基于Sentinel-2A的孙吴地区土壤有机质反演研究

doi:10.11720/wtyht.2022.0038

摘要
图/表
参考文献
相关文章
Metrics

全文: PDF(1863 KB) HTML
输出: BibTeX | EndNote (RIS)

摘要

利用Sentinel-2A多光谱遥感影像,结合实测土壤信息,对黑龙江省孙吴县黑土区土壤有机质含量进行反演研究。对影像进行预处理后,通过相关分析和随机森林(RF)选取特征波段,采用偏最小二乘法和BP神经网络构建土壤有机质含量多光谱模型反演红旗林场土壤有机质含量。研究表明:相关性选取的倒数对数一阶微分反射率波段和RF选择的组合波段能够有效提高土壤反演精度,组合波段的RF-BP神经网络模型反演效果最佳,R²=0.724 5,RMSE=1.312 7%。本次研究可为实现土壤有机质动态监测提供技术支持和参考。

	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
	陈超群
	戴慧敏
	冯雨林
	杨泽
	杨佳佳

关键词 ：黑土, 有机质, Sentinel-2A, 随机森林, BP神经网络

Abstract：

This study conducted the inversion of the organic matter content in the soil of the black soil area in Sunwu County, Heilongjiang Province using the Sentinel-2A multispectral remote sensing images and the surveyed soil data. After preprocessing the images, the characteristic bands were selected through correlation analysis and using the random forest (RF) method. Subsequently, a multispectral inversion model for the organic matter content of the soil was built using the partial least square method and the BP neural network, and the inversion of the organic matter content of the soil in the Hongqi Forest Farm was conducted. According to the obtained results, the bands selected based on the reciprocal of the logarithm of the first-order differential of reflectance through the correlation analysis and the combined bands selected using the RF method can effectively improve the inversion precision of the organic matter content in the soil, and the RF-BP neural network model for the combined bands yielded the optimal inversion performance (R²=0.7245 and RMSE=1.3127%). The results of this study will provide technical support and reference for the dynamic monitoring of the organic matter content in soils.

Key words： black soil organic matter Sentinel-2A random forest BP neural network

收稿日期: 2022-01-25 修回日期: 2022-06-27 出版日期: 2022-10-20

ZTFLH:

P632

基金资助:中国地质调查局项目“东北黑土地1:25万土地质量地球化学调查”(121201007000161312);“兴凯湖平原及松辽平原西部土地质量地球化学调查”(DD20190520)

通讯作者: 杨佳佳

作者简介: 陈超群(1996-),女,硕士研究生,主要研究方向为生态环境遥感与地理信息系统。Email:522110156@qq.com

引用本文:

陈超群, 戴慧敏, 冯雨林, 杨泽, 杨佳佳. 基于Sentinel-2A的孙吴地区土壤有机质反演研究[J]. 物探与化探, 2022, 46(5): 1141-1148.
CHEN Chao-Qun, DAI Hui-Min, FENG Yu-Lin, YANG Ze, YANG Jia-Jia. Sentinel-2A based inversion of the organic matter content of soil in the Sunwu area. Geophysical and Geochemical Exploration, 2022, 46(5): 1141-1148.

链接本文:

https://www.wutanyuhuatan.com/CN/10.11720/wtyht.2022.0038 或 https://www.wutanyuhuatan.com/CN/Y2022/V46/I5/1141

Fig.1 孙吴县遥感影像(a)及红旗林场位置(b)

