Abstract：
The ecosystem gross primary productivity (GPP) is crucial to land-atmosphere carbon exchanges, and changes in global GPP as well as its influencing factors have been well studied in recent years. However, identifying the spatio-temporal variations of global GPP under future climate changes is still a challenging issue. This study aims to develop data-driven approach for predicting the global GPP as well as its monthly and annual variations up to the year 2100 under changing climate. Specifically, Catboost was employed to examine the potential relationship between the GPP and environmental factors, with climate variables, CO2 concentration and terrain attributes being selected as environmental factors. The predicted monthly and annual GPP from Coupled Model Intercomparison Project phase 6 (CMIP6) under future SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5 scenarios were analyzed. The results indicate that the global GPP is predicted to increase under the future climate change in the 21st century. The annual GPP is expected to be 115.122 Pg C, 116.537 Pg C, 117.626 Pg C, and 120.097 Pg C in 2100 under four future scenarios, and the predicted monthly GPP shows seasonal difference. Meanwhile, GPP tends to increase in the northern mid-high latitude regions and decrease in the equatorial regions. For the climate zones form Köppen-Geiger classification, the arid, cold, and polar zones present increased GPP, while GPP in the tropical zone will decrease in the future. Moreover, the high importance of climate variables in GPP prediction illustrates that the future climate change is the main driver of the global GPP dynamics. This study provides a basis for predicting how global GPP responds to future climate change in the coming decades, which contribute to understanding the interactions between vegetation and climate.
摘要：
生态系统总初级生产力(3GPP)对陆地-大气碳交换至关重要，近年来，全球阵的变化及其影响因素得到了很好的研究。然而,在未来的气候变化下，确定全球的时空变化仍然是一个具有挑战性的问题。这项研究旨在开发数据驱动的方法，以预测气候变化下到2100年的全球3GPP及其月度和年度变化。具体来说,Catboost被用来检查3GPP和环境因素之间的潜在关系，随着气候变量，选择CO2浓度和地形属性作为环境因素。分析了未来SSP1-2.6，SSP2-4.5，SSP3-7.0和SSP5-8.5情景下耦合模型比对项目第6阶段（CMIP6）的预测月度和年度3GPP。结果表明，在21世纪未来气候变化的背景下，全球GP1预计将增加。预计年度3GPP为115.122PgC，116.537PgC,117.626PgC,在四种未来情景下，2100年为120.097PgC，而预测的月度3GPP显示出季节性差异。同时,在北部中高纬度地区，PPI趋于增加，而在赤道地区则趋于减少。对于柯本-盖革分类的气候区，干旱的，冷,极地区呈现增加的3GPP，而在未来的热带地区将会减少。此外，气候变量在气候预测中的重要性表明，未来的气候变化是全球气候动力学的主要驱动因素。这项研究为预测未来几十年全球3GPP如何应对未来气候变化提供了基础。这有助于理解植被和气候之间的相互作用。
