摘要
净初级生产力(NPP)是衡量生态系统健康的重要指标,直接影响全球碳循环和气候调节.在全球气候变化背景下,粮食主产区净初级生产力的时空变化及驱动因子的研究较少,故选择粮食产量大省河南省和黑龙江省为研究区,探究两省2001~2020年NPP时空变化和驱动因子的差异性,采用多种机器学习算法估算NPP并模拟共享社会经济路径(SSPs)未来气候情景下两省NPP的演变趋势.结果表明:①相比于单一算法,将AdaBoost和XGBoost算法的预测结果加权融合可以更准确地估算两省NPP.②2001~2020年河南省NPP呈增加趋势,空间上西部和南部高,北部低;该时间段内黑龙江省NPP也呈增加趋势,高值区出现在北部、中部和南部的林区,低值区出现在西部城镇化率高的地区.③河南省NPP变化最重要的气候驱动因子是降水,且较低的温度和较弱的潜在蒸散发的交互作用能增强降水对NPP的正向作用;黑龙江省NPP变化最重要的驱动因子是潜在蒸散发,较弱的潜在蒸散发和较低的年平均温度有助于黑龙江省NPP的增加.④SSP245和SSP585情景下两省NPP均呈减少态势,且SSP585情景下两省NPP减少的幅度更大,这与温度的持续升高显著相关,未来应采取人为手段积极应对NPP减少带来的负面影响.
Net primary productivity(NPP)is a vital indicator of ecosystem health,directly influencing the global carbon cycle and climate regulation,and is closely related to regional food production.In the context of global climate change,research on the spatiotemporal changes and driving factors of NPP in major grain-producing areas is limited.Therefore,this study selected Henan Province and Heilongjiang Province,two significant grain-producing provinces,to explore the differences in NPP changes and their driving factors from 2001 to 2020.Various machine learning algorithms were employed to estimate NPP and simulate its evolutionary trends under future climate scenarios based on shared socioeconomic pathways(SSPs).The results follow:①Compared with the single algorithm,the weighted fusion of the prediction results of the AdaBoost and XGBoost algorithms can more accurately estimate the NPP of the two provinces.This approach significantly reduces the root mean square error and improves the goodness of fit.②From 2001 to 2020,NPP in Henan Province showed an increasing trend,with higher values in the west and south and lower values in the north.Similarly,during this period,NPP in Heilongjiang Province also exhibited an upward trend,with high-value areas located in the northern,central,and southern forest regions and low-value areas in the western regions characterized by high urbanization rates.③In Henan Province,precipitation emerged as the most important climate driver affecting changes in NPP.The interaction between lower temperatures and reduced potential evapotranspiration enhances the positive impact of precipitation on NPP.In contrast,potential evapotranspiration is the primary driving factor for NPP changes in Heilongjiang Province,where lower levels of potential evapotranspiration and annual mean temperature contribute to increased NPP.④Under the SSP245 and SSP585 scenarios,NPP in both provinces is projected to decline,with a more significant reduction under the SSP585 scenario.The continuous increase in temperature poses a substantial threat to NPP in both provinces.Therefore,proactive measures should be taken to mitigate the negative impacts of declining NPP in the future.
作者
解德威
刘健
魏俐宏
丁馨
郑昭佩
XIE De-wei;LIU Jian;WEI Li-hong;DING Xin;ZHENG Zhao-pei(College of Geography and Environment,Shandong Normal University,Jinan 250358,China;Institute of Disaster Risk Science,Faculty of Geographical Science,Beijing Normal University,Beijing 100875,China)
出处
《环境科学》
北大核心
2025年第10期6356-6365,共10页
Environmental Science
基金
国家社会科学基金项目(21BGL026)。
关键词
机器学习
模型融合
净初级生产力(NPP)
共享社会经济路径
产粮大省
machine learning
model fusion
net primary productivity(NPP)
shared socioeconomic pathways
major grain-producing province
