About

Dr. Xiaopeng Song is an Assistant Professor in the Department of Geographical Sciences at the University of Maryland. His research combines satellite observations, machine learning/AI, spatial analysis, and interdisciplinary approaches to examine human-induced land system changes. His work has broad implications for issues such as food and energy security, and climate change. Dr. Song has published 60 peer-reviewed articles in prominent journals including Nature, Nature Climate Change, Nature Food, Nature Sustainability, Proceedings of the National Academy of Sciences, Remote Sensing of Environment, and Science Advances. His research is supported by organizations such as NASA, USGS, the World Resources Institute, and Google. He has served as a Co-Investigator of the Landsat Science Team and contributed to the development of land cover and change map accuracy assessment protocols. Additionally, Dr. Song has served as an area editor for the Land Surface Type volume of the book Comprehensive Remote Sensing and co-edited a special issue on Land Cover and Land Use Change in Science of Remote Sensing. He has delivered invited lectures to various organizations and professional societies, and has contributed to public media outlets including BBC, NPR, and Reuters. His academic background includes a PhD in Geographical Sciences from the University of Maryland and a BS in Geographical Information Science and Economics from Peking University.

Research topics

• Geography
• Environmental science
• Computer Science
• Ecology
• Biology
• Agroforestry
• Artificial Intelligence
• Physical geography
• Demography
• Climatology
• Geology
• Statistics
• Soil science
• Mathematics
• Remote sensing

Selected publications

• Quantifying tropical tree cover loss driven by cropland conversion

  2026-03-14

  articleOpen access1st authorCorresponding

  Cropland conversion is a significant driver of tropical tree cover loss. Recent research has mapped tree cover loss as well as cropland expansion in separate themes using satellite data, but the amount of tree cover loss driven by cropland expansion and the spatiotemporal dynamics are less well understood. The objective of this research is to conduct mapping and quantitative assessments of tree cover to cropland conversion using a combination of satellite-derived land cover change products and design-based inference. We combined the University of Maryland global tree cover loss and global cropland gain maps, both at 30 m resolution, to generate tree cover-to-cropland conversion over five epochs between 2004 and 2023. We employed this new set of maps to aid selection of a stratified random sample for area estimation. We used high-resolution images in Google Earth and time series of Landsat images to interpret the land cover and land use change for each sample unit. This sample allowed us to estimate the area of tree cover-to-cropland conversion over the five epochs and subsequently the temporal trends. Our results suggested that tropical tree cover-to-cropland conversion reached a total of 23.1 ± 2.1 Mha between 2004 and 2023. These results are important for understanding the socioeconomic drivers of deforestation in the tropics.

  PublisherDOI

• Structural changes in the wheat supply chain and their environmental footprints through satellite yield forecasting

  Cleaner and Responsible Consumption · 2026-03-21

  articleOpen access

  Satellite-based crop forecasting in major exporting regions can trigger production shifts in the opposite hemisphere by leveraging seasonal disparities between the Northern and Southern Hemispheres to stabilize global crop supply and markets. However, the impact of these regional shifts on supply chains and environmental footprint remains unclear. Thus, evaluating the influence of yield forecasts on wheat supply chains and carbon footprints and identifying challenges to and strategies for the sustainable use of satellite-based forecasts are crucial. This study analyzed the structural changes in the wheat supply chain and their environmental impacts resulting from satellite yield forecasting by employing the RAS method and wheat production data. The results indicate that Southern Hemisphere wheat production shifts, triggered by favorable Northern Hemisphere crop information, generate an additional $10.80 billion in value-added and 4.08 Mt of carbon dioxide emissions in a favorable crop year relative to the 2019 base year. Moreover, favorable Northern Hemisphere harvests increase carbon dioxide emissions in countries neighboring Russia and Ukraine, including Turkey, Lebanon, and Egypt. Considering carbon leakage in production forecasts for trading agricultural commodities across hemispheres can stabilize prices and reduce emissions, ultimately improving global food security and environmental sustainability. • Satellite forecasts trigger wheat shifts in the Northern and Southern Hemispheres. • Yield forecasts impact supply chains and carbon footprints. • Wheat shifts add $10.80 billion value and 4.08 Mt CO 2 emissions. • Favorable harvests raise CO 2 emissions in Turkey, Lebanon, and Egypt. • Carbon-aware forecasts aid food security and sustainability.

