Dual-polarization Synthetic Aperture Radar Data Research Articles

Land cover mapping of mixed classes using 2D CNN with multi-frequency SAR data

Synthetic aperture radar (SAR) data obtained at multiple frequencies and polarizations offers valuable complementary information for classifying mixed classes that exhibit similar backscattering response. Although deep learning-based convolutional neural networks (CNNs) effectively extract features from multi-frequency SAR data, the arbitrary ordering of SAR features may hinder optimal convolution of the best feature sub-space for a specific class and underutilize available multi-frequency data. To address this, a novel CNN transforming SAR feature-space from 1-D to 2-D and employing varied dilation-rate convolutions is introduced. This transformation maximizes unique and localized feature combinations, efficiently utilizing the available feature sub-spaces and extracting discriminative features for accurate classifications, addressing the challenge of arbitrary band neighborhoods. Utilizing dual-polarization SAR data from ALOS-2 PALSAR-2 and Sentinel-1 sensors, the proposed CNN achieves an average f-score of 0.97 and a kappa coefficient of 0.97, an improvement of 11 %, 7 % and 3 % in OA compared to the 1-D, 2-D and 3-D CNN classifiers, without feature transformation. The classifier's generalization ability is evaluated using ground truth knowledge of various heterogeneous classes, and the proposed CNN classifier outperforms others in terms of accuracy metrics and generalization ability.

Single-Season Rice Area Mapping by Combining Multi-Temporal Polarization Decomposition Components and the Two-Stage Segmentation Method

Recently, Synthetic Aperture Radar (SAR) data, especially Sentinel-1 data, have been increasingly used in rice mapping research. However, current studies usually use long time series data as the data source to represent the differences between rice and other ground objects, especially other crops, which results in complex models and large computational complexity during classification. To address this problem, a novel method for single season rice mapping is proposed, based on the principle that the scattering mechanism of rice paddies in the early flooding period is strongly influenced by water bodies, causing the volume scattering to be lower than that of other crops. Thus, a feature combination that can effectively and stably extract rice planting areas was constructed by combining multi-temporal volume scattering in the early flooding period of rice using dual-polarization SAR data, so that a simple semantic segmentation model could realize high-precision rice mapping tasks. A two-stage segmentation structure was introduced to further improve the mapping result with the Omni-dimensional Dynamic Convolution Residual Segmentation model (ODCRS model) as the bone model. In the experiment, Suihua City, Heilongjiang Province was selected as the study site, and the VH/VV polarized data of Sentinel-1 satellite in 2022 was used as the data source. The mapping accuracy of the ODCRS model was 88.70%, and the user accuracy was 84.19% on the field survey data. Furthermore, experiments with different years and regions also proved the effectiveness and stability of the proposed method.

Assessment of the Impact of Surface Water Content for Temperate Forests in SAR Data at C-Band

This study addresses the escalating challenges posed by forest drought and wildfires, emphasizing the critical need to monitor forest conditions to mitigate associated risks. While traditional optical sensors have proven valuable for vegetation surface water (VSW) assessment, their limitations in regions with persistent cloud cover prompt an exploration of the alternatives. The study advocates the efficacy of Synthetic Aperture Radar (SAR) systems, known for their cloud-penetrating capabilities and sensitivity to changes in dielectric properties. Leveraging Sentinel-1 C-band dual polarization SAR data, the research investigates the impact of Vegetation Surface Water (VSW) on backscatter coefficients in a temperate coniferous forest through the application of generalized linear models. Despite the challenges posed by precipitation and canopy characteristics, the study unveils detectable modulation in backscatter, particularly in VH polarization, indicating the potential of SAR-based methods in forest monitoring. The occurrence of rain on the day of Sentinel-1 image acquisition, and therefore the presence of VSW, triggers an increase of 0.35 dB in VV backscatter, and an increase of 0.45 dB in VH backscatter. The findings underscore the importance of considering surface water content in radar backscatter analyses for accurate biomass estimations and change detection, suggesting avenues for future research and potential correction mechanisms.

Lithology classification of igneous rocks using C-band and L-band dual-polarization SAR data

Abstract Distinguishing different kinds of igneous rocks is difficult because of their subtle differences. Synthetic aperture radar (SAR) is sensitive to rock surface morphology, which can help a lot in classification. Dual-pol SAR data have the advantages of low cost, but few articles are using only dual-pol SAR data for igneous rock classification. In this study, we explored the performance of dual-pol SAR data in distinguishing granitoid, tuff, and syenite porphyry. Backscatter coefficients, polarization decomposition parameters, and texture features from gray-level co-occurrence matrix extracted by Sentinel-1 or PALSAR were classified using several machine learning algorithms. The results are as follows. First, the texture information has greater potential for igneous rock classification, but the polarization decomposition parameters contribute less. Second, after comparing machine learning algorithms, AdaBoost algorithm has the highest overall accuracy for either C-band or L-band SAR data. C-band SAR data provide better classification results than L-band. Finally, tuff is the easiest igneous rock to be successfully classified, and L-band dual-pol SAR data have advantages in the discrimination of syenite porphyry. This study outlines the effectiveness of dual-pol SAR data for igneous rock classification, which will help to select SAR data of appropriate wavelengths for specific types of lithology discrimination.