Table 1 土壤样品中有机质含量统计信息

Fig.2 波段反射率及其变换与土壤有机质含量相关性

Table 2 相关性选取特征波段

Fig.3 对数变化交叉验证曲线

Fig.4 所有波段交叉验证曲线

Table 3 RF重要波段

	数学变换	拟合模型	建模集		测试集
	数学变换	拟合模型	R²	RMSE/%	R²	RMSE/%
相关性	R	$y = 6.3252 - 0.1679 x 1 1 - 0.3836 x 2 - 0.4512 x 3 - 0.6285 x 4 - 0.6827 x 5 - 0.6340 x 6 + 0.5943 x 7 - 0.5943 x 8 - 0.5199 x 9 - 0.5565 x 10$	0.0275	2.0263	0.0155	2.0925
	1/R	$y = 5.2818 + 0.0001 x 1 + 0.000017 x 2$	0.0439	1.9847	0.0657	2.0375
	lgR	$y = 5.4429 - 0.5202 x 1 - 0.2827 x 2 + 1.1558 x 3 - 1.5976 x 4 - 0.5254 x 5 + 3.3470 x 6$	0.0525	1.9757	0.0609	2.0406
	R^a	$y = 6.0894 + 48.0295 x 1 - 110.9565 x 2 + 21.8818 x 3 + 78.5713 x 4 - 33.5780 x 5 - 10.4766 x 6 - 5.2279 x 7$	0.0159	2.0136	0.0121	2.0917
	FDR	y=5.8376-23.9683x₁+67.1411x₂-63.9701x₃+44.2371x₄	0.0223	2.0071	0.0409	2.0632
	SDR	$y = 6.1351 + 4.19 x 1 + 14.4844 x 2 + 59.2323 x 3$	0.0200	2.0093	0.0238	2.0800
	FDLR	$y = 5.0094 + 1.7352 x 1 - 6.2089 x 2 + 6.6447 x 3 - 5.4056 x 4 + 3.6822 x 5 - 11.4086 x 6 - 3.5433 x 7 - 3.0895 x 8$	0.0524	1.9779	0.071	2.0366
	组合	$y = 8.8137 - 0.0162 x 1 + 1.7147 x 2 + 0.4484 x 3 - 0.1764 x 4 - 1.1268 x 5 - 0.4075 x 6 - 0.5233 x 7 - 1.5618 x 8 - 0.8610 x 9 - 1.2728 x 10$	0.0434	1.9852	0.0400	1.9933
随机森林	R	$y = 5.77391 - 8.6669 x 1 - 0.6491 x 2 + 9.5163 x 3 - 26.4628 x 4 + 8.9020 x 5 + 6.6144 x 6$	0.0203	2.0318	0.0035	2.0539
	1/R	$y = 5.80955 + 0.0002 x 1 + 0.0001 x 2 + 0.0009 x 3 - 0.1832 x 4 - 0.0226 x 5 + 0.2048 x 6$	0.0234	2.2798	0.09	2.0390
	lgR	$y = 4.40343 - 0.4786 x 1 - 0.0567 x 2 + 0.0415 x 3 + 0.0548 x 4 + 0.0692 x 5 + 0.0786 x 6$	0.0480	2.0029	0.0408	2.0124
	R^a	$y = 5.95664 - 0.4164 x 1 - 0.8538 x 2 - 2.1607 x 3 - 2.2302 x 4 - 2.3068 x 5 - 2.6665 x 6$	0.0051	2.0475	0.0023	2.0525
	FDR	$y = 5.65109 - 33.6899 x 1 + 85.8885 x 2 - 75.5978 x 3 - 13.8874 x 4 + 28.8994 x 5 - 6.2782 x 6$	0.0365	2.0149	0.0207	2.0356
	SDR	$y = 5.55202 + 13.2022 x 1 - 29.3378 x 2 + 42.5214 x 3 + 53.8592 x 4 - 3.5595 x 5 - 2.0586 x 6$	0.0331	2.0496	0.0200	2.0520
	FDLR	$y = 4.69522 - 0.0386 x 1 - 0.7984 x 2 - 0.6553 x 3 - 0.1570 x 4 - 0.0148 x 5 + 0.0701 x 6$	0.0463	2.0046	0.0418	2.0118
	组合	$y = 1.2967 x 1 + 0.2947 x 2 + 2.6967 x 3 + 0.5528 x 4 + 0.8238 x 5 + 0.7684 x 6 - 0.2272 x 7 - 0.6389 x 8 + 1.0292 x 9 + 1.3686 x 10 - 1.8264 x 11 + 0.7614 x 12 - 0.1926 x 13 + 1.2489 x 14 - 0.1014 x 15 - 0.3382 x 16 + 0.9709 x 17 - 0.3017 x 18 + 1.5403 x 19 + 2.7079 x 20 + 0.0221 x 21 - 3.4156 x 22 + 3.8703 x 23 + 4.6421 x 24 - 0.5339 x 25 - 1.746 x 26 - 1.7177$	0.0760	1.9518	0.0363	2.0183

