基于地统计学与机器学习模型的宝清县表层有机碳含量预测及比较研究

doi:10.11720/wtyht.2025.1488

摘要
图/表
参考文献
相关文章
Metrics

全文: PDF(3219 KB) HTML
输出: BibTeX | EndNote (RIS)

摘要

为了准确预测黑土地区县域土壤有机碳含量,满足县域农业生产和“双碳”目标的需要,本研究利用宝清县黑土地地表基质调查取得的427个土壤样本数据,通过确定性插值法(反距离权重法)、地统计学方法(普通克里金法)和机器学习方法(随机森林模型法)建立评估模型,进行宝清县表层土壤有机碳含量的预测,并比较不同方法的预测精度和效果。结果显示,反距离权重法预测研究区的土壤有机碳含量均值为27.21×10^-3,普通克里金法预测的均值为26.33×10^-3,随机森林模型预测的均值为32.05×10^-3。随机森林模型在均方根误差(RMSE)、平均绝对误差(MAE)和决定系数(R²)等指标上均优于反距离权重法和普通克里金法,且随机森林模型训练集和验证集的决定系数均达到0.73和0.53,精度均显著高于反距离权重法和普通克里金法,反映出随机森林模型通过环境变量的非线性交互作用,能够更充分地挖掘数据中的潜在规律。总体看,结合多环境变量建立的随机森林模型是评估宝清县表层土壤有机碳含量的最佳模型,且预测精度较好,这对评估黑土地区县域农业生产土壤有机质空间差异和“双碳”目标的区域差异分析具有重要的理论和方法参考。

	服务

	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	刘洪博
	史佳卉
	王思引
	裴久渤

关键词 ：黑土地区, 地表基质调查, 有机碳, 地统计方法, 随机森林模型

Abstract：

This study aims to accurately predict the organic carbon content in black soils at the county level, thereby supporting county-level agricultural production and carbon peak and neutrality goals. This study examined 427 soil samples obtained from a surface substrate survey of the black soil area in Baoqing County. Employing deterministic interpolation (inverse distance weighting, IDW), geostatistics (ordinary Kriging method, OK), and machine learning (random forest, RF), this study constructed assessment models to predict the organic carbon content in topsoils in Baoqing County and to compare their prediction accuracy and performance. The results show that the IDW, OK, and RF models yielded average organic carbon contents of 27.21×10^-3, 26.33×10^-3, and 32.05×10^-3, respectively. The RF model outperformed the other two models in terms of root mean square error (RMSE), mean absolute error (MAE), and the coefficient of determination (R²). Specifically, the RF model achieved R² values of 0.73 and 0.53 on training and validation sets, respectively, suggesting significantly higher accuracy. This superior performance demonstrates that the RF model can more fully explore potential patterns in data through the nonlinear interaction of environmental variables. Overall, the RF model, incorporating multiple environmental variables, proved to be the optimal approach for predicting the organic carbon content in topsoils in Baoqing County, demonstrating high prediction accuracy. This study provides valuable theoretical and methodological insights for assessing the spatial variations in soil organic matter relevant to county-level agricultural production and regional differences in carbon peak and neutrality goals within black soil areas.

Key words： black soil area surface substrate survey organic carbon geostatistical method random forest model

收稿日期: 2024-12-12 修回日期: 2025-05-14 出版日期: 2025-10-20

ZTFLH:	X142
	X825

基金资助:自然资源综合调查指挥中心科技创新基金项目(KC20220002);中国地质调查局项目“松嫩平原西部1∶25万地表基质调查”(20240100903)

通讯作者: 裴久渤(1984-),男,博士,教授,主要从事土壤肥力构成与演变研究工作。Email:peijiubo@syau.edu.cn

作者简介: 刘洪博(1991-),男,学士,工程师,主要从事灾害地质调查、地表基质调查工作。Email:2274496829@qq.com

引用本文:

刘洪博, 史佳卉, 王思引, 裴久渤. 基于地统计学与机器学习模型的宝清县表层有机碳含量预测及比较研究[J]. 物探与化探, 2025, 49(5): 1243-1250.
LIU Hong-Bo, SHI Jia-Hui, WANG Si-Yin, PEI Jiu-Bo. Prediction and comparison of organic carbon content in topsoils based on geostatistics and machine learning models: A case study of Baoqing County. Geophysical and Geochemical Exploration, 2025, 49(5): 1243-1250.

