS images, several strategic solutions can be implemented to enhance its utility in scientific research. For addressing accessibility issues, users from low-income countries should foster partnerships with Planet Labs and international organizations to increase the chances of accessing the data freely. Integrating PS data with Landsat and Sentinel-2 imagery can mitigate the challenges posed by the limited spectral range and small swath, providing broader spectral coverage and larger area monitoring (Cheng et al. 2020; Sadeh et al. 2021; Pickering et al. 2021). Applying data fusion techniques which combine PS data with data from other satellite platforms would enable more comprehensive analyses (Csillik et al. 2019; Cheng et al. 2020). By implementing these solutions, users can partly address the current limitations and maximize the potential of PS data for a wide range of scientific applications of Earth science (Frazier and Hemingway 2021). Last but not least, future research should continue to explore innovative applications of PS data and leverage machine learning techniques to enhance data analysis and interpretation (Frazier and Hemingway 2021).

Conclusion

This bibliometric analysis of PS satellite imagery reveals a rapidly expanding body of research that underscores the value of high-resolution data in various scientific fields, particularly Earth and Planetary Sciences, Environmental Science, and Computer Science. The increasing number of publications and global collaboration patterns demonstrate the growing impact of PS imagery on scientific research, highlighting its role in enhancing the observation of land surfaces and in environmental monitoring and management practices. However, this study also highlights key challenges, including limited access to PS data in low-income regions, which may lead to regional disparities in scientific output and collaboration. The findings suggest that although PS data is valuable in remote sensing research, it is often used in a complementary role rather than as the primary dataset, emphasizing the need for a deeper understanding of its contribution to research outcomes as its potential within the remote sensing community continues to be explored.

To address these challenges, the study recommends expanding open-access data initiatives, enhancing data-sharing agreements. For example, expanding programs similar to Google Earth Engine’s public datasets could democratize access to high-resolution imagery, enabling more equitable research contributions globally. Initiatives such as Planet’s Education and Research Program should be expanded to increase access to PS data for scholars, especially those in less developed countries. Additionally, fostering interdisciplinary collaborations through thematic workshops, conferences, and joint research projects that focus on specific applications of PS data, such as climate change mitigation, biodiversity conservation, and disaster management, will further enhance the use of PS data.

Future research should focus on exploring innovative applications of PS data, integrating machine learning techniques for data analysis, and addressing the identified limitations to improve data quality and accessibility. By implementing these strategies, the scientific community can better leverage PS imagery to drive impactful research and develop sustainable solutions for global environmental issues.

Limitations of bibliometric analysis

The main methodology applied in this study is adopted from previous publications; therefore, this study has similar limitations as already reported (Pham-Duc and Nguyen 2022; Pham-Duc et al. 2023). First, the publication collection used in this study only includes articles indexed in the Scopus database, which does not cover all publications related to the used of PS observations available in the literature. Similar analysis can be done with metadata exported from other bibliographic databases, such as the Web of Science. However, due to current technical limitations, it is not recommended to combine metadata from different databases into one product because of the significant amount of time it takes for data unification and repair (Kumpulainen and Seppänen 2022). Second, using only two keywords in the search query string may result in missing some relevant publications that discuss PS data without explicitly mentioning these terms. However, developing a comprehensive search string that returns all publications related to the use of PS data available in the literature is a very challenging task. Third, data cleaning and correction have been done carefully twice by all co-authors, but it might not be perfect, which can slightly affect to our analyses. Fourth, the number of articles related to the used of PS observations is expected to increase quickly in the future; therefore, this type of analysis has to be conducted regularly as the results may change quickly. Similarly, the citation counts of the articles fluctuate rapidly over time, causing the positions of articles in Table 5 to change accordingly. As a result, comparisons of citation counts per article in this bibliometric study can quickly become outdated.