The Zona Bananera region has been the main agricultural zone since the last century in the north of Colombia; soils affected by salts (SAS) are the most important soil degradation problem in all Colombian Caribbean coast. However, the cost and dispensable of conventional soil survey methodologies limited the implementation of SAS detailed maps in the region. In this study, the soil and landscape relationship was implemented more than a tool to identify the presence of primary and secondary salts in relation to soil geomorphological position. The sampling sites were selected using the conditioned Latin hypercube; near-infrared spectroscopy is chosen over an alternative for assessing soil spatial variability through partial least squares, partial least squares regression, and principal components regression models. The spectra were corroborated with conventional analyses, the pretreatments with the Savitsky–Golay filter, and the standard normal variation. A total of 75 evaluated model combinations selected the best R2 for pH (0.61), electric conductivity (EC) (0.52), Cl- (0.55), HCO3- (0.52), and SO42- (0.39). Ordinary Kriging was used to evaluate the variables associated with salt presence, and the correlation between SAS and geomorphological position evidenced the influence of fluviomarine dynamics and ancient salt marshes on the primary and potential salt concentration in the study area.