EGU General Assembly 2023
© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Hierarchical clustering of Sentinel-1 SAR data for soil moisture estimation at the field scale

Giulia Graldi and Alfonso Vitti
Giulia Graldi and Alfonso Vitti
  • Department of Civil, Environmental and Mechanical Engineering, University of Trento, Trento, Italy (

Superficial soil moisture is a key hydrological variable playing a main role in the fluxes of water and heat between land and atmosphere. Its spatial and temporal variations are indeed crucial for applications such as environmental modelling and agricultural management. Soil moisture direct measurements lack spatial representativeness, while soil moisture spatialized information could be derived from satellite data acquired by active microwave sensors. In recent years, missions such as ESA’s (European Space Agency) Sentinel-1 SAR (Synthetic Aperture Radar) mission has provided open data at high resolution (up to 20 m) and temporal frequency (6 days at the equator latitudes before December 2021).

In the proposed work, high resolution data of Sentinel-1 are analyzed on an agricultural area at a resolution of 20 m over a 4 year period. In particular, it is proposed a hierarchical approach for detecting time and space domains where coherently apply a Change Detection method for retrieving soil moisture from the co-polarized band of Sentinel-1 data. The study is conducted at the field scale. Given the agricultural land use of the study area, the total SAR backscattered signal is modelled as the sum of vegetation and attenuated soil contributions.

At first, a classification for masking out the sub-areas dominated by a volumetric response due to vegetation is performed. For doing this, the adaptive thresholding method proposed by Satalino et al., 2014 [1] is performed on proper SAR parameters, such as the VH band, the RVI (Radar Vegetation Index) adapted to Sentinel-1 data [2], and the cross-polarized Interferometric Coherence. The resulting classifications derived from the different parameters are then compared. When working on an agricultural area at the resolution of 20 m, the effects of the soil roughness changes on the backscattering coefficient could not be neglected. For considering them, since no soil roughness data are available on the study area, a time series analysis for detecting steep changes in the co-polarized band is performed. By doing this, it is expected to detect temporal clusters in which no soil roughness variations occur, and thus where a CD method can be applied. The results of the latter classification may be compared with an optical roughness index.


[1] G. Satalino, A. Balenzano, F. Mattia and M. W. J. Davidson, "C-Band SAR Data for Mapping Crops Dominated by Surface or Volume Scattering," in IEEE Geoscience and Remote Sensing Letters, vol. 11, no. 2, pp. 384-388, Feb. 2014, doi: 10.1109/LGRS.2013.2263034.

[2] M. Trudel, F. Charbonneau, R. Leconte, “Using RADARSAT-2 polarimetric and ENVISAT-ASAR dual-polarization data for estimating soil moisture over agricultural fields”. Canadian Journal of Remote Sensing 2012, 38, 514–527.

How to cite: Graldi, G. and Vitti, A.: Hierarchical clustering of Sentinel-1 SAR data for soil moisture estimation at the field scale, EGU General Assembly 2023, Vienna, Austria, 24–28 Apr 2023, EGU23-8533,, 2023.