Enhanced monitoring of rubber plantations in complex tropical regions: Integrating all Landsat/Sentinel data for precise classification and stand age estimation

The rubber tree (Hevea brasiliensis), extensively cultivated in tropical regions, is a primary source of natural rubber. With increasing global demand for natural rubber, precise monitoring of cultivation dynamics of rubber plantations is crucial for regional economic development and ecosystem assessment. However, the significant spatial variability in rubber plantation phenology, interference from non-rubber deciduous forests, and cloud-free optical imagery limitations create considerable uncertainty in applying existing monitoring methods across varying latitudes. This study leverages extensive field surveys and over 30 years of Landsat/Sentinel-2 data to analyze the land use changes, growth processes in rubber seedling stages, and phenological changes during mature stages of rubber plantations. We developed a novel, high-precision identification and stand age estimation algorithm for rubber plantations on the Google Earth Engine (GEE) cloud platform, integrating multi-temporal remote sensing data, machine learning, and advanced algorithm optimization. Applications of this algorithm in countries with complex terrain and climate, such as Vietnam, Laos, Cambodia, and Myanmar, demonstrate higher robustness and accuracy. Ground data validation shows that the overall classification accuracy exceeds 95%, with an average stand age estimation error of less than two years. Spatial statistical analysis at 30-meter resolution aligns closely with data from authoritative sources like the Association of Natural Rubber Producing Countries (ANRPC), underscoring our method's effectiveness and reliability. Beyond mapping distribution and stand age structures of rubber plantations, this research supports future natural rubber production forecasts, environmental impact assessments, and sustainable policy development. Moreover, this study paves the way for novel applications of remote sensing in the monitoring of agriculture and forestry in tropical areas, setting a foundation for future advancements and innovations in these domains.   Keywords: Rubber plantations, classification, Landsat, Sentinel-2, Google Earth Engine