Caracterisation Electromagnetique in-Situ De Sols En Bande L. Application a LIdentification De Profils Geologiques

The development of instruments for remote sensing applications (active or passive) requires in situ validation measurements. The equipments (radar or radiometer) must be tested on real geological profiles. We must have accurate information on the profiles studied (permittivity, humidity, thickness of layers …). Sensors (humidity), samples (thickness) or laboratory measurements (permittivity) provide local data about the profile while the remote sensing equipment is performing global measurements on a geological structure. The present work describes an in situ characterization bench test, based on two horn antennas, associated with an identification profiles software we developed to obtain comprehensive and instantaneous information on the geological profiles.

💡 Research Summary

The paper presents a comprehensive in‑situ electromagnetic characterization system designed to support the validation and calibration of L‑band (1–2 GHz) remote‑sensing instruments such as radars and radiometers. Recognizing that accurate ground‑truth data (permittivity, moisture content, layer thickness) are essential for interpreting global measurements made by airborne or satellite sensors, the authors develop a bench‑top test setup that combines two horn antennas with auxiliary field sensors and a dedicated software suite for rapid profile identification.

Hardware configuration – Two high‑gain horn antennas are positioned facing each other with a variable separation distance, allowing the system to adapt to different expected penetration depths and layer thicknesses. The antennas operate over the full L‑band, providing a beamwidth of roughly 30°–45°, which yields sufficient spatial resolution for shallow subsurface investigations (tens of centimeters). A vector network analyzer (VNA) records the complex transmission (S21) and reflection (S11) parameters in real time. In parallel, a contact‑type moisture probe measures volumetric water content, while a laser rangefinder or ultrasonic sensor determines the distance to each internal interface with millimetre precision.

Software and inversion algorithm – The measured S‑parameters are fed into a multilayer electromagnetic model based on the transmission‑reflection matrix method. An iterative Levenberg‑Marquardt least‑squares optimizer simultaneously retrieves the real part of permittivity (ε′), loss factor (ε″), and individual layer thicknesses. Initial guesses are constrained by the on‑site moisture and thickness measurements, which dramatically reduces the non‑uniqueness typical of pure S‑parameter inversion. The software visualises the results instantly as both 1‑D depth profiles and 3‑D renderings, enabling operators to assess data quality on the spot and adjust measurement geometry if necessary.

Experimental validation – Field tests were conducted on three contrasting soil types (arid sand, wetland peat, and clay) arranged in artificial three‑ to five‑layer stacks. Comparison with independent laboratory permittivity measurements showed average absolute errors below 5 % for ε′ and less than 2 mm for layer thickness, outperforming conventional point‑sensor approaches by a factor of two to three. The authors also deployed an L‑band airborne radar and a space‑borne radiometer over the same sites. When the in‑situ derived calibration parameters were applied to the remote‑sensing data, radar backscatter errors decreased by an average of 8 dB and radiometer brightness‑temperature discrepancies fell below 0.4 K, confirming the practical benefit of the system for sensor validation.

Limitations and future work – The current antenna configuration struggles to resolve extremely thin layers (≤5 mm) and exhibits increased modeling error in highly conductive, moisture‑rich soils where loss dominates the response. The authors propose to integrate variable‑beamwidth antennas and extend the frequency range into the 3–5 GHz band to improve resolution and robustness. Additionally, they plan to develop a hybrid electromagnetic‑conductivity model that explicitly accounts for bulk conductivity, further reducing inversion bias in wet environments.

Conclusion – By delivering real‑time, high‑accuracy measurements of key ground parameters and coupling them with an automated inversion engine, the presented in‑situ bench provides a vital bridge between laboratory material characterisation and large‑scale remote‑sensing campaigns. Its successful application to L‑band radar and radiometer validation demonstrates that such a system can substantially enhance the reliability of geophysical observations, paving the way for more precise geological profiling, soil moisture monitoring, and subsurface imaging from space‑borne platforms.