Journal of Integrative Agriculture ›› 2024, Vol. 23 ›› Issue (4): 1393-1408.DOI: 10.1016/j.jia.2023.09.017

• • 上一篇    下一篇

基于多年月尺度合成影像的耕地土壤有机质空间预测

  

  • 收稿日期:2023-05-11 接受日期:2023-07-21 出版日期:2024-04-20 发布日期:2024-03-30

Mapping soil organic matter in cultivated land based on multi-year composite images on monthly time scales

Jie Song1, 3, Dongsheng Yu1, 3#, Siwei Wang2, Yanhe Zhao2, Xin Wang1, 3, Lixia Ma1, Jiangang Li1, 3   

  1. 1 State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, Nanjing 210008, China

    2 Agricultural and Rural Bureau of Luanping County, Luanping 068250, China 3 Chinese Academy of Sciences University, Beijing 100049, China

  • Received:2023-05-11 Accepted:2023-07-21 Online:2024-04-20 Published:2024-03-30
  • About author:#Correspondence Dongsheng Yu, E-mail: dshyu@issas.ac.cn
  • Supported by:
    Funding and resources for this study came from the special project of the National Key Research and Development Program of China (2022YFB3903302 and 2021YFC1809104).  

摘要:

快速、准确地获取耕地土壤有机质(SOM)空间分布对农业可持续发展和碳平衡管理较为重要。本文提出了基于多年月尺度合成影像预测SOM的方法。利用谷歌地球引擎(GEE)平台获取2016-2021年覆盖整个研究区的哨兵2号遥感影像数据并逐月合成,提取合成影像的光谱波段和植被指数作为环境协变量,并构建随机森林(RF)、支持向量机(SVM)和梯度提升回归树(GBRT)模型比较不同变量组合下SOM预测精度的差异。结果表明:(1) 1341011 合成的光谱波段均与SOM显著相关(P < 0.05);(2)基于单月变量的模型预测精度整体低,其中,1份变量的RF模型预测精度最高,决定系数R2为0.36但将不同月份变量按四个季度进行组合,第一季度(Q1)和第四季度(Q4)模型预测性能较好,三个季度变量任意组合的模型预测精度差异较小。当所有月份的变量被纳入模型时,模型预测性能最佳;(3)三种机器学习算法中RF模型的预测精度始终高于 SVM GBRT 模型,其决定系数R2为0.56。除 12 份的Band12波段外,其余变量的重要性无显著差异。该研究为高精度空间分辨率的SOM数字制图提供了理论参考

Abstract:

Rapid and accurate acquisition of soil organic matter (SOM) information in cultivated land is important for sustainable agricultural development and carbon balance management.  This study proposed a novel approach to predict SOM with high accuracy using multiyear synthetic remote sensing variables on a monthly scale.  We obtained 12 monthly synthetic Sentinel-2 images covering the study area from 2016 to 2021 through the Google Earth Engine (GEE) platform, and reflectance bands and vegetation indices were extracted from these composite images.  Then the random forest (RF), support vector machine (SVM) and gradient boosting regression tree (GBRT) models were tested to investigate the difference in SOM prediction accuracy under different combinations of monthly synthetic variables.  Results showed that firstly, all monthly synthetic spectral bands of Sentinel-2 showed a significant correlation with SOM (P<0.05) for the months of January, March, April, October, and November.  Secondly, in terms of single-monthly composite variables, the prediction accuracy was relatively poor, with the highest R2 value of 0.36 being observed in January.  When monthly synthetic environmental variables were grouped in accordance with the four quarters of the year, the first quarter and the fourth quarter showed good performance, and any combination of three quarters was similar in estimation accuracy.  The overall best performance was observed when all monthly synthetic variables were incorporated into the models.  Thirdly, among the three models compared, the RF model was consistently more accurate than the SVM and GBRT models, achieving an R2 value of 0.56.  Except for band 12 in December, the importance of the remaining bands did not exhibit significant differences.  This research offers a new attempt to map SOM with high accuracy and fine spatial resolution based on monthly synthetic Sentinel-2 images.

Key words: soil organic matter , Sentinel-2 , monthly synthetic images , machine learning model , spatial prediction