The increasing demand for high-quality, temporally consistent satellite imagery has highlighted gaps in data processing resources, particularly in China. Unlike the United States, which has a preprocessed dataset (ARD) for Landsat imagery, Chinese researchers have lacked such an efficient resource. This has led to time-consuming processes, limiting the accuracy and scope of their research. The new Landsat composite data cube fills this gap, offering a reliable, seamless dataset for environmental monitoring and land use studies, which is crucial for large-scale analyses and informed decision-making. Based on these challenges, there is a pressing need to develop such comprehensive and easily accessible datasets.
The research team, led by Yaotong Cai and colleagues, published (DOI: 10.34133/remotesensing.0698) their findings in Journal of Remote Sensing on July 2, 2025. They present the first-ever seamless, annual Leaf-On Landsat composite data cube for China, covering the years 1985 to 2023. This data cube addresses major challenges such as cloud contamination, sensor calibration differences, and data gaps that hindered previous efforts. The newly developed composite dataset simplifies satellite data processing, making it an invaluable tool for monitoring vegetation dynamics and supporting land use policies.
The study introduces an innovative Landsat data cube, which aggregates images from multiple Landsat sensors. The team utilized advanced techniques like medoid compositing and gap-filling to generate high-quality, cloud-free composites. By leveraging segmented linear interpolation, they effectively addressed data gaps caused by cloud cover and sensor failures. The dataset spans 39 years, capturing vegetation dynamics across diverse regions of China. Its high temporal consistency and improved spectral fidelity make it a valuable resource for environmental assessments and long-term land use research.
The research team employed a systematic approach to generate the dataset, starting with Landsat imagery from 1985 to 2023. They used surface reflectance data from Landsat 4/5, 7, and 8/9, all processed through Google Earth Engine for consistency. Cloud and shadow contamination were removed using a quality assessment (QA) band, and data from different sensors were harmonized to reduce discrepancies. The core innovation lies in the medoid compositing method, which selects the most representative pixel each year, minimizing outliers and maintaining the integrity of the data. This method significantly reduces noise and enhances temporal consistency. To address missing data, the team applied gap-filling techniques, using segmented linear interpolation to seamlessly integrate data across years. The result is a comprehensive and reliable dataset that will serve as a foundation for future remote sensing studies.
Dr. Yaotong Cai, lead author, commented, "This dataset is a significant breakthrough for environmental monitoring in China. It not only simplifies satellite data processing but also provides a long-term resource for research on land use, climate change, and biodiversity conservation. Our methodology offers a robust solution for handling the challenges posed by cloud cover and sensor inconsistencies, and we hope it will drive future research."
The study used Landsat imagery processed into surface reflectance using the Land Surface Reflectance Code (LaSRC). The medoid compositing method was employed to select the most representative pixel for each year, while data gaps were filled using a segmented linear interpolation algorithm. The study applied rigorous cloud and shadow masking, ensuring that the composite images were predominantly clear-sky, enhancing data accuracy. The dataset was assessed for consistency using correlation coefficients, ensuring the quality and reliability of the final product.
The newly developed dataset is poised to transform environmental research and land use management in China. Its applications include monitoring forest cover, assessing the effects of climate change, and supporting biodiversity conservation efforts. Future work will focus on refining the cloud and shadow masking algorithms, integrating additional satellite data sources for enhanced coverage, and expanding the dataset to include leaf-off periods. This ongoing improvement will further increase its usefulness in real-time environmental monitoring and global climate studies.
###
References
DOI
10.34133/remotesensing.0698
Original Source URL
https://spj.science.org/doi/10.34133/remotesensing.0698
Funding information
This work was supported in part by the National Science Foundation for Distinguished Young Scholars of China under Grant 42225107; in part by the National Key Research and Development Program of China under Grant 2022YFB3903402; and in part by the National Natural Science Foundation of China under Grant 42171409 and Grant 42171410.
About Journal of Remote Sensing
The Journal of Remote Sensing, an online-only Open Access journal published in association with AIR-CAS, promotes the theory, science, and technology of remote sensing, as well as interdisciplinary research within earth and information science.