Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Scientific Data volume 13, Article number: 116 (2026)
Agrivoltaics is a farming method that strategically integrates solar panels with agricultural production, a dual-use system that boosts food production while generating clean energy. China is the one of leading countries in agrivoltaics. However, no robust vectorized dataset has been available to verify the distribution of agrivoltaics in China. This study aims to provide the first nationwide agrivoltaics distribution and type dataset in China using comprehensive identification methods based on published spatial data of photovoltaic power stations and agrivoltaics records. The overall accuracy of agrivoltaics through visual examination is 89.71%. The results show that: (1) By 2022, there are 1,678 agrivoltaics projects in China with a total installation capacity of 134.55 GW. (2) China launched its first commercial agrivoltaics in 2010, reaching a peak of 347 projects in 2017, after which the number of new agrivoltaics projects has remained no less than 140 annually. (3) The three most common agrivoltaics types are crop-based, fishery-based, and greenhouse-based. This vectorized agrivoltaics dataset will support macro-level management and the sustainable development of agrivoltaics.
Global photovoltaic capacity rose to more than 2.2TW in 2024, supplying more than 10% of global electricity consumption for the first time1. China is the largest and fastest-growing country in terms of photovoltaics (PV) installed capacity2. By the end of 2024, China’s cumulative installed PV capacity had reached 1 TW, with an additional installation of 277.1 GW3. For the conservation and land protection priorities by the land-use and land-cover change necessary for PV deployment, Kruitwagen et al.4 developed the first global inventory of PV generating stations over 10 kilowatts nameplate capacity, which located and verified 68661 facilities at the end of 2018. The report of Global Photovoltaic Power Potential by Country, developed by World Bank, provided solar resource and PV power potential from the perspective of countries and regions5. In the in-depth study of China, Liu et al.2 developed and released the vectorized solar photovoltaic installation dataset across China in 2015 and 2020, which met a medium resolution of 30 meters based on Landsat-8. Fang et al.6 provided a remote sensing-derived dataset for a large-scale ground-mounted photovoltaic power station in China in 2020, with a high spatial resolution of 10 meters. Chen et al.7 extracted the spatial extent and installation data of PV power plants from Sentinel-2 and Landsat data. Evidently, photovoltaic power generation facilities across China have been effectively identified, accompanied by detailed spatial positioning data.
Agrivoltaics is a farming method that strategically integrates solar panels with crops, livestock, or aquaculture. This dual-use system boosts food production while generating clean energy, maximizing land use efficiency. Agrivoltaics offers a promising solution in a world grappling with increasing energy demand and competition for agricultural land8. According to the data collected by the National Renewable Energy Laboratory of the United States, there are 314 agrivoltaics projects in the United States representing over 2.8 GW of solar capacity, of which tracking the global latest 689 research in agrivoltaics and only 131 projects located in Asia., with relatively integrated crop production by March 20239. Geospatial data describing agrivoltaics is required to manage grid connection of renewable energy, land protection, mitigation, and adaptation of agriculture, and identify trade-offs with food and energy production caused by agrivoltaics expansion. According to the estimation of Global Infrastructure Review, China leads the way in agrivoltaics, with over 500 projects showcasing the technology’s potential10, including crop cultivation, livestock grazing, aquafarming, greenhouses, and tea plantations11. However, there is no robust vectorized dataset, so far, to verify the type and distribution of agrivoltaics in China, which seriously restricts the macro management and sustainable development of agrivoltaics.
In this study, we aim to provide the first agrivoltaics distribution and type data across China using comprehensive identification methods. The release of this dataset can provide valuable reference for researchers and users in fields such as agricultural management and renewable energy. The potential application of this dataset includes (1) providing distribution of different type of agrivoltaics; (2) estimating the development trends of agrivoltaics; (3) evaluation the adaptation and mitigation impact of agrivoltaics for agricultural production; (4) offering reference for future national and provincial agricultural and renewable energy planning.
Four available photovoltaic spatial distribution datasets can serve as the basis for identifying agrivoltaics projects. Kruitwagen et al.4 developed and provided the first Global Photovoltaic Inventory using a longitudinal corpus of remote sensing imagery, machine learning, and an extensive cloud computation infrastructure (GIPV, https://zenodo.org/record/5005868). Liu et al.2 developed a vectorized solar photovoltaic installation dataset across China using Landsat-8 imagery in Google Earth Engine with an overall accuracy of over 96% for 2015 and 2020 (VSPV, https://zenodo.org/records/14292571). Feng et al.6 developed a national-scale map of ground-mounted photovoltaic power stations in China using Sentinel-2 imagery based on the GEE via field survey and visual interpretation, with an accuracy of over 89% (GMPV, https://doi.org/10.57760/sciencedb.o00121.00001). Chen et al.7 developed a rapid expansion of photovoltaic power plants in China using Sentinel-2 and Landsat data with an overall accuracy of 89.86% (REPV, https://doi.org/10.6084/m9.figshare.25347880.v1) (Table 1).