链接本文:

https://www.wutanyuhuatan.com/CN/10.11720/wtyht.2025.1488 或 https://www.wutanyuhuatan.com/CN/Y2025/V49/I5/1243

Fig.1 研究区地形及采样点位置

Fig.2 数据正态分布曲线

Table 1 半方差函数模型参数

Fig.3 土壤有机碳含量半方差函数

Fig.4 各环境变量的相关性热力图

Fig.5 环境变量特征重要性分析

Fig.6 反距离权重法预测宝清县土壤有机碳空间分布

Fig.7 普通克里金法预测宝清县土壤有机碳空间分布

Fig.8 随机森林预测宝清县土壤有机碳空间分布

Table 2 有机碳含量空间预测结果统计分析

Table 3 有机碳插值精度分析结果

Fig.9 反距离权重法预测结果与实际值对照

Fig.10 普通克里金法预测结果与实际值对照

Fig.11 随机森林预测结果与实际值对照

[1]	唐宽燕. 内蒙古草原土壤有机碳空间变化及其驱动因子研究[D]. 呼和浩特: 内蒙古大学, 2023.
[1]	Tang K Y. Study on the spatial variation of soil organic carbon and its driving factors in the Inner Mongolian steppes[D]. Hohhot: Inner Mongolia University, 2023.
[2]	向婷. 黄土丘陵区小流域土壤有机碳空间分布特征及驱动因素分析[D]. 延安: 延安大学, 2024.
[2]	Xiang T. Spatial distribution characteristics and driving factors of soil organic carbon in small watershed in Loess Hilly Region[D]. Yan’an: Yan’an University, 2024.
[3]	侯红星, 葛良胜, 孙肖, 等. 地表基质在中国黑土地资源调查评价中的应用探讨——基于黑龙江宝清地区地表基质调查[J]. 自然资源学报, 2022, 37(9):2264-2276. doi: 10.31497/zrzyxb.20220905
[3]	Hou H X, Ge L S, Sun X, et al. A study on the application of ground substrate in the survey and evaluation of China’s black soil resources:Based on ground substrate survey in Baoqing,Heilongjiang Province[J]. Journal of Natural Resources, 2022, 37(9):2264-2276.
[4]	梅帅, 童童, 应纯洋, 等. 基于机器学习的数字土壤制图研究进展[J]. 农业资源与环境学报, 2024, 41(4):744-756.
[4]	Mei S, Tong T, Ying C Y, et al. Advances in digital soil mapping based on machine learning[J]. Journal of Agricultural Resources and Environment, 2024, 41(4):744-756.
[5]	韩杏杏, 陈杰, 王海洋, 等. 基于随机森林模型的耕地表层土壤有机质含量空间预测——以河南省辉县市为例[J]. 土壤, 2019, 51(1):152-159.
[5]	Han X X, Chen J, Wang H Y, et al. Spatial prediction of SOM content in topsoil based on random forest algorithm:A case study of Huixian City,Henan Province[J]. Soils, 2019, 51(1):152-159.
[6]	乔婷, 姚彩燕, 于东升, 等. 水田土壤有机碳时空演变下的最优插值方法[J]. 福建农林大学学报:自然科学版, 2020, 49(5):683-694.
[6]	Qiao T, Yao C Y, Yu D S, et al. Optimal interpolation method for spatial-temporal evolution of soil organic carbon in paddy fields[J]. Journal of Fujian Agriculture and Forestry University:Natural Science Edition, 2020, 49(5):683-694.
[7]	陈琳, 任春颖, 王宗明, 等. 基于克里金插值的耕地表层土壤有机质空间预测[J]. 干旱区研究, 2017, 34(4):798-805.
[7]	Chen L, Ren C Y, Wang Z M, et al. Prediction of spatial distribution of topsoil organic matter content in cultivated land using Kriging methods[J]. Arid Zone Research, 2017, 34(4):798-805.
[8]	杨其坡, 武伟, 刘洪斌. 基于地形因子和随机森林的丘陵区农田土壤有效铁空间分布预测[J]. 中国生态农业学报, 2018, 26(3):422-431.
[8]	Yang Q P, Wu W, Liu H B. Prediction of spatial distribution of soil available iron in a typical hilly farmland using terrain attributes and random forest model[J]. Chinese Journal of Eco-Agriculture, 2018, 26(3):422-431.
[9]	王雨雪, 杨柯, 高秉博, 等. 基于两点机器学习方法的土壤有机质空间分布预测[J]. 农业工程学报, 2022, 38(12):65-73.
[9]	Wang Y X, Yang K, Gao B B, et al. Prediction of the spatial distribution of soil organic matter based on two-point machine learning method[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(12):65-73.
[10]	沈琛琛, 肖文发, 朱建华, 等. 基于机器学习算法的华中天然林土壤有机碳特征与关键影响因子[J]. 林业科学, 2024, 60(3):65-77.
[10]	Shen C C, Xiao W F, Zhu J H, et al. Characterization of soil organic carbon and key influencing factors of natural forests in Central China based on machine learning algorithms[J]. Scientia Silvae Sinicae, 2024, 60(3):65-77.
