Modelling soil erosion is hampered by the complexity intrinsic to the short-term variability and dynamics between its physical properties, especially on the soil surface layer. Soil crust development, in particular, is one of the most important factors controlling runoff and erosion, but its properties are difficult to assess. Accurate and spatially distributed data, particularly at fine scale, are increasing in demand by the scientific community to tackle this issue. Remote sensing with multispectral imaging could partly meet this demand, but another series of issues related to the complexity of the signal to be detected must be solved to provide qualitative data on specific target soil variables.

The work of this thesis focused on the detection capability of the development dynamics of soil physical crust. The general objective was to understand whether soil physical crust development could be detected and modelled by means of multispectral remote sensing, using low-altitude uncrewed aerial vehicles (UAS), with a case study on loamy soils of the Belgian loess belt. This work has been structured on an ideological pathway to test a set of sensors and their detection capability for soil properties, starting from laboratory analysis and progressively scaling to the remote sensing domain.

After a preliminary assessment of spectral sensor technology and analytical methods, this thesis proposed an innovative methodology to model soil physical crust development, as defined by the spectral changes caused on the soil surface by cumulative amounts of rainfall kinetic energy. This was achieved by developing a data processing protocol for multispectral imaging in the visible and near-infrared range. By validating the laboratory methodology in real field conditions, the results offered some evidence that it is possible to map the development of soil physical crust with a remote sensing application. The last part of this thesis tackled a critical issue with the negative effects of soil anisotropy on multispectral imaging, proposing a semi-empirical correction methodology with good preliminary results, and easing the way for wide scale UAS-based assessment of soil properties with multispectral imaging.