Furthermore, Global Energy Monitor12 developed a worldwide dataset of utility-scale solar photovoltaic (Global Solar Power Tracker, GSPT, https://globalenergymonitor.org/projects/global-solar-power-tracker/) based on government data on individual solar power farms, reports by power companies, news and media reports, and local non-governmental organizations, with 13490 records related to photovoltaic power stations in China. The data validation is performed by comparing against proprietary and public data, such as S&P’s Global World Energy Power Plant Database and World Resource Institute’s Global Power Plant Database, etc. The GSPT dataset provides important information such as the name of photovoltaic projects, their administrative districts (county level), approximate geographical locations, construction time, and installed capacity, which forms the basis for our identification of agrivoltaics projects.
The China’s Land use/Cover Datasets (CLUD) was developed by Yang et al.13 based on 335,709 Landsat images on the Google Earth Engine with a 30 m spatial resolution from 1985 to 2023 (https://zenodo.org/records/12779975). Major land-cover types were classified as cropland, forest, brush, grassland, water, snow/ice, bare land, impervious surface, and wetland. The tea map developed by Peng et al.14. Crops map and greenhouse datasets downloaded from the National Ecosystem Science Data Center in the National Science & Technology Infrastructure (https://www.nesdc.org.cn/).
The overall workflow is depicted in Fig. 1. We developed comprehensive identification methods, including PV data collections and integration, agrivoltaics extraction from PV datasets, Installation time analysis and classification system, LUCC analysis, jurisdiction analysis, and proximity analysis. The workflow is described as follows.
Framework of vectorized agrivoltaics dataset developed.
We overlaid four Chinese photovoltaic datasets, namely GIPV, GMPV, VSPV, and REPV, and observed that these datasets exhibit variations in their PV identification capabilities. However, their combination can enhance the accuracy of PV identification (Fig. 2). Consequently, GIPV, GMPV, VSPV, and REPV were integrated into a single spatial dataset (named UnionPV) using the Union Module in ArcGIS, with a spatial tolerance of 30 meters. This module was employed to write all input features and their corresponding attributes into the output feature class while eliminating overlaps. UnionPV contains 65,045 PV polygons, which represent the precise spatial distribution of all PV power stations completed in China by 2022.
Overlay and union analysis for GIPV, GMPV, REPV, and VSPV datasets. The green, yellow, blue and red in the middle column represents GIPV GMPV, REPV, and VSPV respectively. The pink in the right side represents UnionPV.
We obtained data records of photovoltaic projects of China from the GSPT dataset, and collaborated with the records of GEM WIKI to obtain the names, districts and estimated geographic coordinate data of 13,490 photovoltaic projects.
According to the regulations of Chinese engineering project construction, the project name should accurately reflect the main content of the project. Therefore, we can identify agrivoltaics from the photovoltaics project name in the GSPT datasets. The feature table of GSPT datasets contained the names of photovoltaic projects, and GEM Wiki provided more detailed information about photovoltaic power stations in China. We used the keywords screening framework to extract information on agrivoltaics from the GSPT datasets. The framework operates as follows: Criterion selection and keywords. We first developed single dimensional criteria to help determine whether the name of photovoltaics power pertains to one type of agrivoltaics, such as the string including “Agro-photovoltaic complementary”, “Greenhouse-Photovoltaic Complementary”, “Chinese traditional medicine-Photovoltaic Complementary”, “Husbandry – Photovoltaic Complementary”, “Fishery – Photovoltaic Complementary”, “Tea-Photovoltaic Complementary”, and “Forestry – Photovoltaic Complementary” from the feature table and obtained the records of agrivoltaics. Then, 2,864 projects with the semantic meaning of “agricultural-photovoltaic complementarity” in their names were extracted. The data tables and Python code used for agricultural photovoltaic extraction have been shared online (https://doi.org/10.57760/sciencedb.26240)15.
During the process of screening agricultural photovoltaic projects from the GSPT dataset via keyword-based retrieval, certain project names—while relevant—exhibit inherent ambiguity. Representative cases include “photovoltaic poverty alleviation farm projects” and “fishery-photovoltaic integrated park projects,” which are prone to causing misjudgments in keyword-driven screening. Accordingly, these projects have been categorized into the “crop-photovoltaic complementarity” and “fishery-photovoltaic complementarity” classifications respectively through manual identification.