[11]	王志远, 汤哲, 周萍, 等. 面向亚热带丘陵区小流域土壤有机碳空间预测的四种模型构建及性能比较[J]. 农业现代化研究, 2023, 44(3):558-566.
[11]	Wang Z Y, Tang Z, Zhou P, et al. Comparison of four machine learning in predicting soil organic carbon content in a small watershed in the subtropical hilly area[J]. Research of Agricultural Modernization, 2023, 44(3):558-566.
[12]	夏晓莹, 李思瑶, 王杰, 等. 地形因子对天山北坡天山云杉林土壤有机碳的影响[J]. 新疆农业科学, 2023, 60(4):965-973. doi: 10.6048/j.issn.1001-4330.2023.04.022
[12]	Xia X Y, Li S Y, Wang J, et al. Effects of topographic factors on soil organic carbon in Picea schrenkiana forest on the northern slope of Tianshan Mountain[J]. Xinjiang Agricultural Sciences, 2023, 60(4):965-973.
[13]	Lal R. Soil carbon sequestration impacts on global climate change and food security[J]. Science, 2004, 304(5677):1623-1627. doi: 10.1126/science.1097396 pmid: 15192216
[14]	Minasny B, McBratney A B. Digital soil mapping:A brief history and some lessons[J]. Geoderma, 2016,264:301-311.
[15]	刘洪博, 孔繁鹏, 赵建, 等. 地表基质调查技术方法探索与实验——以黑龙江省宝清县黑土地调查为例[J]. 地理信息世界, 2022, 29(6):1-5.
[15]	Liu H B, Kong F P, Zhao J, et al. Exploration and experiment of surface substrate investigation technique:A case study of black soil investigation in Baoqing County,Heilongjiang Province[J]. Geomatics World, 2022, 29(6):1-5.
[16]	Lu J N, Feng S, Wang S K, et al. Patterns and driving mechanism of soil organic carbon,nitrogen,and phosphorus stoichiometry across northern China’s desert-grassland transition zone[J]. Catena, 2023,220:106695.
[17]	李海涛, 邵泽东. 空间插值分析算法综述[J]. 计算机系统应用, 2019, 28(7):1-8.
[17]	Li H T, Shao Z D. Review of spatial interpolation analysis algorithm[J]. Computer Systems & Applications, 2019, 28(7):1-8.
[18]	杜可, 王乐, 张淑香, 等. 黑土区县域土壤养分空间分布特征及其影响因子[J]. 植物营养与肥料学报, 2018, 24(6):1465-1474.
[18]	Du K, Wang L, Zhang S X, et al. Spatial distribution characteristics and influence factors of soil nutrients in black soil region counties[J]. Journal of Plant Nutrition and Fertilizers, 2018, 24(6):1465-1474.
[19]	凌威, 王新杰, 武文昊. 新疆年均降水量的空间插值方法比较[J]. 辽宁林业科技, 2020(4):5-9,58.
[19]	Ling W, Wang X J, Wu W H. Comparison on spatial interpolation methods of average annual precipitation in Xinjiang[J]. Liaoning Forestry Science and Technology, 2020(4):5-9,58.
[20]	Breiman L. Random forests[J]. Machine Learning, 2001, 45(1):5-32.
[21]	李治军, 王华凡, 侯岳, 等. 缺资料地区降雨量空间插值方法比较[J]. 水利科学与寒区工程, 2022, 5(8):68-71.
[21]	Li Z J, Wang H F, Hou Y, et al. Comparison of spatial interpolation methods for precipitation in regions lacking data[J]. Hydro Science and Cold Zone Engineering, 2022, 5(8):68-71.
[22]	齐伟恒, 彭琳, 郜鲁涛, 等. 基于ArcGIS地统计分析模块的土壤养分与pH值空间变异分析——以云南省寻甸县为例[J]. 江苏农业科学, 2018, 46(23):287-291.
[22]	Qi W H, Peng L, Gao L T, et al. Study on spatial variability of soil nutrients and pH values based on GIS geostatistical analysis module—Taking Xundian Country of Yunnan Province as an example[J]. Jiangsu Agricultural Sciences, 2018, 46(23):287-291.
[23]	张志坚, 刘苑秋, 吴春生, 等. 基于地统计学和GIS的江西省森林土壤养分空间分布特征[J]. 水土保持研究, 2018, 25(1):38-46.
[23]	Zhang Z J, Liu Y Q, Wu C S, et al. Spatial distribution characteristics of forest soil nutrients in Jiangxi Province based on geostatistics and GIS[J]. Research of Soil and Water Conservation, 2018, 25(1):38-46.