Table 4 基于PLSR模型的土壤有机质反演

Table 5 基于BP神经网络模型的土壤有机质反演

Fig.5 红旗林场土壤有机质遥感反演和地球化学对比

[1]	孔牧, 杨少平. 森林沼泽景观区有机质对元素表生地球化学特征的影响机制[J]. 物探与化探, 2008, 32(1):31-32,74.
[1]	Kong M, Yang S P. Preliminary research into the disturbed principle of organic material to character of supergene-geochemistry in forest marsh landscape andscape area[J]. Geophysical and Geochemical Exploration, 2008, 32(1):31-32,74.
[2]	Rasmussen C, Heckman K, Wieder W R, et al. Beyond clay:Towards an improved set of variables for predicting soil organic matter content[J]. Biogeochemistry, 2018, 137(5):297-306. doi: 10.1007/s10533-018-0424-3
[3]	戴慧敏, 刘凯, 宋运红, 等. 东北地区黑土退化地球化学指示与退化强度[J]. 地质与资源, 2020, 29(6):510-517.DOI:10.13686/j.cnki.dzyzy.2020.06.002. doi: 10.13686/j.cnki.dzyzy.2020.06.002
[3]	Dai H M, Liu K, Song Y H, et al. Black soil degradation and intensity in northeast China: Geochemical indication[J]. Geology and Resources, 2020, 29(6):510-517.DOI:10.13686/j.cnki.dzyzy.2020.06.002. doi: 10.13686/j.cnki.dzyzy.2020.06.002
[4]	刘焕军, 张美薇, 杨昊轩, 等. 多光谱遥感结合随机森林算法反演耕作土壤有机质含量[J]. 农业工程学报, 2020, 36(10):134-140.
[4]	Liu H J, Zhang M W, Yang H X, et al. Invertion of cultivated soil organic matter content combining multi-spectral remote sensing and random forest algorithm[J]. Transactions of the Chinese Society of Agricultural Engineering, 2020, 36(10):134-140.
[5]	屈冉, 张雅琼, 聂忆黄, 等. 基于多光谱遥感影像的富川县表层土壤有机质含量反演[J]. 环境与可持续发展, 2019, 44(1):154-157.
[5]	Qu R, Zhang Y Q, Nie Y H, et al. Inversion of surface soil organic matter content in Fuchuan county based on multi spectral remote sensing image[J]. Environment and Sustainable Development, 2019, 44(1):154-157.
[6]	陈德宝, 陈桂芬. 基于Landsat8遥感图像的黑土区土壤有机质含量反演研究[J]. 中国农机化学报, 2020, 41(6):194-198. doi: 10.13733/j.jcam.issn.2095-5553.2020.06.031
[6]	Chen D B, Chen G F. Inversion of soil organic matter content in black soil region based on landsat8 remote sensing image[J]. Journal of Chinese Agricultural Mechanization, 2020, 41(6):194-198.
[7]	陈思明, 邹双全, 毛艳玲, 等. 土壤光谱重建的湿地土壤有机质含量多光谱反演[J]. 光谱学与光谱分析, 2018, 38(3):912-917.
[7]	Chen S M, Zou S Q, Mao Y L, et al. Inversion of soil organic matter content in wetland using multispectral data based on soil spectral reconstruction[J]. Spectroscopy and Spectral Analysis, 2018, 38(3):912-917.
[8]	Dhawale N M, Adamchuk V I, Prasher S O, et al. Proximal soil sensing of soil texture and organic matter with a prototype portable mid-infrared spectrometer[J]. European Journal of Soil Science, 2015, 66(4): 661-669. doi: 10.1111/ejss.12265
[9]	马驰. 基于Sentinel-2A遥感影像土壤有机质含量的反演研究[J]. 北方园艺, 2020(2):94-100.
[9]	Ma C. Inversion of soil organic matter content based on sentinel-2A remote sensing image[J]. Northern Horticulture, 2020(2):94-100.
[10]	刘鹏. 孙吴县耕地质量评价[D]. 哈尔滨: 东北农业大学, 2020.
[10]	Liu P. Evaluation of cultivated land quality in Sunwu County[D]. Harbin: Northeast Agricultural University, 2020.
[11]	李丹丹. 黑河市耕地地力评价与土壤改良对策研究[D]. 哈尔滨: 东北农业大学, 2018.
