基于SHAP-XGBoost模型的黄河三角洲碳储量可解释性分析

Q2 Environmental Science

环境科学 Pub Date : 2026-09-08 DOI:10.13227/j.hjkx.202505105

Zhi-Long Ma, Mei Han, Xiang-Lun Kong, Jin-Xin Sun, Bin Zhang

{"title":"基于SHAP-XGBoost模型的黄河三角洲碳储量可解释性分析","authors":"Zhi-Long Ma, Mei Han, Xiang-Lun Kong, Jin-Xin Sun, Bin Zhang","doi":"10.13227/j.hjkx.202505105","DOIUrl":null,"url":null,"abstract":"Ecosystem carbon storage is an indicator of climate change. Exploring the spatial and temporal evolution characteristics and driving mechanism of carbon storage can provide natural solutions to climate change. Based on the five-period land use data of the Yellow River Delta from 1980 to 2020， the InVEST model was used to analyze the spatial and temporal differentiation characteristics of terrestrial ecosystem carbon storage in the Yellow River Delta. The CA-Markov model was coupled to predict the carbon storage of terrestrial ecosystems in the Yellow River Delta under the two scenarios of natural development and ecological protection in 2030 and 2060. Finally， the SHAP-XGBoost model was used to explain the spatial and temporal evolution of carbon storage. The results showed that： ① From 1980 to 2020， the land use types changed significantly. The construction land increased by 503.40 km2， while the cultivated land and water area lost 123.47 km2 and 313.18 km2， respectively. The proportion of ecological land and unused land such as forest land and grassland was small and showed a downward trend. ② From 1980 to 2020， the carbon stocks were 52.88×106， 76.77×106， 36.92×106， 32.71×106， and 49.94×106 t， respectively. Under the natural development scenario， the carbon stocks in 2030 and 2060 were 41.92×106 t and 41.21×106 t， respectively. Under the ecological protection scenario， the carbon stocks in 2030 and 2060 were 51.24×106 t and 60.33×106 t， respectively， indicating that ecological protection could significantly increase the carbon stocks of terrestrial ecosystems. ③ The main driving forces of the spatial and temporal evolution of carbon storage from 1980 to 2020 were the distance from high speed， distance from railway， and NDVI. The relationship between annual average temperature and carbon storage was more complex， and the overall effect was positive. The results of this study can provide a scientific reference for the restoration and management of terrestrial ecosystems in the Yellow River Delta.","PeriodicalId":35937,"journal":{"name":"环境科学","volume":"47 9","pages":"5848-5862"},"PeriodicalIF":0.0000,"publicationDate":"2026-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"[Interpretability-driven Analysis of Carbon Storage in the Yellow River Delta Based on SHAP-XGBoost Model].\",\"authors\":\"Zhi-Long Ma, Mei Han, Xiang-Lun Kong, Jin-Xin Sun, Bin Zhang\",\"doi\":\"10.13227/j.hjkx.202505105\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Ecosystem carbon storage is an indicator of climate change. Exploring the spatial and temporal evolution characteristics and driving mechanism of carbon storage can provide natural solutions to climate change. Based on the five-period land use data of the Yellow River Delta from 1980 to 2020， the InVEST model was used to analyze the spatial and temporal differentiation characteristics of terrestrial ecosystem carbon storage in the Yellow River Delta. The CA-Markov model was coupled to predict the carbon storage of terrestrial ecosystems in the Yellow River Delta under the two scenarios of natural development and ecological protection in 2030 and 2060. Finally， the SHAP-XGBoost model was used to explain the spatial and temporal evolution of carbon storage. The results showed that： ① From 1980 to 2020， the land use types changed significantly. The construction land increased by 503.40 km2， while the cultivated land and water area lost 123.47 km2 and 313.18 km2， respectively. The proportion of ecological land and unused land such as forest land and grassland was small and showed a downward trend. ② From 1980 to 2020， the carbon stocks were 52.88×106， 76.77×106， 36.92×106， 32.71×106， and 49.94×106 t， respectively. Under the natural development scenario， the carbon stocks in 2030 and 2060 were 41.92×106 t and 41.21×106 t， respectively. Under the ecological protection scenario， the carbon stocks in 2030 and 2060 were 51.24×106 t and 60.33×106 t， respectively， indicating that ecological protection could significantly increase the carbon stocks of terrestrial ecosystems. ③ The main driving forces of the spatial and temporal evolution of carbon storage from 1980 to 2020 were the distance from high speed， distance from railway， and NDVI. The relationship between annual average temperature and carbon storage was more complex， and the overall effect was positive. The results of this study can provide a scientific reference for the restoration and management of terrestrial ecosystems in the Yellow River Delta.\",\"PeriodicalId\":35937,\"journal\":{\"name\":\"环境科学\",\"volume\":\"47 9\",\"pages\":\"5848-5862\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2026-09-08\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"环境科学\",\"FirstCategoryId\":\"1087\",\"ListUrlMain\":\"https://doi.org/10.13227/j.hjkx.202505105\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"Environmental Science\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"环境科学","FirstCategoryId":"1087","ListUrlMain":"https://doi.org/10.13227/j.hjkx.202505105","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"Environmental Science","Score":null,"Total":0}

