# Experimental Research on Evaluation of Soil Water Content Using Ground Penetrating Radar and Wavelet Packet-Based Energy Analysis

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Methods

#### 2.1. Construction of Biorthogonal Wavelet Basis

^{N′}[52,53]

#### 2.2. Decomposition of Wavelet Packet Transform

#### 2.3. The WPEA Method

#### 2.4. GPR Methodology

#### 2.4.1. Detection Principle

#### 2.4.2. Forward Simulation

## 3. Experimental Investigation

#### 3.1. Experimental Setup

#### 3.2. GPR Data Collection

## 4. Results

#### 4.1. Preprocessing of GPR Signals

#### 4.2. Time-Domain Signals Response

#### 4.3. WPEI of GPR Signals

#### 4.4. Comparative Analysis

## 5. Discussion

## 6. Conclusions

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Acknowledgments

## Conflicts of Interest

## References

- Shen, X.; Foster, T.; Baldi, H.; Dobreva, I.; Burson, B.; Hays, D.; Tabien, R.; Jessup, R. Quantification of soil organic carbon in biochar-amended soil using ground penetrating radar (GPR). Remote Sens.
**2019**, 11, 2874. [Google Scholar] [CrossRef] [Green Version] - Zajícová, K.; Chuman, T. Application of ground penetrating radar methods in soil studies: A review. Geoderma
**2019**, 343, 116–129. [Google Scholar] [CrossRef] - SU, S.L.; Singh, D.N.; Shojaei Baghini, M. A critical review of soil moisture measurement. Measurement
**2014**, 54, 92–105. [Google Scholar] [CrossRef] - Venmans, A.A.M.; van de Ven, R.; Kollen, J. Rapid and non-intrusive measurements of moisture in road constructions using passive microwave radiometry and GPR—Full scale test. Procedia Eng.
**2016**, 143, 1244–1251. [Google Scholar] [CrossRef] [Green Version] - Klewe, T.; Strangfeld, C.; Kruschwitz, S. Review of moisture measurements in civil engineering with ground penetrating radar—Applied methods and signal features. Constr. Build. Mater.
**2021**, 278, 122250. [Google Scholar] [CrossRef] - Cui, X.; Zhang, Z.; Guo, L.; Liu, X.; Quan, Z.; Cao, X.; Chen, X. The root-soil water relationship is spatially anisotropic in shrub-encroached grassland in North China: Evidence from GPR investigation. Remote Sens.
**2021**, 13, 1137. [Google Scholar] [CrossRef] - Liu, J.; Cheng, Y.S. A new approach of detecting and evaluation railway by using the train-mounted ground penetrating radar in China. Adv. Mater. Res.
**2011**, 243–249, 4351–4355. [Google Scholar] [CrossRef] - Cao, Q.; Song, X.; Wu, H.; Gao, L.; Liu, F.; Yang, S.; Zhang, G. Mapping the response of volumetric soil water content to an intense rainfall event at the field scale using GPR. J. Hydrol.
**2020**, 583, 124605. [Google Scholar] [CrossRef] - Tong, Z.; Gao, J.; Zhang, H. Innovative method for recognizing subgrade defects based on a convolutional neural network. Constr. Build. Mater.
**2018**, 169, 69–82. [Google Scholar] [CrossRef] - Barca, E.; De Benedetto, D.; Stellacci, A.M. Contribution of EMI and GPR proximal sensing data in soil water content assessment by using linear mixed effects models and geostatistical approaches. Geoderma
**2019**, 343, 280–293. [Google Scholar] [CrossRef] - Shamir, O.; Goldshleger, N.; Basson, U.; Reshef, M. Laboratory measurements of subsurface spatial moisture content by ground-penetrating radar (GPR) diffraction and reflection imaging of agricultural soils. Remote Sens.