  PublisherDOI

• An accurate 10 m annual crop map product of maize and soybean across the United States

  Earth system science data · 2026-03-25 · 1 citations

  articleOpen access

  Abstract. High-resolution crop maps over large spatial extents are fundamental to many agricultural applications; however, generating high-quality crop maps consistently across space and time remains a challenge. In this study, we improved a workflow for crop mapping and developed an openly available, annual, 10 m spatial resolution maize and soybean map product over the Contiguous United States (CONUS) from 2019 to 2022 (available at https://glad.umd.edu/dataset/mapping-crops-10-m-resolution-united-states, last access: 26 December 2025). We obtained all available Sentinel-2 surface reflectance data between May and October for every year, applied quality assurance, corrected the bidirectional reflectance distribution function (BRDF) effects, and generated 10 d analysis ready data (ARD) composites. We then derived multi-temporal metrics from the 10 d ARD as training features for the national-scale wall-to-wall mapping. We implemented a stratified, two-stage cluster sampling, and then conducted annual field surveys and collected ground data. Utilizing the training data with Sentinel-2 multi-temporal metrics and topographic factors, we trained random forest models generalized for annual maize and soybean classification separately. Validated using field data from the two-stage cluster sample, our annual maps achieved consistent overall accuracies (OA) greater than 95 % with standard errors of less than 1 %. User's accuracies (UAs) and producer's accuracies (PAs) for maize were higher than 91 % and 84 % across the years, and UAs and PAs for soybean were greater than 88 % and 82 %, respectively. To illustrate the substantial improvement of the 10 m map over existing datasets, e.g., the 30 m Cropland Data Layer (CDL), we aggregated the 10 m maps to 30 m spatial resolution and quantified the number of mixed pixels that can be reduced by improving the mapping from 30 to 10 m. The counties with the most maize and soybean production in Iowa, Illinois and Nebraska had the lowest reduction in mixed pixels, ranging from 1 % to 7 %, whereas southern counties had a higher reduction in mixed pixels. Overall, the median percentages of mixed maize and soybean pixels reduction across all counties were 8 % and 9 %, respectively. With more Sentinel-2-like data available from continuous observations and incoming satellite missions, we anticipate that 10 m crop maps will greatly benefit long-term monitoring for agricultural practices from the field to global scales. The dataset is also available at https://doi.org/10.6084/m9.figshare.28934993.v2 (Li et al., 2025).

  PublisherDOI

• Assessing the Influence of Extreme Climate Events on Soybean Yield Losses in Brazil and Argentina from 2001 to 2022

  SSRN Electronic Journal · 2025-01-01

  preprintOpen access
  PublisherDOI

• Supplementary material to "HIStory of LAND transformation by humans in South America (HISLAND-SA): annual and 1-km crop-specific gridded data (1950–2020)"

  2025-01-27

  preprintOpen access
  PublisherDOI

• Open cotton boll detection using LiDAR point clouds and RGB images from unmanned aerial systems

