Monitoring of Temporal Variations in Shallow Soil Water Content using Multi-Frequency GPR Common Mid-Point Soundings
Ground penetrating radar (GPR) common mid-point (CMP) soundings provide a non-invasive method of monitoring vertical variations in near-surface soil water content. This process is achieved by determining the velocity structure corresponding to the normal move-out (NMO) of hyperbolic reflection events in the CMP sounding. An estimate of volumetric soil water content can be obtained by using an appropriate petrophysical relationship. We have collected CMP soundings using multiple antenna frequencies (e.g., 225, 450 and 900 MHz) at three locations with different soil textures (e.g., sand, sandy loam and silt loam) for a complete annual cycle characterized by wet, dry and frozen soil conditions. Soundings were collected at a fixed site at each textural location with temporal measurement intervals ranging from one to four weeks. The application of multiple antenna frequencies permits the examination of the trade-off between depth of investigation and soil water content measurement resolution. CMP soundings were used to construct time-lapse vertical soil water content models using a NMO semblance analysis procedure. Temporal variations in soil water content distribution in upper 2-3 m were compared between the different antenna frequencies and textural locations to examine robustness of GPR technique for characterizing and quantifying vadose zone hydrological conditions. Preliminary results show the ability of GPR to provide high resolution soil water content information within the shallow vadose zone.
Assessment of Multi-frequency Electromagnetic Induction for Determining Soil Moisture Patterns at the Hillslope Scale
We present an assessment of electromagnetic induction (EM) as a potential rapid and non-invasive method to map soil moisture patterns at the Panola (GA, USA) hillslope. We address the following questions regarding the applicability of EM measurements for hillslope hydrological investigations: (1) Can EM be used for soil moisture measurements in areas with shallow soils?; (2) Can EM represent the temporal and spatial patterns of soil moisture throughout the year?; and (3) can multiple frequencies be used to extract additional information content from the EM approach and explain the depth profile of soil moisture? We found that the apparent conductivity measured with the multi-frequency GEM-300 was linearly related to soil moisture measured with an Aqua-pro capacitance sensor below a threshold conductivity and represented the temporal patterns in soil moisture well. During spring rainfall events that wetted only the surface soil layers the apparent conductivity measurements explained the soil moisture dynamics at depth better than the surface soil moisture dynamics. All four EM frequencies (7290, 9090, 11250, and 14010 Hz) were highly correlated and linearly related to each other and could be used to predict soil moisture. This limited our ability to use the four different EM frequencies to obtain a soil moisture profile with depth. The apparent conductivity patterns represented the observed spatial soil moisture patterns well when the individually fitted relationships between measured soil moisture and apparent conductivity were used for each measurement point. However, when the same (master) relationship was used for all measurement locations, the soil moisture patterns were smoothed and did not resemble the observed soil moisture patterns very well. In addition, the range in calculated soil moisture values was reduced compared to observed soil moisture. Part of the smoothing was likely due to the much larger measurement area of the GEM-300 compared to the Aqua-pro soil moisture measurements.
Ground penetrating radar (GPR) evidence for variations in free phase carbon gas accumulation as a function of peatland landforms: a comparison between near-crest bogs and mid-slope lawns
Northern peatlands serve as atmospheric sources of biogenic free-phase gas (FPG) produced under anaerobic conditions below the water table (mostly methane and carbon dioxide). Recent evidence suggest that FPG accumulates in the subsurface under confining layers and is released during sudden ebullition events, often triggered by sudden drops in atmospheric pressure. Accurate quantification of the impact of FPG releases on the global carbon budget is needed given recent observations of increasing atmospheric methane concentrations. One important step towards understanding the dynamics of FPG in peatlands is to investigate whether certain peatland landforms (i.e. areas with significantly different vegetation patterns) may be more conducive to FPG accumulation and/or release. Additionally, it is important to determine the vertical distribution of FPG within the peat soil and the potential role of peat stratigraphy on gas accumulation and release. In this study, we used common mid-point (CMP) velocity surveys to predict vertical profiles of FPG accumulations by comparing two different peatland landforms: historically forested near-crest bogs and non- forested mid-slope lawns in the Glacial Lake Agassiz Peatland of Minnesota, USA. We show that there is a statistically significant difference in electromagnetic (EM) wave velocities calculated over gas-rich intervals in the peat strata compared to gas-poor intervals. Common-offset radar profiles identified laterally continuous woody confining layers responsible for FPG accumulation. Chaotic GPR facies containing diffraction hyperbolae at the forested near-crest sites are interpreted as deformation of the peat matrix due to FPG accumulation and/or peat fabric disturbance during FPG release events. In contrast, non-forested mid-slope lawn sites, are characterized by planar GPR facies with no evidence of peat fabric disturbance and small relative changes in interpreted EM velocity distribution along the peat column. Using the complex refractive index method (CRIM), we estimate gas content from the 1D velocity profiles. These estimates are coupled with direct gas-sampling from the zones below woody layers to allow for the volumetric calculation of methane fraction within the peat strata. This study demonstrates the potential of 1-D and 2-D GPR methods, to quickly and non-invasively identify potential areas of FPG accumulation in peatlands.