**2018**, 10, 1667. [Google Scholar] [CrossRef] [Green Version] - Galagedara, L.W.; Parkin, G.W.; Redman, J.D.; von Bertoldi, P.; Endres, A.L. Field studies of the GPR ground wave method for estimating soil water content during irrigation and drainage. J. Hydrol.
**2005**, 301, 182–197. [Google Scholar] [CrossRef] - Tian, J.; Yue, J.; Philpot, W.D.; Dong, X.; Tian, Q. Soil moisture content estimate with drying process segmentation using shortwave infrared bands. Remote Sens. Environ.
**2021**, 263, 112552. [Google Scholar] [CrossRef] - Reynolds, S.G. The gravimetric method of soil moisture determination Part I A study of equipment, and methodological problems. J. Hydrol.
**1970**, 11, 258–273. [Google Scholar] [CrossRef] - Plati, C.; Loizos, A. Estimation of in-situ density and moisture content in HMA pavements based on GPR trace reflection amplitude using different frequencies. J. Appl. Geophys.
**2013**, 97, 3–10. [Google Scholar] [CrossRef] - Khakiev, Z.; Shapovalov, V.; Kruglikov, A.; Morozov, A.; Yavna, V. Investigation of long term moisture changes in trackbeds using GPR. J. Appl. Geophys.
**2014**, 110, 1–4. [Google Scholar] [CrossRef] - Walker, J.P.; Willgoose, G.R.; Kalma, J.D. In situ measurement of soil moisture: A comparison of techniques. J. Hydrol.
**2004**, 293, 85–99. [Google Scholar] [CrossRef] - Zhang, J.; Lin, H.; Doolittle, J. Soil layering and preferential flow impacts on seasonal changes of GPR signals in two contrasting soils. Geoderma
**2014**, 213, 560–569. [Google Scholar] [CrossRef] - Zreda, M.; Desilets, D.; Ferré, T.P.A.; Scott, R.L. Measuring soil moisture content non-invasively at intermediate spatial scale using cosmic-ray neutrons. Geophys. Res. Lett.
**2008**, 35, L21402. [Google Scholar] [CrossRef] [Green Version] - Bablet, A.; Vu, P.V.H.; Jacquemoud, S.; Viallefont-Robinet, F.; Fabre, S.; Briottet, X.; Sadeghi, M.; Whiting, M.L.; Baret, F.; Tian, J. MARMIT: A multilayer radiative transfer model of soil reflectance to estimate surface soil moisture content in the solar domain (400–2500 nm). Remote Sens. Environ.
**2018**, 217, 1–17. [Google Scholar] [CrossRef] [Green Version] - Verstraeten, W.W.; Veroustraete, F.; Feyen, J. Assessment of evapotranspiration and soil moisture content across different scales of observation. Sensors
**2008**, 8, 70–117. [Google Scholar] [CrossRef] [Green Version] - Hu, F.; Wei, Z.; Zhang, W.; Dorjee, D.; Meng, L. A spatial downscaling method for SMAP soil moisture through visible and shortwave-infrared remote sensing data. J. Hydrol.
**2020**, 590, 125360. [Google Scholar] [CrossRef] - Ferrazzoli, P.; Paloscia, S.; Pampaloni, P.; Schiavon, G.; Solimini, D.; Coppo, P. Sensitivity of microwave measurements to vegetation biomass and soil moisture content: A case study. IEEE Trans. Geosci. Remote Sens.
**1992**, 30, 750–756. [Google Scholar] [CrossRef] - Qiu, J.; Crow, W.T.; Wagner, W.; Zhao, T. Effect of vegetation index choice on soil moisture retrievals via the synergistic use of synthetic aperture radar and optical remote sensing. Int. J. Appl. Earth Obs.
**2019**, 80, 47–57. [Google Scholar] [CrossRef] - Kornelsen, K.C.; Coulibaly, P. Advances in soil moisture retrieval from synthetic aperture radar and hydrological applications. J. Hydrol.