  Current Plant Biology · 2025-07-17 · 2 citations

  articleOpen accessSenior author

  Accurate quantification of open bolls and their distribution is crucial for understanding cotton growth, development, and yield in optimized crop management and enhanced plant breeding. Manual boll counting methods are time-consuming, labor-intensive, and subjective. Leveraging the potential of high-resolution images for high-throughput phenotyping offers a promising avenue for efficient trait quantification. The objectives of this study were to develop methods to detect and count open cotton bolls using LiDAR point cloud and RGB images and to compare the effectiveness of these two data sources. A DJI Phantom 4 RTK Unmanned Aerial System (UAS) equipped with a 4 K RGB camera was used to acquire high-resolution RGB images, and a DJI Matrice 300 RTK with a Zenmuse L1 sensor was used to acquire LiDAR point cloud data. The RGB images were converted to point cloud using photogrammetry by measuring multiple points of overlapping images. The boll detection workflow involved data filtering and clustering using the density-based spatial clustering of applications with noise (DBSCAN) method. Evaluation of the methods involved 48 plots representing small, medium, and large plant sizes using metrics including mean absolute percentage error (MAPE), root mean square error (RMSE), and coefficient of determination (r²). The methods using both data sources performed well in estimating open bolls, with LiDAR point cloud data slightly outperforming those derived from RGB images. Generally, the performance of the DBSCAN method in boll detection improved with decreasing plant sizes. Specifically, LiDAR data yielded MAPE values of 5.03 %, 8.05 %, and 13.46 %, RMSE values of 7.26, 14.33, and 23.40 bolls per m², and r 2 values of 0.93, 0.84, and 0.84 for small, medium, and large plant sizes, respectively. RGB image-based data exhibited MAPE values of 7.21 %, 6.49 %, and 16.41 %, RMSE values of 11.05, 13.66, and 26.49 bolls per m², and r 2 values of 0.82, 0.74, and 0.83 for small, medium, and large plant sizes, respectively. The method demonstrates the potential of RGB imagery and LiDAR data for estimating boll counts, offering valuable tools for enhanced plant phenotyping in plant breeding and site-specific crop management. Both data sources underestimated boll counts, with smaller plants showing less undercounting, likely due to improved light penetration and separation of bolls. These findings highlight the influence of plant structure on boll detection accuracy and the need to address challenges posed by dense canopies to enhance detection reliability. • The study developed methods to count open cotton bolls using LiDAR point clouds and RGB images. • LiDAR slightly outperformed the RGB image-derived point cloud, with better accuracy for smaller plants due to less canopy density. • Dense canopies reduced detection accuracy, highlighting the influence of plant structure.

  PublisherDOI

• HIStory of LAND transformation by humans in South America (HISLAND-SA): annual and 1-km crop-specific gridded data (1950–2020)

  2025-01-27

  preprintOpen access

  Abstract. South America is a global hotspot for land use and land cover (LULC) change, marked by dramatic agricultural land expansion and deforestation. Developing high-resolution, long-term crop-specific data is essential for gaining a deeper understanding of natural-human interactions and addressing the impacts of human activities on regional biogeochemical, hydrological cycles, and climate. In this study, we integrated multi-source data, including high-resolution remote sensing data, model-based data, and historical agricultural census data, to reconstruct the historical dynamics of four major commodity crops (i.e., soybean, maize, wheat, and rice) in South America at annual time scale and 1 km×1 km spatial resolution from 1950 to 2020. The results showed that cropland in South America has expanded rapidly through encroachment into other vegetation over the past 70 years. Specifically, soybean is one of the most dramatically expanded crops, increasing from essentially zero in 1950 to 48.8 Mha in 2020, resulting in a total loss of 23.92 Mha of other vegetation (i.e., forest, pasture/rangeland, and unmanaged grass/shrubland). In addition, the area of maize increased by a factor of 2.1 from 12.7 Mha in 1950 to 26.9 Mha in 2020, while rice and wheat areas remained relatively stable. The newly developed crop type data provide important insights for assessing the impacts of agricultural land expansion on crop production, greenhouse gas emissions, and carbon and nitrogen cycles in South America. Moreover, these data are instrumental for developing national policies, sustainable trade, investment, and development strategies aimed at securing food supply and other human and environmental objectives in South America and other regions. The datasets are available at https://doi.org/10.5281/zenodo.14002960 (Xu et al., 2024).

