Refine
Has Fulltext
- no (4)
Year of publication
- 2017 (4) (remove)
Language
- English (4)
Is part of the Bibliography
- yes (4)
Keywords
- African Humid Period (1)
- Baragoi (1)
- Be-10 (1)
- erosion (1)
- northern Kenya Rift (1)
- paleo-delta (1)
Institute
Direct current systems employing a kinematic surveying strategy allow to analyze the electrical resistivity of the subsurface for large areas (i.e., several hectares). Typical applications are found in precision agriculture, archaeological prospecting and soil sciences. With the typical survey setting, the collected data sets are often characterized by a rather high level of noise and a rather coarse lateral sampling compared to data acquired with fixed electrodes. We therefore present an efficient one-dimensional inversion approach in which we put special attention on modeling the effects of noise. We apply this method to data recorded with a five-offset equatorial dipole-dipole system employing rolling electrodes. By performing several synthetic tests with realistic noise levels, we found that the considered five-configuration soundings allow for a reliable imaging of two-layer cases in the uppermost two meters of the subsurface, where the subsurface can be assumed to follow a horizontally layered geometry within 3 m around the system. By analyzing the corresponding sensitivity functions, we also show that the equatorial dipole-dipole array is relatively well suited for a 1D inversion approach compared to standard in-line electrode arrays. To illustrate this aspect, we show that our method can provide results similar to those obtained with a 2D Wenner imaging procedure for data recorded across a well-constrained 2D target. We finally apply our method to a large five-offset data set acquired in an agricultural study. The final pseudo-3D model of electrical resistivity is in accordance with borehole data available for the surveyed area. Our results demonstrate the applicability and the versatility of the presented inversion approach for large-scale data sets as they are typically collected with such rolling electrode systems. (C) 2017 Elsevier B.V. All rights reserved.
Electromagnetic induction (EMI) sensors using sufficiently low-frequency harmonic sources and sufficiently small loop separations operate in the low-induction-number (LIN) domain for a relatively wide range of background conductivity. These systems are used in diverse near-surface investigations including applications from soil sciences, hydrology, and archaeology. The special case of portable multiconfiguration EMI sensors operating at frequencies <= 20 kHz offers the possibility of using a fast linear deconvolution method to interpret multichannel data sets in three dimensions. Here, we have developed a fast 3D inversion/deconvolution method regularized with 3D smoothness constraints and formulated in the hybrid spectral-spatial domain. Compared with other linear approaches, the spectral-spatial domain formulation significantly reduces the computational cost of the processing and opens the door for real-time 3D interpretation of large data sets consisting of more than 100,000 data points. First, we test our proposed algorithm on synthetic data sets computed with the full Maxwell theory. Then, we apply our method to a real four-configuration EMI data set acquired to map the thickness of peat layers embedded in a sandy environment. For the synthetic and the field example, we compared our result with the result obtained using a standard point-by-point 1D nonlinear inversion approach. This comparison demonstrates that the proposed methodology provides superior lateral resolution compared with the 1D nonlinear inversion, at the same time significantly reducing the computational cost of the processing.
The African Humid Period (AHP) between similar to 15 and 5.5 cal. kyr BP caused major environmental change in East Africa, including filling of the Suguta Valley in the northern Kenya Rift with an extensive (similar to 2150 km(2)), deep (similar to 300 m) lake. Interfingering fluvio-lacustrine deposits of the Baragoi paleo-delta provide insights into the lake-level history and how erosion rates changed during this time, as revealed by delta-volume estimates and the concentration of cosmogenic Be-10 in fluvial sand. Erosion rates derived from delta-volume estimates range from 0.019 to 0.03 mm yr(-1). Be-10-derived paleo-erosion rates at similar to 11.8 cal. kyr BP ranged from 0.035 to 0.086 mm yr(-1), and were 2.7 to 6.6 times faster than at present. In contrast, at similar to 8.7 cal. kyr BP, erosion rates were only 1.8 times faster than at present. Because Be-10-derived erosion rates integrate over several millennia; we modeled the erosion-rate history that best explains the 10Be data using established non-linear equations that describe in situ cosmogenic isotope production and decay. Two models with different temporal constraints (15-6.7 and 12-6.7 kyr) suggest erosion rates that were 25 to 300 times higher than the initial erosion rate (pre-delta formation). That pulse of high erosion rates was short (similar to 4 kyr or less) and must have been followed by a rapid decrease in rates while climate remained humid to reach the modern Be-10-based erosion rate of,similar to 0.013 mm yr(-1). Our simulations also flag the two highest Be-10-derived erosion rates at 11.8 kyr BP related to nonuniform catchment erosion. These changes in erosion rates and processes during the AHP may reflect a strong increase in precipitation, runoff, and erosivity at the arid-to-humid transition either at 15 or similar to 12 cal. kyr BP, before the landscape stabilized again, possibly due to increased soil production and denser vegetation.