ALS-derived raster visualization techniques have become common in recent years, opening up new possibilities for subtle landform detection in earth sciences and archaeology, but they have also introduced confusion for users. As a consequence, the choice between these visualization techniques is still mostly supported by empirical knowledge. Some attempts have been made to compare these techniques, but there is still a lack of analytical data. This work proposes a new method, based on gradient modelling and spatial statistics, to analytically assess the efficacy of these visualization techniques. A selected panel of outstanding visualization techniques was assessed first by a classic non-analytical approach, and secondly by the proposed new analytical approach. The comparison of results showed that the latter provided more detailed and objective data, not always consistent with previous empirical knowledge. These data allowed us to characterize with precision the terrain for which each visualization technique performs best. A combination of visualization techniques based on DEM manipulation (Slope and Local Relief Model) appeared to be the best choice for normal terrain morphometry, occasionally supported by illumination techniques such as Sky-View Factor or Negative Openness as a function of terrain characteristics.

In near-surface geophysics, small portable loop-loop electromagnetic induction (EMI) sensors using harmonic sources with a constant and rather small frequency are increasingly used to investigate the electrical properties of the subsurface. For such sensors, the influence of electrical conductivity and magnetic permeability on the EMI response is well-understood. Typically, data analysis focuses on reconstructing an electrical conductivity model by inverting the out-of-phase response. However, in a variety of near-surface applications, magnetic permeability (or susceptibility) models derived from the in-phase (IP) response may provide important additional information. In view of developing a fast 3D inversion procedure of the IP response for a dense grid of measurement points, we first analyze the 3D sensitivity functions associated with a homogeneous permeable half-space. Then, we compare synthetic data computed using a linear forward-modeling method based on these sensitivity functions with synthetic data computed using full nonlinear forward-modeling methods. The results indicate the correctness and applicability of our linear forward-modeling approach. Furthermore, we determine the advantages of converting IP data into apparent permeability, which, for example, allows us to extend the applicability of the linear forward-modeling method to high-magnetic environments. Finally, we compute synthetic data with the linear theory for a model consisting of a controlled magnetic target and compare the results with field data collected with a four-configuration loop-loop EMI sensor. With this field-scale experiment, we determine that our linear forward-modeling approach can reproduce measured data with sufficiently small error, and, thus, it represents the basis for developing efficient inversion approaches.

We have developed an inversion process of electromagnetic induction (EMI) data based on a two-step approach with 1D inversion of the entire studied surface and a fast 3D inversion applied over limited areas. This process is similar to that formerly used in resistivity prospection. For the study of soil (environmental, engineering, or archaeological explorations), low-frequency electromagnetic instruments (referred to as Slingram EMI) have highly useful specificities. They are light, are easy to move in the field, and can simultaneously measure the ground’s electric conductivity and magnetic susceptibility; they have thus been used to map these properties over large surface areas, within relatively short periods of time, and at reasonable expense. The possibility of combining several coil geometries has opened up the potential for multidepth techniques and systematic 1D inversion, which are found to be sufficiently revealing to allow larger portions of surveyed areas to be analyzed. In the “targeted areas” selected for 3D inversion, the geometries of the 3D features and the resistivity and/or susceptibility contrasts are determined. This step is based on the method of moments, where only 3D heterogeneities are meshed, and only a small number of major characteristics, such as contrast, thickness, width, etc., are sought. We first applied this process to synthetic data, then to data acquired at an experimental test site, and finally to field cases. The rapid 3D inversion complements the 1D inversion by solving a series of issues: correction for the apparent anisotropy generated by the instrument configuration, multiarched anomalies, precise location of lateral changes, and determination of the properties contrasts. Our inversion results highlighted the importance of the instrument geometry. We also have determined that apparent magnetic susceptibility data can be more appropriate for the determination of the volume of man-made features and can be highly complementary to conductivity data.