  PublisherDOI

• Global annual cropland dynamics 2015 – 2024

  Remote Sensing of Environment · 2025-01-01

  articleOpen access

  Food security worldwide is increasingly threatened by population growth, shifting diets, geopolitical conflicts, and climate change impacts. Annual operational cropland monitoring is required to support the United Nations Zero Hunger Sustainable Development Goal. Landsat satellite data provide a foundation for such global, independent, high-cadence monitoring at 30-m spatial resolution suitable for agricultural policy and management interventions, policy responses, and market adjustments. Here, we used Landsat Analysis Ready Data developed by the Global Land Analysis and Discovery Lab (GLAD-ARD) and machine learning to map global cropland extent annually from 2015 to 2024. We showed that the global cropland area increased by more than 6% over the past decade. By combining sample-based cropland area estimates from our research and the earlier analysis (2003–2019), we estimate that the global cropland area has expanded by nearly 14% since 2003. Between 2015 and 2024, Africa accounted for the largest regional increase (+24.5 Mha). At the national scale, Brazil experienced the largest gain (+16.5 Mha) and Morocco the largest loss (˗0.38 Mha). A third (33.3%) of all new cropland was established through natural vegetation clearing or irrigation expansion within natural drylands. The overall accuracies of the 2015 and 2024 cropland maps were 97.8% (Standard Error 0.3%) and 97.3% (SE 0.4%), respectively. Despite cropland expansion, population growth has outpaced cropland gains; between 2015 and 2024, per-capita cropland area declined from 0.166 to 0.161 ha per person. Our data illustrate the combined effects of changes in land use priorities, climate, water supply, international trade, and armed conflicts on global cropland extent dynamics during the last decade. • Presented global annual cropland maps with overall accuracy >97%. • Global cropland area increased by more than 6% from 2015 to 2024. • Cropland area increased by 14% over the past two decades. • Per-capita cropland area has declined by 3% from 2015 to 2024.

  PublisherDOI

• Effects of igneous contact metamorphism on the mineral and toxic trace element composition of low-sulfur coal in the Huaibei Coalfield, eastern China

  Applied Geochemistry · 2025-01-06 · 2 citations

  article
  PublisherDOI

• HIStory of LAND transformation by humans in South America (HISLAND-SA): annual and 1 km gridded data for soybean, maize, wheat, and rice (1950–2020)

  Earth system science data · 2025-11-24

  articleOpen accessCorresponding

  Abstract. South America is a global hotspot for land use and land cover (LULC) change, marked by dramatic agricultural land expansion and deforestation. While previous studies have documented land use and land cover changes in South America over recent decades, there is still a lack of spatially explicit and time-series maps of crop types that capture shifts in crop distribution. Therefore, developing high-resolution, long-term, and crop-specific datasets is crucial for advancing our understanding of human–environment interactions and for assessing the impacts of agricultural activities on carbon and biogeochemical cycles, biodiversity, and climate. In this study, we integrated multi-source data, including high-resolution remote sensing data, model-based data, and historical agricultural census data, to reconstruct the historical dynamics of four major commodity crops (i.e., soybean, maize, wheat, and rice) in South America at an annual timescale and 1 km × 1 km spatial resolution from 1950 to 2020. The results showed that soybean and maize cultivation expanded rapidly in South America by encroaching on other vegetation (i.e., forest, pasture/rangeland, and unmanaged grass/shrubland) over the past 70 years, whereas wheat and rice areas remained relatively stable. Specifically, soybean is one of the most dramatically expanded crops, increasing from essentially zero in 1950 to 48.8 Mha in 2020, resulting in a total loss of 23.92 Mha of other vegetation. In addition, the area of maize increased by a factor of 2.1 from 12.7 Mha in 1950 to 26.9 Mha in 2020. The newly developed crop type dataset provides important insights for assessing the impacts of cropland expansion on crop production, biodiversity, greenhouse gas emissions, and carbon and nitrogen cycles in South America. Moreover, these data are instrumental for developing national policies, sustainable trade, investment, and development strategies aimed at securing food supply and other human and environmental objectives in South America. The datasets are available at https://doi.org/10.5281/zenodo.14002960 (Xu et al., 2024).

  PublisherDOI

Frequent coauthors

• Matthew C. Hansen

  Humboldt-Universität zu Berlin

  31 shared

• Philippe Ciais

  Laboratoire des Sciences du Climat et de l'Environnement

  27 shared

• Peter Potapov

  26 shared

• Stephen V. Stehman

  State University of New York

  24 shared

• Lei Wang

  Southwest University of Science and Technology

  21 shared

• John Townshend

  16 shared

• Joe Sexton

  TerraMetrics (United States)

  16 shared

• Chengquan Huang

  University of Maryland, College Park

  16 shared

Education

• Ph.D., Geography

  University of Maryland

  2008

• M.S., Geography

  University of Maryland

  2003

• B.S., Geography

  University of Science and Technology of China

  2000

Awards & honors

• Highly Cited Researcher in Cross-Field by Clarivate