Spatially Detailed Porosity Prediction From Airborne Electromagnetics and Sparse Borehole Fluid Sampling
Sub-surface porosity is of importance in estimating fluid contant and salt-load parameters for hydrological modelling. While sparse boreholes may adequately sample the depth to a sub-horizontal water-table and usually also adequately sample ground-water salinity, they do not provide adequate sampling of the spatial variations in porosity or hydraulic permeability caused by spatial variations in sedimentary and other geological processes.. We show in this presentation that spatially detailed porosity can be estimated by applying Archie's law to conductivity estimates from airborne electromagnetic surveys with interpolated ground-water conductivity values. The prediction was tested on data from the Chowilla flood plain in the Murray-Darling Basin of South Australia. A frequency domain, helicopter-borne electromagnetic system collected data at 6 frequencies and 3 to 4 m spacings on lines spaced 100 m apart. This data was transformed into conductivity-depth sections, from which a 3D bulk-conductivity map could be created with about 30 m spatial resolution and 2 to 5 m vertical depth resolution. For that portion of the volume below the interpolated water-table, we predicted porosity in each cell using Archie's law. Generally, predicted porosities were in the 30 to 50 % range, consistent with expectations for the partially consolidated sediments in the floodplain. Porosities were directly measured on core from eight boreholes in the area, and compared quite well with the predictions. The predicted porosity map was spatially consistent, and when combined with measured salinities in the ground water, was able to provide a detailed 3D map of salt-loads in the saturated zone, and as such contribute to a hazard assessment of the saline threat to the river.
A simple GPR full-waveform inversion method for investigating fractured rock hydrology
The 3-D geometry of fractures, along with the type and distribution of the fill material, substantially impacts bulk hydraulic properties. Coring is largely inadequate for a complete hydrologic characterization of fractured rock, and minimally invasive methods are desirable. A method is described for full-waveform inversion of GPR transmission data to quantitatively determine the fracture aperture and electromagnetic properties of the fill, based on a thin-layer model. This model is based on a plane wave assumption and assumes idealized geometry of the fracture, but it allows for quick exploration of the error space in inversion. Analysis of this method on synthetic data indicates that the properties of fractures with an aperture of greater than 5% of the dominant wavelength can be determined, when prior knowledge of the fracture orientation and electromagnetic properties of the background exist. This inversion method is applied to a study of the interaction of invasive brush vegetation with the shallow fractured rock hydrology of the Edward Aquifer of central Texas, USA. The transmission data are acquired with the transmitting antenna placed on the wall of a trench and the receiver moved out along the surface, and the background properties and fracture orientations are constrained by 3-D polarimetric GPR reflection data. The inversion results from field data show consistency with the location of fractures from reflection data.
Inversion of dispersive GPR data recorded across precipitation and thawing induced waveguides
High frequency GPR is particularly well suited for monitoring the shallow subsurface due to its non-invasive nature and ability to measure the soil water content. In case of precipitation or thawing, a low velocity waveguide can be induced due to the strong influence of the change in water content on the radar velocity. In this way, the lower substrate medium has a lower permittivity than the middle waveguide layer and causes total internal reflection when the angle of incidence at both interfaces is larger than the critical angle which enables the signal to propagate over relatively large distances. Low-velocity waveguides are induced by thawing of frozen sand and by precipitation events. In both cases, the waveguide properties can be obtained by calculating phase-velocity spectra, followed by picking dispersion curves from the maxima in the spectra. The picked dispersion curve is then inverted for a single- or multi-layer subsurface model; inversion involves adjusting the model parameters until the difference between the picked dispersion curve and the model-predicted dispersion curve is minimized. Here, we show that a waveguide develops after a significant precipitation event soaks a dry surface layer on a test site in New England. The newly wet surface layer has a higher relative permittivity and associated lower velocity than the immediately underlying dry material. TE and TM data collected after the rainfall show clear dispersion. A joint inversion of the TE and TM fundamental modes provides more reliable estimates of the medium parameters than separate inversions. Analysis of another GPR data set collected after the soaking of an existing wet layer show the presence of 4 TE and 4 TM modes. Including higher order modes in the joint TE- TM inversion resulted in better constrained models than fundamental mode inversion. Common mid-point gathers using high-frequency 900 MHz antennas were collected on a test site in Waterloo. Surveys were conducted during the seasonal thaw. The low velocity waveguide, which was caused by thawing of the shallow part of the frozen ground, shows clear dispersion of the fundamental TE and TM mode for the broadside and endfire antenna configurations, respectively. Here, a joint inversion was carried out to invert for the medium properties.