**2013**, 476, 460–489. [Google Scholar] [CrossRef] - Sonkamble, S.; Chandra, S. GPR for earth and environmental applications: Case studies from India. J. Appl. Geophys.
**2021**, 193, 104422. [Google Scholar] [CrossRef] - Dinh, K.; Gucunski, N.; Zayed, T. Automated visualization of concrete bridge deck condition from GPR data. NDT E Int.
**2019**, 102, 120–128. [Google Scholar] [CrossRef] - Fernandes, F.M.; Pais, J.C. Laboratory observation of cracks in road pavements with GPR. Constr. Build. Mater.
**2017**, 154, 1130–1138. [Google Scholar] [CrossRef] - Lombardi, F.; Lualdi, M. Step-frequency ground penetrating radar for agricultural soil morphology characterisation. Remote Sens.
**2019**, 11, 1075. [Google Scholar] [CrossRef] [Green Version] - Chen, Y.; Zhou, Z.; Wang, J.; Zhao, Y.; Dou, Z. Quantification and division of unfrozen water content during the freezing process and the influence of soil properties by low-field nuclear magnetic resonance. J. Hydrol.
**2021**, 602, 126719. [Google Scholar] [CrossRef] - Wai-Lok Lai, W.; Dérobert, X.; Annan, P. A review of ground penetrating radar application in civil engineering: A 30-year journey from locating and testing to imaging and diagnosis. NDT E Int.
**2018**, 96, 58–78. [Google Scholar] [CrossRef] - Roth, C.H.; Malicki, M.A.; Plagge, R. Empirical evaluation of the relationship between soil dielectric constant and volumetric water content as the basis for calibrating soil moisture measurement by TDR. J. Soil Sci.
**1992**, 43, 1–13. [Google Scholar] [CrossRef] - Lunt, I.A.; Hubbard, S.S.; Rubin, Y. Soil moisture content estimation using ground-penetrating radar reflection data. J. Hydrol.
**2005**, 307, 254–269. [Google Scholar] [CrossRef] - Shapovalov, V.; Yavna, V.; Kochur, A.; Khakiev, Z.; Sulavko, S.; Daniel, P.; Kruglikov, A. Application of GPR for determining electrophysical properties of structural layers and materials. J. Appl. Geophys.
**2020**, 172, 103913. [Google Scholar] [CrossRef] - Benedetto, A. Water content evaluation in unsaturated soil using GPR signal analysis in the frequency domain. J. Appl. Geophys.
**2010**, 71, 26–35. [Google Scholar] [CrossRef] - Bablet, A.; Viallefont-Robinet, F.; Jacquemoud, S.; Fabre, S.; Briottet, X. High-resolution mapping of in-depth soil moisture content through a laboratory experiment coupling a spectroradiometer and two hyperspectral cameras. Remote Sens. Environ.
**2020**, 236, 111533. [Google Scholar] [CrossRef] - Rasol, M.A.; Pérez-Gracia, V.; Fernandes, F.M.; Pais, J.C.; Santos-Assunçao, S.; Santos, C.; Sossa, V. GPR laboratory tests and numerical models to characterize cracks in cement concrete specimens, exemplifying damage in rigid pavement. Measurement
**2020**, 158, 107662. [Google Scholar] [CrossRef] - Rasol, M.A.; Pérez-Gracia, V.; Solla, M.; Pais, J.C.; Fernandes, F.M.; Santos, C. An experimental and numerical approach to combine ground penetrating radar and computational modeling for the identification of early cracking in cement concrete pavements. NDT E Int.
**2020**, 115, 102293. [Google Scholar] [CrossRef] - Elshorbagy, A.; Parasuraman, K. On the relevance of using artificial neural networks for estimating soil moisture content. J. Hydrol.
