Impacts of Climate Change and Human Activities on Vegetation NDVI Changes in Henan Province from 2000 to 2020

PENGFEI HOU^1*Yue Wang²Jingxu Wang¹Shike Qiu¹Shuangquan Li¹Hao Wang³Sanjun Yin⁴Du Jun^1*

• ¹Henan Academy of Sciences, Zhengzhou, China
• ²Henan University College of Geography and Environmental Science, Kaifeng, China
• ³Chinese Academy of Surveying and Mapping, Beijing, China
• ⁴Forestry Survey and Planning Institute of Henan Province, Zhengzhou, China

The synergistic impacts of climate change and human activities have profoundly shaped vegetation dynamics, making the elucidation of their underlying driving mechanisms critical for regional ecological conservation and sustainable development. This study investigates these complex interactions in Henan Province, China, by integrating multi-source datasets from 2000 to 2020. We comprehensive analytical framework, which spans from traditional statistical methods to advanced machine learning models (Random Forest and Shapley Additive exPlanations), was employed to systematically decipher the spatiotemporal patterns of NDVI and its intricate driving forces. The results indicate: (1) During the past two decades, the NDVI in Henan Province exhibited a significant upward trend (an average increase of 0.049 per decade), which reflected the continuous improvement in ecological quality. Spatially, high NDVI values were mainly distributed in the mountainous areas in the west and south (Funiu Mountains and Tongbai Mountains), while the low-value areas were concentrated in the Central Plains urban agglomerations, which have shown signs of recovery. (2) Feature importance analysis based on machine learning precisely identified grassland, cropland, and barren land as the dominant drivers regulating the spatial pattern of NDVI, while impervious surfaces exerted the relatively weakest direct influence. (3) The SHAP model further revealed complex nonlinear relationships between key factors and NDVI. For instance, cropland exhibited a pronounced inverted U-shaped pattern, indicating that moderate agricultural activity positively contributes to vegetation cover, while excessive saturation may produce inhibitory effects. Although climatic factors establish the background conditions for vegetation growth, human activities are the primary drivers shaping the current spatiotemporal heterogeneity of NDVI.