MaxEnt with remote sensing for tea plantation suitability under climate change.

Premium tea cultivation is highly vulnerable to climate change, yet its future suitability remains insufficiently understood. In this study, we integrated spatially de-biased tea occurrence records derived from Gaofen-6 imagery and a U-Net deep learning framework with the MaxEnt model to project tea suitability in Nanping, Southeastern China, under multiple CMIP6 climate scenarios from 2021 to 2080. Random forest was used to cross-check model robustness. The results show that precipitation seasonality and precipitation of the wettest quarter are the main climatic drivers of tea suitability, and future warming is likely to shift highly suitable areas southward while increasing spatial fragmentation under high-emission scenarios. These findings provide a transferable framework for evaluating climate-sensitive high-value crops and support more adaptive agricultural planning under global environmental change.