[1]	夏炎, 王润涛, 杜倩倩, 王喜宽, 吕红杰, 侯进凯, 李炳辉. 洛阳市农田土壤碳汇规律研究[J]. 物探与化探, 2025, 49(1): 215-228.
[2]	段星星, 刘小龙, 韩宝华, 阿地来·赛提尼亚孜, 金梦婷, 刘彤. 黄土高原地区土壤有机碳和无机碳储量及含量特征[J]. 物探与化探, 2025, 49(1): 239-247.
[3]	刘凯, 戴慧敏, 刘国栋, 梁帅, 魏明辉, 杨泽, 宋运红. 基于土壤有机碳含量的黑土层厚度预测及影响因素分析[J]. 物探与化探, 2024, 48(5): 1368-1376.
[4]	刘庆宇, 马瑛, 程莉, 沈骁, 张亚峰, 苗国文, 黄强, 韩思琪. 青海东部表层土壤有机碳密度及其空间分布特征[J]. 物探与化探, 2023, 47(4): 1098-1108.
[5]	张一鹤, 杨泽, 戴慧敏, 刘国栋, 韩晓萌, 李秋燕. 穆棱河—兴凯湖平原土壤有机碳、全氮的时空变异特征[J]. 物探与化探, 2022, 46(5): 1050-1055.
[6]	杨泽, 张一鹤, 戴慧敏, 刘国栋, 刘凯, 许江. 兴凯湖平原表层土壤有机碳空间变异的主控因素[J]. 物探与化探, 2022, 46(5): 1076-1086.
[7]	刘国栋, 李禄军, 戴慧敏, 许江, 刘凯, 张一鶴, 杨泽. 松辽平原土壤碳库变化及其原因分析[J]. 物探与化探, 2021, 45(5): 1109-1120.
[8]	胡梦颖, 徐进力, 朱颖涛, 张鹏鹏, 张灵火, 徐红纳, 白金峰, 张勤. 氧化燃烧—气相色谱法测定土壤中的有机碳含量[J]. 物探与化探, 2020, 44(6): 1435-1440.
[9]	王月平, 张立, 崔玉军, 吕石佳. 宝清县东部土壤硒含量特征及其与土壤性质的关系[J]. 物探与化探, 2019, 43(4): 904-911.
[10]	鲍丽然, 严明书, 贾中民, 龚媛媛. 重庆西部表层土壤有机碳储量与密度分布[J]. 物探与化探, 2015, 39(1): 180-185.
[11]	刘文辉. 甘肃省张掖—永昌地区土壤有机碳密度估算及其空间分布特征[J]. 物探与化探, 2013, 37(3): 552-556.
[12]	刘文辉, 李春亮, 吴永强. 甘肃省兰州—白银地区土壤有机碳库储量估算与空间分布特征[J]. 物探与化探, 2012, 36(3): 367-371.
[13]	刘海良, 苏文利, 罗振佳, 李磊. 页岩气、油页岩资源的电阻率法勘查[J]. 物探与化探, 2012, 36(3): 503-506.

Viewed

Full text

Abstract

Cited

Shared

Discussed