Mapping of Soil Moisture Variability, Using the Sentinel-2 Data Optical-Trapezoid Approach

Abstract

Understanding soil moisture is crucial in various aspects of daily life and scientific pursuits. Among these, knowledge about water stress conditions holds particular significance for both agriculture and soil conservation. The objective of this research is to explore the application of satellite imagery in the cartography and surveillance of moisture levels within an agricultural region. Soil moisture content was assessed using the optical trapezoid model (OPTRAM). Developed by Sadeghi et al. (2015), the OPtical TRApezoidal model (OPTRAM) was designed to gauge soil water content (SWC) by assuming a linear correlation between soil moisture content and shortwave transformed reflectance (STR). The parameters essential for calculating moisture content were identified by scrutinizing pixel distribution in the STR-NDVI (Normalized Difference Vegetation Index) space. The examination period spanned from April to October 2021. The models were employed to compute the spatial fluctuation of soil moisture and its deviation for three satellite images during the summer of 2021. The study site was located in Hungary's Bács-Kiskun county, encompassing agricultural fields with a total expanse of 5500 km2. The study region exhibited variability in terms of soil composition and topography. Meteorological parameters recorded at 19 stations within a drought monitoring network, along with soil moisture measurements at different depths, were also taken into account. To validate the data obtained from the soil moisture sensor and model, soil samples were collected at a depth of 10 cm for laboratory moisture assessments. The present condition can be depicted through the analysis of a spatial image, while time series analyses enable continuous monitoring of soil moisture. The eCognition software environment, employing the object-based (OBIA) approach, was used to process satellite data. Statistical methods were utilized to establish correlations between the datasets measured at the site and estimated from satellite images.