Please login first
Sentinel data and machine learning algorithms for soil moisture land classification
* 1 , 1 , 2
1  GREEN-TEAM Laboratory (LR17AGR01-INAT), University of Gabes, Gabes National School of Engineers (ENIG), Gabes. Tunisia.
2  University of Sfax, Olive Tree Institute- Tunis Station, Ariana, Tunisia.
Academic Editor: Gianni Bellocchi


In recent years, Tunisia has experienced severe drought, highlighting the need for careful water resource management to ensure both sufficient water availability and long-term sustainability. The use of field data to determine the water use of cultivated crops is a challenging task, with uncertainties due to the lack of representativeness of measurements. The use of remote sensing tools appears as a promising solution.

The study region of our work is located in the south-west of Tunisia, at Kairwan (latitudes 35° to 35° 45’ N and longitudes 9° 30’ to 10° 15’ E). The region is characterized by its arid climate and the overexploitation of the groundwater. We consider three different soil coverage types namely: cereals (LC1), fallow (LC2) and bare soil (LC3). Data used (nine Sentinel-1 and 12 Sentinel-2 images) are downloaded from the Copernicus platform. This paper proposes the following contributions: (1) a formula called 'ER' to estimate soil water status from Sentinel-1 data, (2) two color composition images to monitor changes in soil moisture and crop cover, and (3) the monitoring of the impact of climate change on land use through unsupervised classification using ISODATA and K-Means.

The results demonstrate the importance of the synergy between optical and radar satellite data in determining the soil water status via remote sensing.

Keywords: Soil Moisture; Remote Sensing; Water; Sentinel; Radar; Rainfall; ML.