In recent decades, the Earth Observation (EO) wave interaction modelling domain has witnessed a proliferation of both forward and inverse models. These models are developed by the scientific community to understand the relationship between electromagnetic waves and natural surfaces, and to support methodologies for extracting bio-geophysical variables from remotely sensed data. However, the current landscape exposes certain limitations such as the absence of systematic implementation, validation on limited datasets, and a scarce integration with emerging Artificial Intelligence (AI)-based inversion techniques. This manuscript introduces the Wave Interaction Models Exploitation Framework (WIMEX), developed to address these challenges in the frame of an ESA-funded project. Leveraging EO data available today, and exploiting Graphical Processing Unit and parallel computing, the framework proposes a systematic approach to create, validate, and disseminate forward and inverse models. WIMEX aims to offer a flexible development environment supporting the evolving needs of the scientific community.

We propose in this article a case-based reasoning algorithm for enhancing intelligent transport systems. To deal effectively with a current real situation, the human brain tends to resort to similar situations encountered previously. Case Based Reasoning mimics human reasoning by storing historical cases in a case base, such as a human memory, for reuse when facing similar cases. In this work, the knowledge and experience gained in the past using optimization algorithms is used to search for solutions of a real on-demand transport problem. The proposed case based reasoning algorithm helps make the right decisions at the right time.

The aim of this study is to retrieve the NDVI values during the wheat cycle using the radar data over reference fields located in the Kairouan plain in the center of Tunisia. The developed approach is based on the use of C-band Sentinel-1 acquisition specifically the cross-polarization ratio and the estimated coherence as features of curve fitting equations and machine learning algorithms such as the random forest and the support vector machine regressors. The NDVI retrieve according to the wheat growth stage, at a field scale, was marked by RMSE values lower than 0.13 and bias values under -0.03

We propose a methodology combining a change detection approach with a neural network algorithm to monitor soil moisture. The methodology utilizes Sentinel-1 and Sentinel-2 data, incorporating various metrics such as radar signals (VV and VH polarization), surface soil moisture index (I_SSM), radar incidence angle, normalized difference vegetation index (NDVI), and VH/VV ratio. In situ data from the International Soil Moisture Network (ISMN) across diverse climatic contexts are used for testing. The results demonstrate improved soil moisture estimations using the hybrid algorithms.

The objective of this study is to analyze GNSS-R data variations as a function of land cover using airborne measurements obtained with the GLObal Navigation Satellite System Reflectometry Instrument (GLORI), which is a polarimetric instrument. GNSS-R measurements were acquired at the agricultural Urgell site in Spain in July 2021. In situ measurements describing the soil and vegetation properties were then obtained simultaneously with flight measurements. The behavior of the observable copolarization (right-right) reflectivity ΓRRand the cross-polarization (right-left) reflectivity ΓRLas a function of land use is discussed. The distribution of coherent and incoherent components in the reflected power is estimated for different types of land cover.

In this paper, we investigate the potential of the modified water cloud model to estimate soil moisture in pepper crop fields with drip irrigation in a semiarid area in Tunisia using cross-polarized L-band data (ALOS-2) and C-band data (Sentinel-1) data in Horizontal-Horizontal (L-HH) and Vertical-Vertical (C- VV) polarization, respectively. Within the context of spatially heterogeneous soil moisture, the total backscattering is the sum of pepper row scattering weighted by the vegetation fraction cover (Fe) and the inter-row soil scattering weighted by (1-Fc). The vegetation row contribution is calculated as the sum of volume scattering contribution of pepper and underlying soil components attenuated by the vegetation cover. Due to the presence of drip irrigation, the underlying soil zone is divided into two parts: irrigated and non-irrigated parts. To assess the calibrated model performance, various simulations are performed under different conditions of soil moisture and vegetation biophysical properties. Under various conditions of soil moisture, the results revealed the potential of the suggested model to simulate SAR signal where cover fraction and pepper height values are under 0.4 and 0.5 m, respectively, using L-HH and cover fraction value under 0.3 and vegetation height value 0.3 m, using C-VV data.

Radar remote sensing has shown a high potential for soil surface parameters estimation in different pedo-climatic context. In the present study, we investigated Sentinel-l radar signal in order to analyze its behavior as function of soil moisture and soil roughness. In addition, we evaluated the approach combining the modified Integral Equation Model (IEM-B) and the Water Cloud Model (WCM) for estimating soil moisture in western France. Soil surface parameters were acquired over 4 campaigns during which composite soil samples were collected simultaneously to Sentinel-l acquisition dates. The dates of those campaigns were defined according to the evolution of the soil surface condition, during the agricultural season. The sensitivity of radar signal$\sigma 0$to soil moisture was studied over the 22 reference fields and over the Thiessen polygons created around the measurement points. Linear relationships are observed between the radar signal and volumetric soil moisture less than 35 vol. % with higher sensitivity for VH polarization (0.41 dB/vol.% in VH against 0.26 dB/vol.% in VV). The best correlation coefficients (R) were observed for the VH polarization with the Zs roughness parameter$(\mathrm{R}={}$0.53 and 0.29 for reference fields and Thiessen polygons, respectively). Following that, a comparison of in situ soil moisture with that predicted based on approach proposed by [1], using Neural network algorithm with a training using the two models IEM-B and Water Cloud Model (WCM) allowed an accuracy with an RMSE ranging between 6.1 and 6.5 vol. % for reference fields and Thiessen polygons respectively. These results confirm that the proposed algorithm is accurate to estimate soil moisture.

In this paper, we discuss the potential of L-band Advanced Land Observing Satellite-2 (ALOS-2) images for retrieving soil moisture over cereal fields in a semi -arid area (Merguellil- Tunisia). SAR signal sensitivity was studied as function of in-situ measurements: roughness and soil moisture. Sensitivity to soil moisture was illustrated for three classes of Normalized Difference Vegetation Index (NDVI). Results reveal the impact of soil moisture on L-band data even in dense vegetation class (NDVI > 0.6). High correlations characterize linear relationships between radar signal and vegetation biophysical properties (Leaf Area Index, vegetation height and Vegetation Water Content). Signal modeling over bare soils was evaluated through empirical equation, modified Dubois model (Dubois-B) and modified Integral Equation Model (IEM-B). For covered fields, Water Cloud Model (WCM) was parametrized for HH and HV polarizations (with and without soil-vegetation interactions component) coupled with the best accuracy bare soil backscattering models: IEM-B for co-polarization and empirical models for the entire dataset. WCM coupled to IEM - B illustrates the best performance to estimate soil water content in HH polarization. The integration of soil-vegetation interaction component provides a stable accuracy of soil moisture estimation in HH polarization and improve soil moisture accuracy in HV polarization mode.

One of the current challenges in the development of robot control systems is making them capable of intelligent and suitable responses to changing environments. But, the control of the robot’s behavior in uncertain and dynamic environments is very challenging when the problem is how to guarantee the robot’s safety by minimizing the interaction with other actors. The most popular methods are based on reactive local navigation schemes that tightly couple the robot actions to the sensor information. These approaches are well based on distance between the robot and the obstacles. This information does not allow the robot to make an intelligent decision while it navigates in unknown environments. Whereas, the robot needs to anticipate environment evolution in order to minimize interaction and avoid conflict with other agents. In this paper, we present a multi-agent simulation model of an autonomous robot in dynamic and uncertain environments. We focus on cases of interactions between agents sharing the same space. The robot should minimize the conflict with other agents when it navigates headed for its goal. Our model is based on fuzzy logic technique in order to deal with the uncertainty of perception.