The latest available photovoltaic distribution data encompasses projects completed prior to 2022 (Table 1), from which 2,532 such projects were selected (Fig. 1). Then, we established three first-level classifications of planting, breeding, and forestry, and seven second-level classifications of crops, greenhouses, Chinese traditional medicines, tea, livestock, fisheries, and forestry for the agricultural photovoltaic polygon data (Table 2).
As is widely recognized, PV installations situated on urban areas or impervious surfaces do not serve agricultural production purposes. Such PV systems are primarily deployed in the form of rooftop photovoltaics and rural distributed photovoltaics. Based on the China Land Use Dataset (CLUD), this study employed spatial overlay analysis to identify the PV-related segments where the land use type, as recorded in the dataset, corresponds to impervious surfaces. Through this analytical process, a total of 55,606 PV polygons were extracted and compiled into the subsequent PV dataset.
In the process of refining the agrivoltaics spatial dataset, we took the jurisdiction information of the already recorded agrivoltaics projects as a core constraint condition—this jurisdiction information was derived from the project registration materials and had been verified for consistency with the basic administrative division data in the early stage.
On this basis, we further carried out a targeted filtering operation on the initially identified photovoltaic power station polygon dataset. The key objective of this filtering step was to screen out the photovoltaic power station polygons that are geographically located within the same administrative jurisdiction of a prefecture-level city as the recorded agrivoltaics projects.
This operation not only ensured that the filtered photovoltaic power station polygons had a clear spatial correlation with the agrivoltaics projects (avoiding the inclusion of polygons from unrelated administrative regions) but also laid a solid foundation for subsequent regional-scale statistical analysis of agrivoltaics and associated photovoltaic facilities, as well as the study of their spatial distribution patterns within prefecture-level administrative units.
Taking the approximate latitude and longitude coordinates, from GSPT, of the agricultural photovoltaic projects as the center, with radii of 3 km, 5 km, 8 km, and 10 km, the nearest photovoltaic project polygons were searched and matched, and links were established between the agricultural photovoltaic projects and the nearest photovoltaic project polygons. As the results, there are 1678 photovoltaics matched to 1174 polygons, with instances where multiple photovoltaic power stations are situated in close proximity and share a single land parcel, and reported match rates 47.61%, 67.64%, 91.83% and 100% respectively within 3 km, 5 km 8 km, and 10 km.
Following the completion of the aforementioned 6 sequential analytical steps—including data matching, spatial overlay analysis, administrative jurisdiction verification, and targeted polygon filtering as outlined earlier—a total of 1,678 agrivoltaics projects that fully comply with the pre-established comprehensive evaluation method were successfully identified and screened out.
Following the aforementioned methodological framework, a vectorized dataset documenting the spatial distribution and functional types of agrivoltaics across China was constructed by the end of 2022, and this dataset was officially named “China Agrivoltaics Map”.
To facilitate academic research and industrial application, the dataset has been made publicly accessible via the following URL: https://doi.org/10.57760/sciencedb.2624015. As of the end of 2022, the dataset encompasses a total of 1,678 agrivoltaics projects in China, with an aggregated installed capacity reaching 134.55 gigawatts (GW) (see Fig. 3 for details).
The map of agrivoltaics in China.
In terms of the development timeline of China’s agrivoltaics industry, the construction of commercial agrivoltaics projects was initiated in 2010, with 4 projects launched in that year. The industry experienced rapid growth thereafter, peaking in 2017 with 347 new projects completed within the year. Since then, the annual number of newly added agrivoltaics projects has remained stable at no less than 140 (refer to Fig. 4 for the annual project count trend).
The number of agrivoltaics projects from 2010 to 2022.
From a regional perspective, the top 5 provinces in China by agrivoltaics installed capacity are Guangdong, Guizhou, Shandong, Hebei, and Anhui, with respective installed capacities of 16.6 GW, 14.96 GW, 14.61 GW, 13.69 GW, and 9.02 GW. When ranked by the number of agrivoltaics projects, the top 5 provinces are Guizhou (173 projects), Guangdong (158 projects), Shandong (155 projects), Jiangsu (133 projects), and Hebei (129 projects).
In terms of functional types, the three most prominent categories of agrivoltaics projects are crop – based agrivoltaics, fishery – based agrivoltaics, and greenhouse – based agrivoltaics, accounting for 884, 528, and 103 projects respectively (as illustrated in Fig. 5).