[11]	Li D D. Investigation and evaluation on cultivated land fertility of Heihe City[D]. Harbin: Northeast Agricultural University, 2018.
[12]	Breiman L. Random forests[J]. Machine Learning, 2001, 45(1):5-32. doi: 10.1023/A:1010933404324
[13]	彭刘亚, 解惠婷, 冯伟栋. 基于随机森林算法的砂土液化预测方法[J]. 物探与化探, 2020, 44(6):1429-1434.
[13]	Peng L Y, Xie H T, Feng W D. The method of predict sand liquefaction based on random forest algorithm[J]. Geophysical and Geochemical Exploration, 2020, 44(6):1429-1434.
[14]	王琨, 肖克炎, 丛源. 对数比变换和偏最小二乘法在地球化学组合异常提取中的应用——以湘西北铅锌矿为例[J]. 物探与化探, 2015, 39(1):141-148.
[14]	Wang K, Xiao K Y, Cong Y. Log-ratio transformation and PLS methods for identifying integrated geochemical anomalies: A case study of lead-zinc mineralization in northwestern Hunan[J]. Geophysical and Geochemical Exploration, 2015, 39(1):141-148.
[15]	陈昊宇, 杨光, 韩雪莹, 等. 基于连续小波变换的土壤有机质含量高光谱反演[J]. 中国农业科技导报, 2021, 23(5):132-142.DOI:10.13304/j.nykjdb.2020.0742. doi: 10.13304/j.nykjdb.2020.0742
[15]	Chen H Y, Yang G, Han X Y, et al. Hyperspectral inversion of soil organic matter content based on continuous wavelet transform journal of agricultural science and technology[J]. Journal of Agricultural Science and Technology, 2021, 23(5):132-142.DOI:10.13304/j.nykjdb.2020.0742. doi: 10.13304/j.nykjdb.2020.0742
[16]	叶勤, 姜雪芹, 李西灿, 等. 基于高光谱数据的土壤有机质含量反演模型比较[J]. 农业机械学报, 2017, 48(3):164-172.
[16]	Ye Q, Jiang X Q, Li X C, et al. Comparison on inversion model of soil organic matter content based on hyperspectral data[J]. Transactions of the Chinese Society for Agricultural Machinery, 2017, 48(3):164-172.
[17]	王启元, 赵艳玲, 房铄东, 等. 基于多光谱遥感的裸土土壤含水量反演研究[J]. 矿业科学学报, 2020, 5(6):608-615.
[17]	Wang Q Y, Zhao Y L, Fang S D, et al. Inversion of soil moisture in bare soil based on multi-spectral remote sensing[J]. Journal of Mining Science and Technology, 2020, 5(6): 608-615.
[18]	汤超. 淮北矿区有机质含量反演[J]. 农业与技术, 2021, 41(13):123-128.DOI:10.19754/j.nyyjs.20210715035. doi: 10.19754/j.nyyjs.20210715035
[18]	Tang C. Inversion of organic matter content in Huaibei mining area[J]. Agriculture and Technology, 2021, 41(13):123-128.DOI:10.19754/j.nyyjs.20210715035. doi: 10.19754/j.nyyjs.20210715035
[19]	谢树刚. 基于高光谱的黄河三角洲土壤有机质含量估测模型研究[D]. 泰安: 山东农业大学, 2021.DOI:10.27277/d.cnki.gsdnu.2021.000631. doi: 10.27277/d.cnki.gsdnu.2021.000631
[19]	Xie S G. Research on estimation model of soil organic matter content in Yellow River Delta based on hyperspectral[D]. Tai’an: Shandong Agricultural University, 2021.DOI: 10.27277/d.cnki.gsdnu.2021.000631. doi: 10.27277/d.cnki.gsdnu.2021.000631
[20]	陶培峰, 王建华, 李志忠, 等. 基于高光谱的土壤养分含量反演模型研究[J]. 地质与资源, 2020, 29(1):68-75,84.DOI:10.13686/j.cnki.dzyzy.2020.01.006. doi: 10.13686/j.cnki.dzyzy.2020.01.006
[20]	Tao P F, Wang J H, Li Z Z, et al. Research of soil nutrient content inversion model based on hyperspectral data[J]. Geology and Resources, 2020, 29(1):68-75,84.DOI:10.13686/j.cnki.dzyzy.2020.01.006. doi: 10.13686/j.cnki.dzyzy.2020.01.006

