东北林业大学学报2026，Vol.54Issue(6)：112-124,133,14.
应用广义加性模型和广义加性混合模型的云南省归一化植被指数时空变化驱动机制研究
Spatiotemporal Variation and Driving Mechanism of NDVI in Yunnan Province Based on Generalized Additive Models and Generalized Additive Mixed Models
摘要
Abstract
To explore the driving mechanism of spatiotemporal variations in the normalized difference vegetation index(NDVI)in Yunnan Province,quantitative analysis of spatiotemporal vegetation dynamics was conducted using Theil-Sen Median trend analysis combined with Mann-Kendall significance test.Generalized additive models(GAM)and generalized addi-tive mixed models(GAMM)were introduced to quantitatively analyze the relationships between NDVI and climatic,topo-graphic,and socioeconomic factors,with random effects set at county level,land-use type,and their nested structure.The results showed that:(1)From 2000 to 2020,vegetation coverage in Yunnan Province exhibited an overall significant im-proving trend,while localized degradation and high fluctuation occurred in core urbanized areas.(2)All environmental factors showed highly significant nonlinear relationships with NDVI(P＜0.001).Among them,land surface temperature and annual mean temperature were dominant climatic factors with threshold effects(F=337.880,P＜0.001.F=306.081,P＜0.001),whereas population and GDP density displayed significant negative driving effects.(3)The introduction of different random effect structures(county,land-use type,and their nesting)effectively disentangled the driving mecha-nisms of influencing factors.The influence of annual precipitation was more manifested as a regional background effect,and its local explanatory power weakened after controlling for county differences.Population density exhibited a stable neg-ative effect independent of land-use type,and its local impact became more prominent after controlling for regional varia-tions.The effect of aspect became insignificant after incorporating the most complex nested effects(F=2.882,P=0.089),indicating that its influence was absorbed by higher-order geographic and land-use contexts.The results verify the superiority of the GAMM in analyzing geographically hierarchical data.关键词
归一化植被指数/广义加性模型/广义加性混合模型/非线性响应/云南省Key words
Normalized difference vegetation index(NDVI)/Generalized additive model(GAM)/Generalized ad-ditive mixed model(GAMM)/Nonlinear response/Yunnan Province分类
农业科技引用本文复制引用
闫爽,殷晓洁,王婧,何雨佳,叶江霞..应用广义加性模型和广义加性混合模型的云南省归一化植被指数时空变化驱动机制研究[J].东北林业大学学报,2026,54(6):112-124,133,14.基金项目
云南省重点研发计划项目(202503AP140004) （202503AP140004）
云南省"兴滇英才支持计划"青年人才项目(XDYC-QNRC-2022-0251) （XDYC-QNRC-2022-0251）
云南省基础研究专项项目(202401AT070294). （202401AT070294）
