This paper analyzes the accuracy on the detection of rapeseed fields using Sentinel-1 (S1) time series. Random Forest (RF) and three deep learning (DL) algorithms namely Long Short-Term Memory Fully Convolutional Network (LSTM-FCN), InceptionTime, and Multi-layer Perceptron (MLP) were tested in this study. All four algorithms were used to classify the S1 time series with a large number of ground samples. To test the transferability of classification models, the algorithms were trained on a given year, and then tested on different years. The results demonstrated the high performance of all four algorithms in mapping rapeseed fields when using different years in training and testing phases (F1 between 85.5% and 92.7%, kappa between 0.85 and 0.93).

This study presents the potential of the Sentinel-1 (S1) Synthetic Aperture Radar (SAR) data to detect irrigation events over summer crops, including Maize, Soybean, Sorghum and Potato. The potential of the S1 to detect the irrigation events was carried out using the Irrigation Event Detection Model (IEDM) in five study sites in south Europe and the Middle East. The IEDM is a decision tree model initially developed to detect irrigation events using the change detection algorithm applied to the S1 time series data. Results showed generally good overall accuracy for irrigation detection using the S1 data, reaching 67% for all studied sites together. This accuracy varied according to the studied area, with the highest accuracy for semi-arid areas and lowest for temperate areas. In addition, the accuracy of irrigation detection decreases as the vegetation becomes well developed.

International audience

Despite the current short temporal coverage of high spatial resolution SM maps estimated from Synthetic Aperture Radars such as Sentinel-1(S1), their evaluation is important in the context of the ESA CCI as potential future high resolution (HR) SM long time series, and also as benchmarking references for HR SM data sets that could be obtained by downscaling coarser resolution sensors. In this context, 1 km HR SM maps obtained making a synergistic use of S 1 and Sentinel 2 (or Sentinel 3) using the$S^{2}MP$algorithm were compared to the HR SM data sets from the Copernicus Global Land Service produced from S 1 data over three regions in Europe and one in Tunisia. In addition, the$S^{2}MP$maps were also compared to the SMAP+S 1 downscaled product in those regions and in two additional ones in North America and Australia. The HR SM maps show an overall good agreement for croplands and herbaceous land covers while showing significant differences for other land cover classes. All the 1 km SM maps data sets, in addition to coarse scale SMAP, SMOS and CCI data, were evaluated against in-situ measurements. The results show that the HR products are in good agreement but they show a lower correlation with respect to in-situ data than the coarse resolution products.

Better management of water consumption in irrigated agriculture is essential in order to save water resources. The objective of this study is to propose a new model capable of detecting the irrigation events using the Sentinel-1 (S1) C-band SAR (synthetic-aperture radar) in a near real-time approach. The proposed irrigation detection model relies on the change detection in the S1 backscattering coefficients at plot scale. A tree-based approach has been constructed to detect irrigation events by studying the behavior of the S1 backscattering coefficients following irrigation events at plot scale over three study sites located in Montpellier (southeast France), Tarbes (southwest France) and Catalonia (northeast Spain). Auxiliary data such as the NDVI (Normalized Difference Vegetation Index) and the soil moisture estimations were integrated as additional filters to reduce ambiguities related to vegetation growth and surface roughness. The results shows that the proposed model was capable of detecting 84% of the irrigation events over Montpellier. Over Catalonia site, 90.2% of the non-irrigated plots had no detected irrigation events whereas 72.4% of the irrigated plots had one and more detected irrigation events. In Tarbes, the analysis shows that irrigation events could still be detected even in the presence of abundant rainfall events during the summer season.

The TREIA land data and services center was created with the objective of increasing the use of space data in complementarity with other types of data (in particular in situ, airborne data) by the scientific community and more generally the public actors. TREIA is structuring the French science community through 1) a mutualized Service and Data Infrastructure (SDI) distributed between several centers, allowing access to a variety of products; 2) the setup of Regional Animation Networks (RAN) to federate users (scientists and public / private actors) and 3) Scientific Expertise Centers (SEC) clustering virtual research groups on a thematic domain. A strong relationship between SECs and RANs is being developed to both disseminate the outputs to the user communities and aggregate the user needs. The research works carried out in two SECs are presented. They are organized around the design and development of value-added products and services.