[1]	李朋飞, 管后春, 王翔, 陈岩滨, 王耀, 吴衡, 史春鸿. 皖北潮土与砂姜黑土锌含量分布及影响因素[J]. 物探与化探, 2022, 46(6): 1545-1554.
[2]	肖红叶, 刘国栋, 杨泽, 房娜娜, 戴慧敏, 刘凯, 韩晓萌, 朱恒. 东北黑土区近半世纪土地利用变化时空特征分析[J]. 物探与化探, 2022, 46(5): 1037-1049.
[3]	宋运红, 杨凤超, 刘凯, 戴慧敏, 许江, 韩晓萌. 黑龙江省海伦地区黑土剖面常量元素地球化学特征及其对物源的指示意义[J]. 物探与化探, 2022, 46(5): 1105-1113.
[4]	刘凯, 戴慧敏, 刘国栋, 宋运红, 梁帅, 杨泽. 基于主成分聚类法的典型黑土区土壤地球化学分类[J]. 物探与化探, 2022, 46(5): 1132-1140.
[5]	王鹏, 赵君, 刘拓, 周一凡, 魏锦萍, 王磊. 半干旱区有机质与全氮空间变异的尺度效应特征——以延安市为例[J]. 物探与化探, 2022, 46(4): 1011-1020.
[6]	殷启春, 王元俊, 周道容, 张丽, 孙桐. 复电阻率法在安徽南陵盆地海相页岩气勘探中的应用[J]. 物探与化探, 2022, 46(3): 668-677.
[7]	郭建宏, 张占松, 张超谟, 周雪晴, 肖航, 秦瑞宝, 余杰. 用地球物理测井资料预测煤层气含量——基于斜率关联度—随机森林方法的工作案例[J]. 物探与化探, 2021, 45(1): 18-28.
[8]	彭刘亚, 解惠婷, 冯伟栋. 基于随机森林算法的砂土液化预测方法[J]. 物探与化探, 2020, 44(6): 1429-1434.
[9]	严明书, 吴春梅, 蒙丽, 丁相伦, 董攀, 邓海, 雷家立, 龚媛媛, 鲍丽然. 重庆市黔江猕猴桃果园土壤养分状况分析[J]. 物探与化探, 2019, 43(5): 1123-1130.
[10]	付康伟, 张学强, 彭炎. BP神经网络算法在陆域天然气水合物成藏预测中的应用[J]. 物探与化探, 2019, 43(3): 486-493.
[11]	张兆龙, 王跃钢, 腾红磊, 王乐. 一种基于遗传算法和BP神经网络的选星方法[J]. 物探与化探, 2017, 41(5): 946-950.
[12]	李春鹏, 隋桂梅, 刘志国, 杨松岭, 闫青华, 尹川. 成熟—过成熟烃源岩有机质类型识别[J]. 物探与化探, 2017, 41(2): 219-223.
[13]	段晓梦, 陈培元, 吕栋, 孔令武, 蒋百召. 增生楔盆地烃源岩特征综合评价——以缅甸某区块为例[J]. 物探与化探, 2016, 40(2): 257-263.
[14]	张瑞, 陈刚, 潘保芝, 蒋必辞, 杨雪, 刘丹. 基于细菌觅食优化广义回归神经网络的煤层气含量预测[J]. 物探与化探, 2016, 40(2): 327-332.
[15]	周印明, 刘雪军, 张春贺, 朱永山. 快速识别页岩气“甜点”目标的时频电磁勘探技术及应用[J]. 物探与化探, 2015, 39(1): 60-63，83.

Viewed

Full text

Abstract

Cited

Shared

Discussed