Potential of high spatial resolution radars for the characterization of soil properties in agricultural environments

[Departement_IRSTEA]Territoires [TR1_IRSTEA]SYNERGIE [Axe_IRSTEA]TETIS-ATTOS

International audience

Английский. Soil characteristics play a key role in different hydrological processes (floods, runoff, evapotranspiration, infiltration, soil erosion and related contamination, and imbalances in the water and carbon cycles). In the last decades, important scientific activity has been devoted to demonstrate the potential of radar remote sensing and in particular high resolution measurements with Synthetic Aperture Radar (SAR) to describe surface properties of agricultural areas. The objective of this chapter is the gathering of principal theoretical and experimental scientific developments in the hopes of understanding research instances in different topics linked to the use of SAR remote sensing. In the last two decades, radar measurements showed a high evolution in comparison to the first radar generation (ERS…) from one configuration (angle, frequency, polarization) to multi-configuration radars (ASAR/ENVISAT, RADARSAT ..), and finally the arrival in the last years of radar constellations (TerraSAR-X, COSMO-SkyMed) allowed the proposal of high repetitivity and a very high spatial resolution imagery (about 1 m). The first part of this chapter illustrates the sensitivity of radar signals to different surface parameters (roughness, moisture, salinity, texture). In this part, techniques and methods of measurement, and the description of soil parameters are detailed. The high sensitivity of the radar signal to the soil moisture and roughness is demonstrated. The high potential of radar signals to retrieve textures or salinity for particular meteorological conditions or radar configurations is established. The second part describes the backscattering models developed to analyze the mathematical relationship between soil parameters and radar signals. Different approaches (analytical physical, semi-empirical or empirical) are detailed. Empirical models based on the use of real data only, and semi-empirical models combining physical modeling with empirical parameterization are particularly useful for data inversion in an operational context. Analytical models, based on theoretical modeling, are useful for the understanding of radar signal physical behavior and retrieving of radar configuration optimization for soil parameters characterization. The third part provides a description of different inversion methods applied for bare soils or soils covered with vegetation. Several approaches considering one, or more configurations (multi-incidence, multi-polarization, polarimetric) are described. Inversion methods, based on different techniques: regression, inversion of analytical direct models, Artificial Neural network, change detection or others are illustrated. At field agriculture scale, radar signals produce a precision of about 4-5 vol% for soil moisture estimation; it permits the separation of soils into three roughness classes: smooth, medium and rough. The launch of the SENTINEL-1 and RADARSAT constellations in the upcoming years should allow the development of surface characteristics by operational mapping. High resolution products (with spatial resolution approximately between 100 m and 1 km) will be proposed to the scientific community for assimilation in process models.