Although satellite imaging has been present as a source of valuable spatial data for a long time, it was not until very recently that high quality satellite imagery products produced by high resolution multispectral instruments became affordable and broadly available. On the other hand, information contained in such measurements proved to have significant impact on the overall improvement of the best practices in agricultural production and environmental monitoring. One of the applications that could benefit from the large scale satellite based measurements is characterization of topsoil properties of arable land. More exactly, bare soil spectra acquired by multispectral instruments can directly provide information about soil texture, represented by the content of clay, sand, or silt, over the observed vegetation free area. There have been a few attempts to investigate such possibilities in the context of the current and forthcoming multispectral and hyperspectral imagers. In a recently published study, a comprehensive evaluation of the capabilities of several imagers in the task of soil texture estimation was performed. However, those findings were based only on the simulated and resampled spectral responses derived from the soil spectral signature libraries acquired under controlled laboratory conditions using high precision hyperspectral instruments. Among the simulated imagers was also Sentinel-2 MSI. In line with these efforts, aim of this paper is to further investigate applicability of this instrument in the real working environment, characterized by the challenging factors introduced by the atmosphere, tillage and plant remains, missing data due to cloud coverage, variable soil moisture as a consequence of climate and volatile weather conditions, as well as natural soil spatial variability, due to the large spatial extent of the performed analysis.