**2008**, 362, 1–18. [Google Scholar] [CrossRef] - Bai, H.; Sinfield, J.V. Improved background and clutter reduction for pipe detection under pavement using ground penetrating radar (GPR). J. Appl. Geophys.
**2020**, 172, 103918. [Google Scholar] [CrossRef] - Szymczyk, P.; Szymczyk, M. Non-destructive building investigation through analysis of GPR signal by S-transform. Automat. Constr.
**2015**, 55, 35–46. [Google Scholar] [CrossRef] - Tosti, F.; Patriarca, C.; Slob, E.; Benedetto, A.; Lambot, S. Clay content evaluation in soils through GPR signal processing. J. Appl. Geophys.
**2013**, 97, 69–80. [Google Scholar] [CrossRef] - Ghasemi, F.S.A.; Abrishamian, M.S. A novel method for FDTD numerical GPR imaging of arbitrary shapes based on Fourier transform. NDT E Int.
**2007**, 40, 140–146. [Google Scholar] [CrossRef] - Ling, T.; Liu, H.; Gong, S.; Huang, F. Construction and application of a new biorthogonal wavelet basis for a quantitative analysis of GPR signals. J. Appl. Geophys.
**2019**, 170, 103837. [Google Scholar] [CrossRef] - Javadi, M.; Ghasemzadeh, H. Wavelet analysis for ground penetrating radar applications: A case study. J. Geophys. Eng.
**2017**, 14, 1189–1202. [Google Scholar] [CrossRef] - Lu, G.; Zhao, W.; Forte, E.; Tian, G.; Li, Y.; Pipan, M. Multi-frequency and multi-attribute GPR data fusion based on 2-D wavelet transform. Measurement
**2020**, 166, 108243. [Google Scholar] [CrossRef] - Baili, J.; Lahouar, S.; Hergli, M.; Al-Qadi, I.L.; Besbes, K. GPR signal de-noising by discrete wavelet transform. NDT E Int.
**2009**, 42, 696–703. [Google Scholar] [CrossRef] - Artagan, S.S.; Borecky, V. Advances in the nondestructive condition assessment of railway ballast: A focus on GPR. NDT E Int.
**2020**, 115, 102290. [Google Scholar] [CrossRef] - Ni, S.; Huang, Y.; Lo, K.; Lin, D. Buried pipe detection by ground penetrating radar using the discrete wavelet transform. Comput. Geotech.
**2010**, 37, 440–448. [Google Scholar] [CrossRef] - Zhang, L.; Ling, T.; Yu, B.; Huang, F.; Zhang, S. Intensive interferences processing for GPR signal based on the wavelet transform and F-K filtering. J. Appl. Geophys.
**2021**, 186, 104273. [Google Scholar] [CrossRef] - Verma, V.S.; Kumar Jha, R. Improved watermarking technique based on significant difference of lifting wavelet coefficients. Signal. Image Video Process.
**2014**, 9, 1443–1450. [Google Scholar] [CrossRef] - Li, H.; Wang, Q.; Wu, L. A novel design of lifting scheme from general wavelet. IEEE Trans. Signal Process.
**2001**, 49, 1714–1717. [Google Scholar] - Ling, T.; Zhang, L.; Liu, H.; Huang, F.; Gu, D.; Yu, B. Quantitative identification and analysis of tunnel lining defect based on the wavelet lifting scheme theory. China J. Highw. Transp.
**2019**, 32, 197–204. [Google Scholar] - Daubechies, I.; Sweldens, W. Factoring wavelet transforms into lifting steps. J. Fourier Anal. Appl.
**1998**, 4, 247–269. [Google Scholar] [CrossRef] - Tzanis, A. The curvelet transform in the analysis of 2-D GPR data: Signal enhancement and extraction of orientation-and-scale-dependent information. J. Appl. Geophys.
**2015**, 115, 145–170. [Google Scholar] [CrossRef] - Ling, T.; Zhang, L.; Huang, F.; Gu, D.; Yu, B.; Zhang, S. OMHT method for weak signal processing of GPR and its application in identification of concrete micro-crack. J. Cent. South Univ.
**2020**, 26, 3057–3065. [Google Scholar] [CrossRef] - Liao, K.; Lai, X.; Zhou, Z.; Zhu, Q. Applying fractal analysis to detect spatio-temporal variability of soil moisture content on two contrasting land use hillslopes. Catena
**2017**, 157, 163–172. [Google Scholar] [CrossRef] - Kong, Q.; Chen, H.; Mo, Y.L.; Song, G. Real-time monitoring of water content in sandy soil using shear mode piezoceramic transducers and active sensing-A feasibility study. Sensors
**2017**, 17, 2395. [Google Scholar] [CrossRef] [Green Version] - Xu, B.; Li, B.; Song, G. Active debonding detection for large rectangular CFSTs based on wavelet packet energy spectrum with piezoceramics. J. Struct. Eng.
**2013**, 139, 1435–1443. [Google Scholar] [CrossRef] - Wang, Z.; Wei, L.; Cao, M. Damage quantification with embedded piezoelectric aggregates based on wavelet packet energy analysis. Sensors
**2019**, 19, 425. [Google Scholar] [CrossRef] [Green Version] - Lu, W.; Li, P.; Chen, M.; Zhou, C.; Shu, D. Comparison of vibrations induced by excavation of deep-buried cavern and open pit with method of bench blasting. J. Cent. South Univ.
**2011**, 18, 1709–1718. [Google Scholar] [CrossRef] - Chen, G.; Li, Q.; Li, D.; Wu, Z.; Liu, Y. Main frequency band of blast vibration signal based on wavelet packet transform. Appl. Math. Model.
**2019**, 74, 569–585. [Google Scholar] [CrossRef] - Jiang, T.; Kong, Q.; Patil, D.; Luo, Z.; Huo, L.; Song, G. Detection of debonding between fiber reinforced polymer bar and concrete structure using piezoceramic transducers and wavelet packet analysis. IEEE Sens. J.
**2017**, 17, 1992–1998. [Google Scholar] [CrossRef] - Luo, T.X.H.; Lai, W.W.L.; Giannopoulos, A. Forward modelling on GPR responses of subsurface air voids. Tunn. Undergr. Space Tech.
**2020**, 103, 103521. [Google Scholar] [CrossRef] - Ercoli, M.; Di Matteo, L.; Pauselli, C.; Mancinelli, P.; Frapiccini, S.; Talegalli, L.; Cannata, A. Integrated GPR and laboratory water content measures of sandy soils: From laboratory to field scale. Constr. Build. Mater.
**2018**, 159, 734–744. [Google Scholar] [CrossRef] - Warren, C.; Giannopoulos, A.; Giannakis, I. gprMax: Open source software to simulate electromagnetic wave propagation for ground penetrating radar. Comput. Phys. Commun.
**2016**, 209, 163–170. [Google Scholar] [CrossRef] [Green Version] - Giannopoulos, A. Modelling ground penetrating radar by GprMax. Constr. Build. Mater.
**2005**, 19, 755–762. [Google Scholar] [CrossRef] - Topp, G.C. Electromagnetic determination of soil water content: Measurements in coaxial transmission lines. Water Resour. Res.
**1980**, 16, 574–582. [Google Scholar] [CrossRef] [Green Version] - Liu, J. Methods of Detection on Moisture of Railway Subgrage by Using the Ground Penetrating Radar. Ph.D. Thesis, China University of Mining and Technology, Beijing, China, 2015. [Google Scholar]
- Cao, M.; Qiao, P. Integrated wavelet transform and its application to vibration mode shapes for the damage detection of beam-type structures. Smart Mater. Struct.
**2008**, 17, 055014. [Google Scholar] [CrossRef] - Yi, T.H.; Li, H.N.; Zhao, X.Y. Noise smoothing for structural vibration test signals using an improved wavelet thresholding technique. Sensors
**2012**, 12, 11205–11220. [Google Scholar] [CrossRef] [Green Version]