Land Cover Classification of SAR Based on 1DCNN-MRF Model Using Improved Dual-Polarization Radar Vegetation Index

Accurate land cover classification (LCC) is essential for studying global change. Synthetic aperture radar (SAR) has been used for LCC due to its advantage of weather independence. In particular, the dual-polarization (dual-pol) SAR data have a wider coverage and are easier to obtain, which provides an unprecedented opportunity for LCC. However, the dual-pol SAR data have a weak discrimination ability due to limited polarization information. Moreover, the complex imaging mechanism leads to the speckle noise of SAR images, which also decreases the accuracy of SAR LCC. To address the above issues, an improved dual-pol radar vegetation index based on multiple components (DpRVIm) and a new LCC method are proposed for dual-pol SAR data. Firstly, in the DpRVIm, the scattering information of polarization and terrain factors were considered to improve the separability of ground objects for dual-pol data. Then, the Jeffries-Matusita (J-M) distance and one-dimensional convolutional neural network (1DCNN) algorithm were used to analyze the effect of difference dual-pol radar vegetation indexes on LCC. Finally, in order to reduce the influence of the speckle noise, a two-stage LCC method, the 1DCNN-MRF, based on the 1DCNN and Markov random field (MRF) was designed considering the spatial information of ground objects. In this study, the HH-HV model data of the Gaofen-3 satellite in the Dongting Lake area were used, and the results showed that: (1) Through the combination of the backscatter coefficient and dual-pol radar vegetation indexes based on the polarization decomposition technique, the accuracy of LCC can be improved compared with the single backscatter coefficient. (2) The DpRVIm was more conducive to improving the accuracy of LCC than the classic dual-pol radar vegetation index (DpRVI) and radar vegetation index (RVI), especially for farmland and forest. (3) Compared with the classic machine learning methods K-nearest neighbor (KNN), random forest (RF), and the 1DCNN, the designed 1DCNN-MRF achieved the highest accuracy, with an overall accuracy (OA) score of 81.76% and a Kappa coefficient (Kappa) score of 0.74. This study indicated the application potential of the polarization decomposition technique and DEM in enhancing the separability of different land cover types in SAR LCC. Furthermore, it demonstrated that the combination of deep learning networks and MRF is suitable to suppress the influence of speckle noise.

Rural land cover mapping using short-term Sentinel-2 temporal composites and machine learning

Multispectral sensors have been invaluable for land cover mapping. Nonetheless, cloud cover is frequent in many parts in the tropics, meaning that relying on single-date multispectral data can be problematic. The limited number of variables, i.e., only two, in conventional dual-polarized synthetic aperture radar data usually drives low classification accuracies. Researchers have explored an augmentation strategy that uses textural features, such as Haralick’s Gray Level Co-occurrence Matrix and/or decomposition features. However, this also does not produce an acceptable outcome. Another approach involves short-term temporal stacking. This masks out atmospheric disturbance and means every pixel has a single value representing the object. Our research explored this last approach to identify woody vegetation in a rural area of Indonesia from Sentinel-2 data. Ten machine learning algorithms were explored to investigate limitations and opportunities. We investigated algorithms capable of reporting variable importance to indicate crucial spectral bands. This study indicated that highly important spectra include red edge and short-wave infrared bands and to a lesser extent, the red band. Part of these bands were confirmed by the variable importance. The spectral similarity between bands 8 and 8A could be seen as a drawback when both bands are employed. Variable importance indicated a negative impact due to this similarity, suggesting that only one should be used in the classification. Overall, classification accuracy was generally high (over 96%), except in the case of the benchmark, i.e., classification and regression trees. Highest accuracy (97.6%) was provided by C5.0 model, although the difference from random forests (RFs) was negligible. The RF was favored as it produced a better estimate of timber plantations. A few algorithms were unable to separate woody vegetation types, including the popular support vector machine. Successful separation was mainly achieved by ensemble tree learners, such as RFs and the variants of extreme gradient boosting.

