Geophysical data, and their integration with hydro-biogeochemical point-scale measurements, are key to quantifying the subsurface structure and characteristics with relevant spatial resolution and coverage as well as for imaging the spatiotemporal distribution of water and solute properties. The Environmental Geophysics group develops and applies novel strategies to estimate properties and processes in the subsurface, and their interaction with above-ground processes, using a variety of geophysical techniques, including complex electrical conductivity tomography, electromagnetic induction methods, ground-penetrating radar, and well logging. These datasets are integrated into physical models and/or with point-scale measurements to help improve understanding and quantification of subsurface hydro-biogeochemical processes. Further, by coupling below-ground investigation with remote- sensing data, we improve the understanding of interactions between above- and below-ground properties critical to predictive understanding of ecosystem functioning and for potentially scaling results from intensive site to larger spatial coverage. These strategies enabled new understanding of subsurface processes in various environments, including imaging of dissolved CO2 plumes and related geochemical concentrations in freshwater aquifers, mapping biogeochemical hotspot in alluvial deposits, quantification of active layer and permafrost distribution and ice-content, and estimation of hydraulic and thermal parameters through joint geophysical and hydrological inverse modelling.
The Environmental Geophysics group has extensive expertise in developing and using advanced remote-sensing techniques to investigate environmental interactions and phenomena. Our expertise ranges from data analysis of several remote-sensing data sources including multi- and hyperspectral, thermal imagery, and LiDAR point-cloud data to the development of algorithms based on signal/image processing, machine learning, and data fusion. These analytical methods are currently being applied to several projects focused on the characterization of natural ecosystems, such as alpine watersheds, arctic tundra, and tropical forest, to retrieve environmental variables critical for quantifying ecosystem behavior and response to stressors.Current projects also evaluate managed (agricultural) ecosystems, applying novel approaches based on remote-sensing techniques to advance sustainable practices, and improve crop monitoring and performance assessment (such as the estimation of plant density and vigor, or greenhouse gases and carbon footprint). The group has extensive experience using satellite and airborne remote-sensing data across scales. The group’s assets include several UAV platforms and sensors (including multispectral and thermal sensors, and high-resolution RGB cameras) for high-resolution investigations and field spectroradiometer for in situ analysis.
Sensing subsurface physical, hydrological, and biogeochemical properties and processes at field scale with relevant accuracy, resolution, and coverage is critical for improving our ability to predict and manage both natural and managed ecosystems. Leveraging recent advances in the field of physics, we develop novel sensing systems for field investigation of subsurface properties and processes. Some examples include:
- dense, low-cost sensor networks for capturing vertically resolved profiles of soil properties at an unprecedented number of locations;
- integrated imaging and modeling toolbox for root development;
- novel fiber-optic distributed sensors to autonomously measure subsurface temperature, strain, acoustic signals, and geochemical properties; and
- UAV-based electromagnetic sensor to sense subsurface electrical conductivity.
These developments take advantage of specialized fabrication and testing facilities at the EESA Geosciences Measurement Facility, and are linked to advances in signal processing, data-model integration, and data management and visualization.
Key Examples
- Distributed Temperature Profiling (DTP) system (website soon), NGEE-Arctic
- EcoSENSE
- UAV-mounted Passive ElectroMagnetic (EM) Sensor for spatiotemporal Imaging of Shallow Subsurface Properties
- TERI
Laboratory research based on column-to-mesoscale platforms provide key capabilities helping our group to understand and build petrophysical correlations and models for mechanistic process understanding. In our petrophysical studies, e jointly use multiple geophysical approaches across scales, including spectral induced polarization, distributed fiber-optic sensing, electrical resistivity tomography, electrochemical impedance spectroscopy, and advance geochemical and imaging methods such as SEM, XRD, and x-ray CT.
The subsurface and its response to natural phenomena or perturbations are sensed using geophysical methods that rely on instrumentation and numerical methods to reconstruct an image of subsurface structure and characteristics. In addition, capturing variation over time enables us to assess the movement of solute in the subsurface. We are using and developing a variety of numerical methods for inverse modeling or parameter estimation of soil physical, hydro-geochemical, and mechanical properties over space and/or time. Our stochastic- and deterministic-based estimation approaches involve various datasets as well as hydrological and geophysical models. Examples include imaging of solute spatiotemporal distribution using electrical conductivity tomography, joint inversion of geophysical and hydrological data for estimation of soil hydraulic parameters, and estimation of organic matter content from temperature and electrical conductivity time-series.
Artificial Intelligence (AI) and machine learning (ML) technologies play a key role in extracting spatiotemporal information from a large volume of data obtained by geophysical approaches and remote sensing datasets. Environmental data pose unique challenges in AI/ML since environmental data tend to have large noise or natural fluctuations; there tends to be many large indirect data and sparse direct data; large indirect data with sparse direct data; data are often highly correlated with space and time; and the amount of training data is always small. We are developing a variety of AI/ML tools to address these challenges, including Bayesian methods for integrating multiscale data and unsupervised learning (e.g., clustering) for capturing co-variability among datasets. Our particular focus has been to exploit above- and belowground co-variability of ecosystems (such as soil-plant interactions) through coupling geophysics and remote-sensing data, and also the covariability among hydro-geochemical-biological properties. Examples include ecosystem zonation (i.e., spatial clustering) in the Arctic permafrost environment, soil moisture estimation through remote-sensing data, geophysical estimation of reactive faces and geochemical hotspots, and multiscale data integration of radiation monitoring data.