The mapping of moisture content, composition and texture of soils is attracting a growing interest, in particular with the goal of evaluating threats to soil quality, such as soil salinization. Fast non-invasive geophysical surveys are often used in this context. The aim of this work was to study constitutive models that can be used to parameterize electrical conductivity and permittivity starting from a unifying conceptual approach, and to evaluate whether the information carried by one measurement type can be used to identify soil parameters that are then used to predict the other geophysical quantity. To this end, a recently-developed constitutive model was here extended and modified to consider also the grain surface conductivity, a critical component in most natural situations. The extended model was successfully tested against laboratory measurements. In addition, the new model was compared against five other equations that use similar soil parameterizations. It was concluded that only three out of the five selected models yield similar predictions, while the remaining two predict a different geophysical response for the same soil texture. Following this analysis, a methodology was developed to estimate soil salinity starting from the simultaneous measurements of bulk electrical conductivity and permittivity and validate this methodology against laboratory experiments. The method is valid in situations where the conductivity of the pore-water remains approximately constant during the measurement period. Key features of the approach proposed to map soil salinization are (i) simplicity, (ii) absence of fitting parameters and (iii) the fact that moisture content does not need to be measured or estimated independently. The methodology was tested on a large number of soil samples and proved robust and accurate.