引用次数: 0

摘要

生态系统碳储量是气候变化的一个指标。探索碳储量的时空演化特征和驱动机制，可以为气候变化提供自然的解决方案。基于黄河三角洲1980—2020年5期土地利用数据，利用InVEST模型分析了黄河三角洲陆地生态系统碳储量的时空分异特征。利用CA-Markov模型耦合预测了2030年和2060年黄河三角洲自然开发和生态保护两种情景下陆地生态系统的碳储量。最后，利用SHAP-XGBoost模型解释了碳储量的时空演变。结果表明：①1980 ~ 2020年，土地利用类型发生显著变化。建设用地增加503.40 km2，耕地和水域面积分别减少123.47 km2和313.18 km2。生态用地和林地、草地等未利用地所占比例较小，且呈下降趋势。②1980 ~ 2020年碳储量分别为52.88×106、76.77×106、36.92×106、32.71×106和49.94×106 t。在自然发展情景下，2030年和2060年碳储量分别为41.92×106 t和41.21×106 t。在生态保护情景下，2030年和2060年的碳储量分别为51.24×106 t和60.33×106 t，说明生态保护可以显著增加陆地生态系统的碳储量。③1980—2020年碳储量时空演变的主要驱动力是高速公路距离、铁路距离和NDVI。年平均温度与碳储量之间的关系较为复杂，总体呈正相关。研究结果可为黄河三角洲陆地生态系统的恢复与管理提供科学参考。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

[Interpretability-driven Analysis of Carbon Storage in the Yellow River Delta Based on SHAP-XGBoost Model].

Ecosystem carbon storage is an indicator of climate change. Exploring the spatial and temporal evolution characteristics and driving mechanism of carbon storage can provide natural solutions to climate change. Based on the five-period land use data of the Yellow River Delta from 1980 to 2020， the InVEST model was used to analyze the spatial and temporal differentiation characteristics of terrestrial ecosystem carbon storage in the Yellow River Delta. The CA-Markov model was coupled to predict the carbon storage of terrestrial ecosystems in the Yellow River Delta under the two scenarios of natural development and ecological protection in 2030 and 2060. Finally， the SHAP-XGBoost model was used to explain the spatial and temporal evolution of carbon storage. The results showed that： ① From 1980 to 2020， the land use types changed significantly. The construction land increased by 503.40 km²， while the cultivated land and water area lost 123.47 km² and 313.18 km²， respectively. The proportion of ecological land and unused land such as forest land and grassland was small and showed a downward trend. ② From 1980 to 2020， the carbon stocks were 52.88×10⁶， 76.77×10⁶， 36.92×10⁶， 32.71×10⁶， and 49.94×10⁶ t， respectively. Under the natural development scenario， the carbon stocks in 2030 and 2060 were 41.92×10⁶ t and 41.21×10⁶ t， respectively. Under the ecological protection scenario， the carbon stocks in 2030 and 2060 were 51.24×10⁶ t and 60.33×10⁶ t， respectively， indicating that ecological protection could significantly increase the carbon stocks of terrestrial ecosystems. ③ The main driving forces of the spatial and temporal evolution of carbon storage from 1980 to 2020 were the distance from high speed， distance from railway， and NDVI. The relationship between annual average temperature and carbon storage was more complex， and the overall effect was positive. The results of this study can provide a scientific reference for the restoration and management of terrestrial ecosystems in the Yellow River Delta.