The obj ective of this paper is to present an approach for mapping irrigated areas at plot scale using the Sentinel-1 radar time series. Over a study site located in Catalonia region of north Spain, a dense temporal series of S1 backscattering coefficients were first obtained at plot scale and grid scale (10km x 10km). The S1 time series at plot and grid scales were conjointly used to remove the ambiguity between rainfall events and irrigation events. The principal component analysis (PCA) and the wavelet transformation were applied to the SAR temporal series. Then, to classify irrigated/non-irrigated plots the random forest (RF) classifier was employed using the obtained principal components (PC) and the wavelet coefficients (WT). A convolutional neural network was also tested using the prepared S1 temporal series. The result of the classification reaches 90.7% and 89.1% using the PC and the WT in a random forest classifier respectively. The accuracy of the classification reaches 94.1% using the CNN.

This letter proposes a deep learning model to deal with the spatial transfer challenge for the mapping of irrigated areas through the analysis of Sentinel-1 data. First, a convolutional neural network (CNN) model called “Teacher Model” is trained on a source geographical area characterized by a huge volume of samples. Then, this model is transferred from the source area to the target area characterized by a limited number of samples. The transfer learning framework is based on a distill and refine strategy, in which the teacher model is first distilled into a student model and, successively, refined by data samples coming from the target geographical area. The proposed strategy is compared with different approaches including a random forest (RF) classifier trained on the target data set and a CNN trained on the source data set and directly applied on the target area as well as several CNN classifiers trained on the target data set. The evaluation of the performed transfer strategy shows that the “distill and refine” framework obtains the best performance compared with other competing approaches. The obtained findings represent a first step toward the understanding of the spatial transferability of deep learning models in the Earth observation domain.

Monitoring the surface soil moisture (SSM) in agricultural areas at plot scale helps in many applications such as irrigation planning and crop management. Over the last decade, SAR (Synthetic Aperture Radar) data have shown great potential in estimating SSM over agriculture areas. Today, Sentinel-1 (S1) and Sentinel-2 (S2) satellites present a good opportunity for operational SSM estimates in agricultural areas because they provide free and open access data at high spatial resolution (10 m x 10 m) and high revisit time (6 days over Europe). The aim of this paper is to present an operational approach for mapping soil moisture at high spatial resolution (plot scale) in agriculture areas by coupling S1 and S2 images. The proposed approach is based on the inversion of the Water Cloud Model (WCM) using the neural network technique.

The potential of Sentinel-1 C-band SAR data in VV polarization for estimating the surface roughness (Hrms) over bare agricultural soils was studied. First, a neural network (NN) is used for estimating the soil moisture (mv). Then, a second neural network is used for retrieving the soil roughness in using as an input to the network the soil moisture that was estimated by the first network. The neural networks are trained using simulated dataset generated from the radar backscattering model IEM (Integral Equation Model). The inversion approach is then validated using Sentinel-1 images collected over two study sites, one in France and one in Tunisia. Results show that the use of C-band in VV polarization does not allow a reliable estimate of the soil roughness. Results show that the accuracy on the estimates of Hrms is about 0.8 cm (RMSE).

The objective of the present paper is to develop an operational approach for soil moisture mapping in agricultural areas at a high spatial resolution over bare soils, as well as soils with vegetation cover. The developed approach is based on the synergic use of radar and optical data and uses the neural network technique to invert the radar signal. Three inversion SAR (Synthetic Aperture Radar) configurations were tested: (1) VV polarization, (2) VH polarization, and (3) both VV and VH polarization, all in addition to the NDVI information extracted from optical images. Neural networks were developed and validated using synthetic and real databases. The results showed that the soil moisture could be estimated in agricultural areas with an accuracy of approximately 5 vol.%.

The paper discusses the use of high integration processes and their applications for telecommunications at microwaves. For comparison, implementation techniques are detailed both in GaAs and Si technologies for active LC filters. A comparison in terms of size and cost is also made with the design of integrated negative resistance circuits. We present the BiCMOS SiGe HBT process (in which the transistor base is doped with germanium) for comparison with the GaAs process. Although earlier results seem to give advantage to SiGe and Si technologies in terms of size, cost and power consumption, there are several limitations in comparison to better-known processes, such as GaAs processes, at microwaves. In particular, there are differences in individual component performances, such as inductors or transmission lines. Moreover, the way of implementing the component on the chip is very different. Also, particular biasing methods and topologies may complicate the design. If all these differences and constraints have been overcome, our design examples show that engineers can take advantages of using these processes at microwaves. However, our results also show that GaAs technologies still provide many advantages, would it only be in terms of process maturity at these frequencies.