首页 > 优先出版
PDF HTML阅读 XML下载 导出引用 引用提醒
基于随机森林的黑土地尺度耕地土壤容重传递函数构建
作者:
作者单位:

1.南京林业大学 南方现代林业协同创新中心，南京210037;2.土壤与农业可持续发展国家重点实验室（中国科学院南京土壤研究所），南京 210008;3.中国科学院大学 现代农业科学学院，北京 100049

中图分类号:

S152.4

基金项目:

国家重点研发计划项目（2021YFD1500102）、中国科学院战略性先导科技专项（XDA28010102）和国家自然科学基金（42107145）


Construction of pedotransfer function for predicting soil bulk density in cultivated land of Northeast China using random forest
Author:
Affiliation:

1.Co‑Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing 210037, China;2.State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, Nanjing 210008, China;3.College of Advanced Agricultural Sciences, University of Chinese Academy of Sciences, Beijing 100049, China

Fund Project:

National Key R&D Program of China (No. 2021YFD1500102), Strategic Priority Research Program of the Chinese Academy of Sciences (No. XDA28010102), and the National Natural Science Foundation of China (No. 42107145)

  • 摘要
    • | |
  • 访问统计
    • |
  • 参考文献
    • |
  • 相似文献
    • |
  • 引证文献
    • | |
  • 文章评论
    摘要:

    土壤容重（Bulk Density，BD）信息对了解东北黑土地耕地土壤物理状况及推进黑土地利用与保护至关重要。传统环刀方法在容重采样环节耗时耗力，难以获得大空间尺度容重数据。土壤传递函数（Pedo-transfer function，PTF）可以利用容易获取的土壤变量实现容重信息估算，然而当前缺少针对东北黑土地尺度的PTF模型研究，而且用于PTF模型构建的潜在土壤属性变量的重要性尚待揭示。本研究针对东北黑土地尺度，基于实测容重数据，使用随机森林（Random forest，RF）机器学习方法，以土壤有机质（Soil organic matter，SOM）、含水量（Moisture content，MC）和土壤质地相关变量作为输入变量，构建面向大尺度耕地容重预测的PTF模型，并分析土壤属性变量的重要性，同时评估国内外已发表PTF模型在东北黑土地耕地容重预测中的适用性。结果显示：已发表PTF模型在表层、亚表层以及全部土壤容重预测中的最优R2分别为0.17、0.22和0.26，RMSE分别为0.16、0.13和0.15 g·cm-3；本研究中基于RF方法的PTF模型在表层、亚表层以及全部土壤中的最优预测R2分别为0.22、0.45和0.37，RMSE分别为0.16、0.11和0.14 g·cm-3。研究表明：已发表PTF模型的容重预测精度较低，难以用于东北黑土地尺度的容重预测；基于RF开发的模型具有预测东北黑土地容重的潜力；有机质是预测东北黑土地容重的最重要变量，其次是含水量，土壤质地相关变量影响较小。

    Abstract:

    【Objective】Soil bulk density (BD) is crucial for understanding the physical condition of black soil in cultivated land of Northeast China and advancing its utilization and protection. The traditional cutting ring method for determining BD is time-consuming and laborious, making the evaluation of BD on a large spatial scale difficult. The pedo-transfer function (PTF) can estimate BD information using readily available soil variables. However, there is currently a lack of research on PTF models specifically targeting the whole of Northeast China, and the importance of potential soil attribute variables for PTF model construction remains to be elucidated.【Method】By incorporating soil organic matter (SOM), moisture content (MC), and soil texture-related variables as input features, we constructed PTF models capable of predicting BD on a large scale. Furthermore, we delved into the significance of these soil attribute variables in the constructed PTF models. Additionally, we assessed the suitability of existing published PTF models for BD prediction in the black soil of Northeast China.【Result】The optimal predicted R2 values of published PTF were 0.17, 0.22, and 0.26, respectively, for the topsoil, subsoil, and all soil samples, and RMSE were 0.16, 0.13, and 0.15 g·cm-3, respectively. Also, the optimal predicted R2 values of PTF for the topsoil, subsoil, and all soil samples based on the proposed RF method were 0.22, 0.45, and 0.37, respectively, while the RMSE values were 0.16, 0.11, and 0.14 g·cm-3, respectively.【Conclusion】The published PTF models had low BD prediction accuracy and were difficult to use for BD prediction on the scale of black soil in Northeast China whereas the PTF model constructed in this study has the potential to predict BD on the scale of black soil in Northeast China. Among the variables, SOM was the most important variable for predicting BD in the black soil of Northeast China, followed by MC, while soil texture-related variables had a relatively small impact.

    参考文献
    相似文献
    引证文献
引用本文

王晓盼,王昌昆,孙海军,郭志英,刘杰,高磊,马海艺,袁自然,姚成硕,潘贤章.基于随机森林的黑土地尺度耕地土壤容重传递函数构建[J].土壤学报,,[待发表]
WANG Xiaopan, WANG Changkun, SUN Haijun, GUO Zhiying, LIU Jie, GAO Lei, MA Haiyi, YUAN Ziran, YAO Chengshuo, PAN Xianzhang. Construction of pedotransfer function for predicting soil bulk density in cultivated land of Northeast China using random forest[J]. Acta Pedologica Sinica,,[In Press]

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2024-06-18
  • 最后修改日期:2024-10-22
  • 录用日期:2024-11-18
文章二维码
今日访问量： 您是第 位访问者
通讯地址：江苏省南京市江宁区麒麟街道创优路298号 邮政编码：211135 电话：025-86881237
网址：http://pedologica.issas.ac.cn/ E-mail：actapedo@issas.ac.cn
版权所有：《土壤学报》编辑部 技术支持：北京勤云科技发展有限公司 ICP:05004320号-1