Flood Detection in Dual-Polarization SAR Images Based on Multi-Scale Deeplab Model

The proliferation of massive polarimetric Synthetic Aperture Radar (SAR) data helps promote the development of SAR image interpretation. Due to the advantages of powerful feature extraction capability and strong adaptability for different tasks, deep learning has been adopted in the work of SAR image interpretation and has achieved good results. However, most deep learning methods only employ single-polarization SAR images and ignore the water features embedded in multi-polarization SAR images. To fully exploit the dual-polarization SAR data and multi-scale features of SAR images, an effective flood detection method for SAR images is proposed in this paper. In the proposed flood detection method, a powerful Multi-Scale Deeplab (MS-Deeplab) model is constructed based on the dual-channel MobileNetV2 backbone and the classic DeeplabV3+ architecture to improve the ability of water feature extraction in SAR images. Firstly, the dual-channel feature extraction backbone based on the lightweight MobileNetV2 separately trains the dual-polarization SAR images, and the obtained training parameters are merged with the linear weighting to fuse dual-polarization water features. Given the multi-scale space information in SAR images, then, a multi-scale feature fusion module is introduced to effectively utilize multi-layer features and contextual information, which enhances the representation of water features. Finally, a joint loss function is constructed based on cross-entropy and a dice coefficient to deal with the imbalanced categorical distribution in the training dataset. The experimental results on the time series of Sentinel-1A SAR images show that the proposed method for flood detection has a strong ability to locate water boundaries and tiny water bodies in complex scenes. In terms of quantitative assessment, MS-Deeplab can achieve a better performance compared with other mainstream semantic segmentation models, including PSPNet, Unet and the original DeeplabV3+ model, with a 3.27% intersection over union (IoU) and 1.69% pixel accuracy (PA) improvement than the original DeeplabV3+ model.

Retrieval of Farmland Surface Soil Moisture Based on Feature Optimization and Machine Learning

Soil moisture is an important parameter affecting environmental processes such as hydrology, ecology, and climate. Synthetic aperture radar (SAR) microwave remote sensing is an important means of farmland surface soil moisture (SSM) measurement. The inversion of farmland SSM by microwave remote sensing is greatly affected by vegetation cover. To address this problem, a multisource remote sensing inversion method of farmland SSM based on feature optimization and machine learning is proposed in this paper. Six typical machine learning algorithms suitable for small sample training, including random forest, radial basis function neural network, generalized regression neural network, support vector regression, genetic algorithm–back propagation neural network, and extreme learning machine, were selected in this paper. The features extracted from Sentinel-1/2 and Radarsat-2 remote sensing data were analyzed by Pearson correlation, and those with high correlation coefficients were selected to form the optimal feature subset as the input for the subsequent machine learning models. Then, the SSM collaborative inversion models under different machine learning algorithms were constructed, and comparative experiments were set up to select the optimal prediction model. The models’ accuracy under different feature parameters were studied, and the difference in the performance between the dual-polarization SAR data and the quad-polarization SAR data in SSM inversion was explored. The experimental results showed that among the six models, the random forest model had a higher inversion accuracy, with a coefficient of determination of 0.6395 and a root mean square error of 0.0264 cm3/cm3. Meanwhile, the inversion accuracy could be greatly improved after feature optimization, and the inversion accuracy of the quad-polarization SAR data combined with optical remote sensing data, was better than that of the dual-polarization SAR data combined with optical remote sensing data.

Oil Spill Detection with Dual-Polarimetric Sentinel-1 SAR Using Superpixel-Level Image Stretching and Deep Convolutional Neural Network

Synthetic aperture radar (SAR) has been widely applied in oil spill detection on the sea surface due to the advantages of wide area coverage, all-weather operation, and multi-polarization characteristics. Sentinel-1 satellites can provide dual-polarized SAR data, and they have high potential for successful application to oil spill detection. However, the characteristics of the sea surface and oil film on different images are not the same when imaging at different locations and in different conditions, which leads to the inconsistent accuracy of these images with the application of the current oil spill detection methods. In order to avoid the above limitation, we propose an oil spill detection method using image stretching based on superpixels and a convolutional neural network. Experiments were carried out on eight Sentinel-1 dual-pol data, and the optimal superpixel number and image stretching parameters are discussed. Mean intersection over union (MIoU) was used to evaluate classification accuracy. The proposed method could effectively improve the classification accuracy; when the expansion and inhibition coefficients of image stretching were set to 1.6 and 1.2 respectively, the experiments achieved a maximum MIoU of 85.4%, 7.3% higher than that without image stretching.

Detection of Flood Extent Using Sentinel-1A/B Synthetic Aperture Radar: An Application for Hurricane Harvey, Houston, TX

The increasing number of flood events combined with coastal urbanization has contributed to significant economic losses and damage to buildings and infrastructure. