Environmental and Engineering Geophysics
Environmental and Engineering Geophysics
Embedded Files
Environmental and Engineering geophysics uses various geophysical methods to address critical environmental and geotechnical issues (e.g., identification of sinkholes, estimation of soil moisture, soil contamination, characterization of groundwater flow, and determination of water table). Dr. Askari's current research on environmental and engineering geophysics focuses on (1) developing an intelligent drone-based radar and (2) site characterization and non-destructive testing using ultra-high-frequency seismic waves.
Conventional geophysical radar measurements are ground-based, termed as ground penetrating radar (GPR). GPR data acquisitions are usually time-consuming, restricting their applications for the environmental monitoring of large areas (e.g., mega-farms). In addition, the conventional GPR processing methods are not intelligent, not allowing real-time investigations of environmental issues. Dr. Askari's team with the collaboration of Prof. Zekavat’s team from the Worcester Polytechnic Institute, Worcester, MA is developing an enhanced drone-based radar, in which real-time measurements become feasible through machine learning techniques. This research aims at estimating soil moisture to the root zone for mega farms. Using soil moisture maps obtained from our method, we can optimize irrigation for big farms. This research is funded by the United States Department of Agriculture (USDA).
The schematic of the intelligent drone-based radar system for soil moisture estimation.
The near S-wave velocity is a critical geophysical property used to determine fundamental geotechnical parameters such as the soil’s shear modulus, natural frequency, and effective vertical stress. The S-wave’s anisotropy can be used to identify near-surface fractures and characterize groundwater flows. In near-surface studies, the surface wave method (SWM) has been widely used to estimate the S-wave velocity. SWM uses the dispersion properties (either phase or group velocity) of Love or Rayleigh waves to obtain an S-wave velocity model of the subsurface via an inversion procedure. My team has conducted extensive research to develop novel approaches to enhance the processing and penetration depth of SWM with a focus on environmental applications.
Currently, my team is working on ultrahigh frequency SWM (frequency range from 100 Hz to a few kilohertz). The ultrahigh-frequency SWM will allow super high-resolution S-wave velocity imaging up to 1m below the surface, which will have broad applications in agriculture and forestry.
In this inter-institutional collaborative research with Dr. Zekavat’s team from Worcester Polytechnic Institute (WPI), we developed an intelligent drone-based GPR system enhanced by machine learning to estimate root-zone soil moisture across large agricultural fields. Using the gprMax electromagnetic simulator, our team generated realistic synthetic GPR datasets that emulate multilayer soil structures, which were later calibrated with real measurements collected at WPI’s SoilX Laboratory. The integrated framework combines feature-engineered GPR data with advanced ML models, such as Random Forest and Neural Networks, to accurately predict soil moisture and subsurface layer thickness with centimeter-level precision. The results demonstrate that our AI-enhanced GPR system significantly outperforms traditional remote sensing methods, enabling the creation of 3D root-zone moisture maps that support optimized irrigation management, efficient water use, and enhanced crop productivity in megafarms.
(a) Model for synthetic GPR dataset creation. (b) Soil Hydraulic model indicating water flow and layer distributions. (c) CLM Model indicating soil horizons O, A, B, C. From Namdari et al. (2024).
In this inter-institutional collaborative research, which received the IEEE ORSS Best Paper Award, we developed a data-driven GPR system for accurate estimation of soil water content at multiple root-zone depths. The system employs a Stepped-Frequency Continuous-Wave (SFCW) radar operating between 0.4–2.0 GHz and was validated through an extensive field campaign at WPI using in-situ soil moisture probes. By integrating machine learning algorithms, including Linear Regression, Lasso, Random Forest, and XGBoost, we achieved soil moisture prediction errors below 6% across four depths. This study demonstrates the capability of AI-enhanced GPR to deliver non-invasive, high-resolution soil moisture mapping, paving the way for efficient irrigation management and sustainable agriculture.
Frequency analysis scatter plots by Lasso feature selection by depth over all 4096 SFCW frequencies. Percentage of trials with non-zero coefficients across 1050 model training stages. From Filardi et al. (2023).
In Jeng et al., we assessed the capability of SWM to forecast near-surface fractures within low-permeable bedrocks. We showed that by optimizing data acquisition and incorporating higher modes of the Rayleigh and Love wave phase velocities in inversion, it is possible to study the S-wave velocity within the bedrock (i.e., 0-90 m below the surface). Moreover, by considering anisotropic inversion, we calculated seismic radial anisotropy of subsurface layers. Seismic radial anisotropy is defined concerning the difference between the velocity of a vertically polarized S-wave (SV, from the dispersion analysis of the Rayleigh waves) and one polarized horizontally (SH, from the dispersion analysis of the Love wave). We evaluated the correlation of seismic radial anisotropy with near-surface fractures at two sites with different bedrock lithologies (one metamorphic-igneous and the other sedimentary). The seismic radial anisotropies at these two sites showed a strong correlation with near-surface fractures, and thus, can be used as a strong attribute to identify near-surface fractures.
(a and c) VSH, VSV and VS (average of VSH and VSV) profiles and (b and d) radial anisotropy parameters for two seismic lines acquired in a hydrological site with a fractured bedrock respectively. (C1) and (C2) the locations of fractures within two nearby wells respectively. (e) and (f) probability of fracture occurrence with the cutoff value of 0.2 for the two lines respectively (from Jeng et al.).
Page updated
Google Sites
Report abuse