The distribution different type of agrivoltaics in provinces of China.
“The China Agrivoltaics Map” comprises an ESRI shapefile with 1,166 polygons (Fig. 3) and a feature list containing 1,678 agrivoltaics records (Table 3; List of China Agrivoltaics sharing online, https://doi.org/10.57760/sciencedb.2624015).
There are 1174 polygons contained in the China Agrivoltaics Map with the features as follows:
Latitude (Lat_Y): latitude of a point representation the land parcel of Agrivoltaics’ location, in decimal degrees, calculated in ArcMap using the calculate geometry tool with the GCS_WGS_1984 coordinate system.
Longitude (Long_X): longitude of a point representation the land parcel of Agrivoltaics’ location, in decimal degrees, calculated in ArcMap using the calculate geometry tool with the GCS_WGS_1984 coordinate system.
Link_ID: a unique stable identification number for each land parcel.
Subsequently, the List of China Agrivoltaics contains a total of 1,678 entries.
Country: China
Province: the province in which the agrivoltaics is located.
City: the city in which the Agrivoltaics is located, based on the longitude/latitude point location to avoid multi-city errors;
Start Year (Year): year in which the agrivoltaics started operating;
Agrivoltaics Type (Type): These sites use the land between panel rows, under arrays, and surrounding arrays for agricultural purposes (i.e., crop production or grazing) and/or ecosystem services (e.g., forest). This dataset classifies agrivoltaics into 7 types, including crops, greenhouse, Chinese traditional medicine, tea, husbandry, fishery, and forest (Table 2).
English Project Name (Name_EN): English name of agrivoltaics projects;
Chinese Project Name (Name_CN): Chinese name of agrivoltaics projects;
Capacity: facility capacity in MW AC;
Link_ID: A unique, stable identifier assigned to each land parcel to link the agrivoltaics installations situated within it.
The spatial positional accuracy of the China Agrivoltaics Map was evaluated, with the agrivoltaics data stratified by provincial-level administrative regions. A total of 350 agrivoltaics projects were randomly selected as the validation dataset, with the sample size for each stratum provided in Fig. 6 and https://doi.org/10.57760/sciencedb.2624015.
The Sampling Allocation Across Provinces in China.
Sentinel-2 imagery from the Google Earth Engine (GEE) platform, which offers a 10-meter spatial resolution, was utilized. Annual composite remote sensing images of China for 2023 (January to December) were generated. To address minor cloud cover issues, verification of agrivoltaics project identification results was conducted using the 2023 World Imagery Basemap (WIB, available at Wayback.maptiles.arcgis.com) published by ESRI. All samples were manually identified by two students and researchers using both GEE and WIB platforms (Fig. 7).
Agrivoltaics validated via visual examination.
The confusion matrix is used to evaluate the effect of agricultural photovoltaic identification, and the calculation formulas for performance indicators are as follows:
Where, TP (True Positive) is the number of cases that are actually positive and predicted as positive. FP (False Positive) is the number of cases that are actually negative but predicted as positive (false alarm). TN (True Negative) is the number of cases that are actually negative and predicted as negative. FN (False Negative) is the number of cases that are actually positive but predicted as negative (missed detection). As the results, the accuracy, precision, recall and F1 score of agrivoltaics identification are 89.71%, 80.48%, 89.71%, and 84.85%, respectively (Fig. 8).
The confusion matrix of agrivoltaics projects validation.
As noted above, the agrivoltaics data from different sources is incomplete, but it is expected to contain most facilities. For some missing records in the “China Agrivoltaics Map”, we checked and searched via GEM Wiki, which describes the details of the whole projects.
Users can employ China Agrivoltaics Map in (1) analysis the installation of agrivoltaics; (2) analysis the spatial and temporal distribution of agrivoltaics; (3) evaluating the mitigation and adaptation contributions of agrivoltaics development; (4) verify the consistency between agrivoltaics and land management policies; (5) assess the impacts of agrivoltaics to food production. This analysis can be processed via GIS software.
The China Agrivoltaics Map contains the agrivoltaics projects before December 2022. Users should be aware that although great efforts were made to validate the locations and type through Sentinel imagery and visual examination, none of the facilities are field verified or real data from government.
The dataset is available at https://doi.org/10.57760/sciencedb.26240.
The code is available at https://doi.org/10.57760/sciencedb.26240.
IEA (International Energy Agency Photovoltaic Power Systems Programme). Snapshot of Global PV Markets 2025. https://iea-pvps.org/snapshot-reports/snapshot-2025/. https://doi.org/10.69766/PBHV9141 (2025).
Liu, J., Wang, J., Li, L. Vectorized solar photovoltaic installation dataset across China in 2015 and 2020. Sci. Data 11, 1446, https://doi.org/10.1038/s41597-024-04356-z. Zenodo https://zenodo.org/records/14292571 (2024).
National Energy Administration, Construction Situation of Photovoltaic Power Generation Projects in 2024. https://www.nea.gov.cn (2024).
Kruitwagen, L. et al. A global inventory of photovoltaic solar energy generating units. Nature 598, 604–610. https://doi.org/10.1038/s41586-021-03957-7. Zenodo https://zenodo.org/record/5005868 (2021).
Suri, M. et al. Global Photovoltaic Power Potential by Country (English). Energy Sector Management Assistance Program (ESMAP) Washington, D.C.: World Bank Group, http://documents.worldbank.org/curated/en/466331592817725242/Global-Photovoltaic-Power-Potential-by-Country (2020).
Feng, Q. L. L. et al. A 10-m national-scale map of ground-mounted photovoltaic power stations in China of 2020. Sci. Data 11, 198, https://doi.org/10.1038/s41597-024-02994-x. ScienceDB https://doi.org/10.57760/sciencedb.o00121.00001 (2024).
Chen, Y. H., Zhou, J. Y., Ge, Y. & Dong, J. W. Uncovering the rapid expansion of photovoltaic power plants in China from 2010 to 2022 using satellite data and deep learning. Remote Sensing of Environment. 305, 114100 (2024).
Article Google Scholar
Widmer, J., Christ, B., Grenz, J. & Norgrove, L. Agrivoltaics, a promising new tool for electricity and food production: A systematic review. Renewable and Sustainable Energy Reviews 192, 114277, https://doi.org/10.1016/j.rser.2023.114277 (2024).
Article Google Scholar
NREL, Solar Resource Maps and Data. Solar Resource Maps and Data | Geospatial Data Science | NREL (2024).
Global Infrastructure Review, Solar-Powered Farms in Asia: The Rise of Agrivoltaics for Food and Energy, Solar-Powered Farms in Asia: The Rise of Agrivoltaics for Food and Energy (2024).
WRI, New Opportunities for the Integration and Development of “Belt and Road” Photovoltaics + Industries. https://wri.org.cn/insights/Review-of-the-Renewable-Energy-Supporting-Belt-and-Road-Initiative-Countries-Green-Development-Series-Salon-4 (2024).
Global Energy Monitor. Global solar power tracker. https://globalenergymonitor.org/projects/global-solar-power-tracker/ (2024).
Yang, J. & Huang, X. The 30 m annual land cover datasets and its dynamics in China from 1985 to 2023. Zenono. https://zenodo.org/records/12779975 (2024).
Peng, Y. et al. Where is tea grown in the world: A robust mapping framework for agroforestry crop with knowledge graph and sentinel’s images. Remote Sensing of Environment, 303, https://doi.org/10.1016/j.rse.2024.114016 (2024).
Zhang, X. & Ma, X. Vectorized Agrivoltaics Dataset in China[DS/OL]. V2. Science Data Bank. https://doi.org/10.57760/sciencedb.26240 (2025).
Download references
This work was jointly supported by the National Key R&D Program of China (2023YFF0805904), National Natural Science Foundation of China (No. 32271638 and 32171561), the Central Public-interest Scientific Institution Basal Research Fund (No. BSRF202502), and the Low Carbon Science Center of the Agricultural Science and Technology Innovation Program (ASTIP—CAAS).
Key Laboratory of Land Surface Pattern and Simulation, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, 11A Datun Rd, Beijing, 100101, China
Xueyan Zhang
Institute of Environment & Sustainable Development in Agriculture, Chinese Academy of Agricultural Sciences, Beijing, 100081, P.R. China
Xin Ma
PubMed Google Scholar
PubMed Google Scholar
Xueyan Zhang: conceptualization, investigation, data curation, formal analysis, writing – original draft, writing – review & editing; Xin Ma: conceptualization, writing – review & editing, supervision.
Correspondence to Xin Ma.
The authors declare no competing interests.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissions
Zhang, X., Ma, X. Vectorized Agrivoltaics Dataset in China from 2010 to 2022. Sci Data 13, 116 (2026). https://doi.org/10.1038/s41597-025-06305-w
Download citation
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s41597-025-06305-w
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Advertisement
Scientific Data (Sci Data)
ISSN 2052-4463 (online)
© 2026 Springer Nature Limited
Sign up for the Nature Briefing: Anthropocene newsletter — what matters in anthropocene research, free to your inbox weekly.
