地球信息科学学报 ›› 2020, Vol. 22 ›› Issue (2): 316-327.doi: 10.12082/dqxxkx.2020.190265

• 遥感科学与应用技术 • 上一篇    下一篇

基于包络检波和STFT谱分析的探地雷达土壤分层信息识别

李俐1,3, 付雪2, 崔佳2, 张超1,3, 朱德海1,3,*(), 吴克宁4   

  1. 1. 中国农业大学土地科学与技术学院,北京 100083
    2. 中国农业大学信息与电气工程学院,北京 100083
    3. 农业农村部农业灾害遥感重点实验室,北京 100083
    4. 中国地质大学(北京)土地科学技术学院,北京 100083
  • 收稿日期:2019-05-30 修回日期:2019-09-16 出版日期:2020-02-25 发布日期:2020-04-13
  • 通讯作者: 朱德海 E-mail:zhudehai@263.net
  • 作者简介:李 俐(1976— ),河南南阳人,副教授,主要从事微波农业应用研究。E-mail: lilixch@163.com
  • 基金资助:
    中国农业大学基本业务费项目(2019TC117)

Soil Layer Identification based on Envelope Detector and STFT Spectrum Analysis of Ground Penetrating Radar Signals

LI Li1,3, FU Xue2, CUI Jia2, ZHANG Chao1,3, ZHU Dehai1,3,*(), WU Kening4   

  1. 1. College of Land Science and Technology, China Agricultural University, Beijing 100083, China
    2. College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China
    3. Key Laboratory of Remote Sensing for Agri-Hazards, Ministry of Agriculture and Rural Affairs, Beijing 100083, China
    4. School of Land Science and Technology, China University of Geosciences, Beijing 100083, China
  • Received:2019-05-30 Revised:2019-09-16 Online:2020-02-25 Published:2020-04-13
  • Contact: ZHU Dehai E-mail:zhudehai@263.net
  • Supported by:
    The Fundamental Research Funds for the Central Universities(2019TC117)

摘要:

土壤分层信息,特别是表土层结构,对土地生产力具有重要影响,是评价土壤质量的一个重要指标。为了快速、准确地获取土壤分层信息,本文利用探地雷达对分层土壤进行了回波信号采集,并分别在时域和频域分析土壤层位置和层厚信息。首先在信号预处理的基础上,借助包络检波方法确定在土壤分层界面在时域上的位置;然后获取电磁波速度,得到土壤分层厚度。考虑到土壤介电常数与电磁波在土壤中传播速度的相关性,采用短时傅里叶变换方法(Short-time Fourier Transform,STFT)获取各土壤层时频域特征值,并利用回归分析建立特征值与介电常数之间的数学关系,实现对各土壤层的介电常数估算,从而计算出电磁波传播速度,进而确定土壤各层厚度。为验证算法的有效性,分别对理想模拟实验环境和农田环境进行了探地雷达实验,结果表明利用包络检波对探地雷达回波信息进行分析,土壤层检出率达到94.5%,借助STFT谱分析进行探地雷达回波速度估计,对于70 cm深度以上土层厚度计算误差大都保持在10%以下,但随着土壤深度的增加,误差变大。总体来说,本方法能有效识别浅层土壤的分层信息,可应用于实际生产中耕层厚度的估测。

关键词: 土壤分层, 探地雷达, 包络检波, 短时傅里叶变换, 土壤介电常数

Abstract:

Soil stratification information, especially surface soil structure, has important impact on land productivity and is an important index for evaluating soil quality. The present study aimed to obtain the information of soil layers quickly and accurately, for which Ground Penetrating Radar (GPR) technology was used. The echo signals of GPR were processed in both the time and frequency domains. In the time domain, the envelope detection method was used to determine the transience of the echo signals and therefore to get the location of soil layers on the time axis. To get the soil layer location in spatial coordinates, the velocity of electromagnetic wave propagation in soil was needed. Considering the velocity of electromagnetic wave propagation in soil layers varying with the soil dielectric constants, the Short-Time Fourier Transform (STFT) method was applied to the echo signals for dielectric constant analysis in the frequency domain. Soil layers with different dielectric constant exhibited different characteristics in the STFT signals. After clustering analysis of the soil layers, the relationship between STFT characteristic value and dielectric constant in a certain layer was established based on regressive analysis. Then, the velocity of electromagnetic wave propagation in each soil layer was determined using the dielectric constants. After the electromagnetic wave velocity of Ground Penetrating Radar (GPR) was estimated, the location of layers' interface was further determined and then the thickness of each soil layer was computed. To valid the effect of the above-mentioned methods, the echo signals of soil, for both the ideal simulated experimental environment which has obvious layered interface and the farmland environment whose layers have changed naturally, were collected. The experimental results show that, with the envelope detection method, layers not deeper than 70cm in the ideal simulated experimental environment were 100% detected and for both the ideal simulated experimental environment and the farmland environment, the detection rate of ground penetrating radar echo information reached 94.5%. The estimation of ground penetrating radar echo velocity using STFT spectrum analysis shows that the calculation error of soil thickness above 70 cm depth was mostly below 10%. Our findings suggest that the proprosed methodology can effectively identify the stratification information of shallow soil and estimate the thickness of the soil. However, surface vegetation, film mulching, soil voids, soil salinity, moisture heterogeneity, gradual change of soil layers, and soil layer depth will all affect the accuracy of the detection. For example, with the increase of soil depth, the error becomes larger. So, if the data acquisition spot is selected rationally, the proposed methodology can be applied to plough layer thickness detection in practical fileds.

Key words: soil stratification, ground penetrating radar, envelope detection, short-time Fourier transform, soil dielectric constant