As an important component of the earth ecosystem, soil moisture monitoring is of great significance in the fields of crop growth monitoring, crop yield estimation, variable irrigation, and other related applications. In order to mitigate or eliminate the impacts of sparse vegetation covers in farmland areas, this study combines multi-source remote sensing data from Sentinel-1 radar and Sentinel-2 optical satellites to quantitatively retrieve soil moisture content. Firstly, a traditional Oh model was applied to estimate soil moisture content after removing vegetation influence by a water cloud model. Secondly, support vector regression (SVR) and generalized regression neural network (GRNN) models were used to establish the relationships between various remote sensing features and real soil moisture. Finally, a regression convolutional neural network (CNNR) model is constructed to extract deep-level features of remote sensing data to increase soil moisture retrieval accuracy. In addition, polarimetric decomposition features for real Sentinel-1 PolSAR data are also included in the construction of inversion models. Based on the established soil moisture retrieval models, this study analyzes the influence of each input feature on the inversion accuracy in detail. The experimental results show that the optimal combination of R² and root mean square error (RMSE) for SVR is 0.7619 and 0.0257 cm³/cm³, respectively. The optimal combination of R² and RMSE for GRNN is 0.7098 and 0.0264 cm³/cm³, respectively. Especially, the CNNR model with optimal feature combination can generate inversion results with the highest accuracy, whose R² and RMSE reach up to 0.8947 and 0.0208 cm³/cm³, respectively. Compared to other methods, the proposed algorithm improves the accuracy of soil moisture retrieval from synthetic aperture radar (SAR) and optical data. Furthermore, after adding polarization decomposition features, the R² of CNNR is raised by 0.1524 and the RMSE of CNNR decreased by 0.0019 cm³/cm³ on average, which means that the addition of polarimetric decomposition features effectively improves the accuracy of soil moisture retrieval results.

Crop classification in agriculture is one of important applications for polarimetric synthetic aperture radar (PolSAR) data. For agricultural crop discrimination, compared with single-temporal data, multi-temporal data can dramatically increase crop classification accuracies since the same crop shows different external phenomena as it grows up. In practice, the utilization of multi-temporal data encounters a serious problem known as a “dimension disaster”. Aiming to solve this problem and raise the classification accuracy, this study developed a feature dimension reduction method using stacked sparse auto-encoders (S-SAEs) for crop classification. First, various incoherent scattering decomposition algorithms were employed to extract a variety of detailed and quantitative parameters from multi-temporal PolSAR data. Second, based on analyzing the configuration and main parameters for constructing an S-SAE, a three-hidden-layer S-SAE network was built to reduce the dimensionality and extract effective features to manage the “dimension disaster” caused by excessive scattering parameters, especially for multi-temporal, quad-pol SAR images. Third, a convolutional neural network (CNN) was constructed and employed to further enhance the crop classification performance. Finally, the performances of the proposed strategy were assessed with the simulated multi-temporal Sentinel-1 data for two experimental sites established by the European Space Agency (ESA). The experimental results showed that the overall accuracy with the proposed method was raised by at least 17% compared with the long short-term memory (LSTM) method in the case of a 1% training ratio. Meanwhile, for a CNN classifier, the overall accuracy was almost 4% higher than those of the principle component analysis (PCA) and locally linear embedded (LLE) methods. The comparison studies clearly demonstrated the advantage of the proposed multi-temporal crop classification methodology in terms of classification accuracy, even with small training ratios.

Accurate and reliable crop classification information is a significant data source for agricultural monitoring and food security evaluation research. It is well-known that polarimetric synthetic aperture radar (PolSAR) data provides ample information for crop classification. Moreover, multi-temporal PolSAR data can further increase classification accuracies since the crops show different external forms as they grow up. In this paper, we distinguish the crop types with multi-temporal PolSAR data. First, due to the “dimension disaster” of multi-temporal PolSAR data caused by excessive scattering parameters, a neural network of sparse auto-encoder with non-negativity constraint (NC-SAE) was employed to compress the data, yielding efficient features for accurate classification. Second, a novel crop discrimination network with multi-scale features (MSCDN) was constructed to improve the classification performance, which is proved to be superior to the popular classifiers of convolutional neural networks (CNN) and support vector machine (SVM). The performances of the proposed method were evaluated and compared with the traditional methods by using simulated Sentinel-1 data provided by European Space Agency (ESA). For the final classification results of the proposed method, its overall accuracy and kappa coefficient reaches 99.33% and 99.19%, respectively, which were almost 5% and 6% higher than the CNN method. The classification results indicate that the proposed methodology is promising for practical use in agricultural applications.

The multisource data fusion technique has been proven to perform better in crop classification. However, traditional fusion methods simply stack the original source data and their corresponding features, which can be only regarded as a superficial fusion method rather than deep fusion. This paper proposes a pixel-level fusion method for multispectral data and dual polarimetric synthetic aperture radar (PolSAR) data based on the polarization extension, which yields synthetic quad PolSAR data. Then we can generate high-dimensional features by means of various polarization decomposition schemes. High-dimensional features usually cause the curse of the dimensionality problem. To overcome this drawback in crop classification using the end-to-end network, we propose a simple network, namely the full tensor decomposition network (FTDN), where the feature extraction in the hidden layer is accomplished by tensor transformation. The number of parameters of the FTDN is considerably fewer than that of traditional neural networks. Moreover, the FTDN admits higher classification accuracy by making full use of structural information of PolSAR data. The experimental results demonstrate the effectiveness of the fusion method and the FTDN model.