Using Sentinel-2 for Simplifying Soil Sampling and Mapping: Two Case Studies in Umbria, Italy

Abstract

Soil-sample collection and strategy are costly and time-consuming endeavors, mainly when the goal is in-field variation mapping that usually requires dense sampling. This study developed and tested a streamlined soil mapping methodology, applicable at the field scale, based on an unsupervised classification of Sentinel-2 (S2) data supporting the definition of reduced soil-sampling schemes. The study occurred in two agricultural fields of 20 hectares each near Deruta, Umbria, Italy. S2 images were acquired for the two bare fields. After a band selection based on bibliography, PCA (Principal Component Analysis) and cluster analysis were used to identify points of two reduced-sample schemes. The data obtained by these samplings were used in linear regressions with principal components of the selected S2 bands to produce maps for clay and organic matter (OM). Resultant maps were assessed by analyzing residuals with a conventional soil sampling of 30 soil samples for each field to quantify their accuracy level. Although of limited extent and with a specific focus, the low average errors (Clay ± 2.71%, OM ± 0.16%) we obtained using only three soil samples suggest a wider potential for this methodology. The proposed approach, integrating S2 data and traditional soil-sampling methods could considerably reduce soil-sampling time and costs in ordinary and precision agriculture applications.

Integration of Geostatistical and Sentinal-2AMultispectral Satellite Image Analysis for Predicting Soil Fertility Condition in Drylands

“…S2A data may help to characterize a given soil gradient, even if only a few soil samples were used during differentiated crop operations (within the process of a precision agriculture) [30]. The following five Equations ( 8)- (12), were used to obtain the values of pH, SOM %, N, P, and K from the reflectance of the satellite image.…”

“…One of the major challenges today is finding the best way to develop low-cost methods for modelling and mapping physical and chemical soil properties over large areas. Remote sensing (RS) has commonly been utilized, mainly at a regional scale, for soil mapping because of the low costs, quick data acquisition, and availability of high temporal and spatial information quality [30]. Satellites and aerial platforms that are used for capturing bare soil images can be analyzed by traditional methods, such as band ratio and discriminant analysis [30].…”

“…Remote sensing (RS) has commonly been utilized, mainly at a regional scale, for soil mapping because of the low costs, quick data acquisition, and availability of high temporal and spatial information quality [30]. Satellites and aerial platforms that are used for capturing bare soil images can be analyzed by traditional methods, such as band ratio and discriminant analysis [30]. An adequate prediction of soil properties using hyperspectral data is not easy due to the soil reflectance phenomenon.…”

“…An adequate prediction of soil properties using hyperspectral data is not easy due to the soil reflectance phenomenon. To overcome this problem, regression analysis, including multiple regression analysis (MLR), principal component regression (PCR), and partial least-squares (PLS) regression, have been used, with these making connections between soil properties and reflectance data or spectral indices [30,31]. The use of hyperspectral VNIR/SWIR imagery was confirmed by a previous study as being a very important tool for mapping soil properties, however, it cannot be implemented for large areas or for long time periods due to high cost and low availability [5].…”

Prediction of Soil Water Content and Electrical Conductivity Using Random Forest Methods with UAV Multispectral and Ground-Coupled Geophysical Data

“…Advances in satellite technology, data processing, and petrophysics ensure that these techniques will be increasingly important in the future. Many studies have used satellite data, such as the Sentinel and Landsat series, to observe and predict soil properties such as pH [21,22], cation exchange capacity [22], soil organic carbon [22,23], soil organic matter [21,24], clay content [21,22,24], salinity [25], and soil water content estimation [26][27][28][29][30][31][32]. Reference [24] showed that using Sentinel-2 satellite data to identify differences in soil properties can guide on-ground soil sampling and significantly reduce the time and cost of conventional sampling efforts.…”