**Figure 4.**A cavity model buried in the stratum for GPR forward simulation and the simulation result. (

**a**) Cavity model; (

**b**) GPR time profile of cavity model.

**Figure 5.**Results of GPR forward simulation of soils with different water contents. (

**a**) water content = 10%; (

**b**) water content = 12%; (

**c**) water content = 14%; (

**d**) water content = 16%; (

**e**) water content = 18%; (

**f**) water content = 20%; (

**g**) water content = 22%; (

**h**) water content = 24%.

**Figure 7.**Relationship between the WPEI and the soil water content. (

**a**) Changing trend of the WPEI as the relative dielectric constant increases for soils, (

**b**) changing trend of the WPEI as the electrical conductivity increases for soils.

**Figure 8.**Clay samples with different water contents. (

**a**) water content = 10%; (

**b**) water content = 12%; (

**c**) water content = 14%; (

**d**) water content = 16%; (

**e**) water content = 18%; (

**f**) water content = 20%; (

**g**) water content = 22%; (

**h**) water content = 24%.

**Figure 10.**GPR imaging results of the clay in the four model boxes. (

**a**) water content = 10%; (

**b**) water content = 12%; (

**c**) water content = 14%; (

**d**) water content = 16%; (

**e**) water content = 18%; (

**f**) water content = 20%; (

**g**) water content = 22%; (

**h**) water content = 24%.

**Figure 11.**GPR time profiles after conventional processing. (

**a**) water content = 10%; (

**b**) water content = 12%; (

**c**) water content = 14%; (

**d**) water content = 16%; (

**e**) water content = 18%; (

**f**) water content = 20%; (

**g**) water content = 22%; (

**h**) water content = 24%.

**Figure 12.**GPR single-channel signals of four model boxes with different water contents. (

**a**) Model box 1; (

**b**) Model box 2; (

**c**) Model box 3; (

**d**) Model box 4.

**Figure 13.**Regression relationship between the WPEI obtained using the WPEA method and the soil water content in the four model boxes. (

**a**) Model box 1; (

**b**) Model box 2; (

**c**) Model box 3; (

**d**) Model box 4.

**Figure 14.**Regression relationship between the WPEI obtained by using db6 wavelet packet transform and the soil water content in four model boxes. (

**a**) Model box 1; (

**b**) Model box 2; (

**c**) Model box 3; (

**d**) Model box 4.

Water Content | 10% | 12% | 14% | 16% | 18% | 20% | 22% | 24% |
---|---|---|---|---|---|---|---|---|

Relative dielectric constant | 5.343 | 6.115 | 6.983 | 7.941 | 8.987 | 10.116 | 11.325 | 12.611 |

Conductivity (μS/cm) | 88.681 | 126.085 | 180.261 | 258.134 | 368.670 | 523.129 | 735.272 | 1021.533 |

Electromagnetic wave velocity (m/ns) | 0.129 | 0.121 | 0.113 | 0.106 | 0.100 | 0.094 | 0.089 | 0.084 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Zhang, S.; Zhang, L.; Ling, T.; Fu, G.; Guo, Y.
Experimental Research on Evaluation of Soil Water Content Using Ground Penetrating Radar and Wavelet Packet-Based Energy Analysis. *Remote Sens.* **2021**, *13*, 5047.
https://doi.org/10.3390/rs13245047

**AMA Style**

Zhang S, Zhang L, Ling T, Fu G, Guo Y.
Experimental Research on Evaluation of Soil Water Content Using Ground Penetrating Radar and Wavelet Packet-Based Energy Analysis. *Remote Sensing*. 2021; 13(24):5047.
https://doi.org/10.3390/rs13245047

**Chicago/Turabian Style**

Zhang, Sheng, Liang Zhang, Tonghua Ling, Guihai Fu, and Youlin Guo.
2021. "Experimental Research on Evaluation of Soil Water Content Using Ground Penetrating Radar and Wavelet Packet-Based Energy Analysis" *Remote Sensing* 13, no. 24: 5047.
https://doi.org/10.3390/rs13245047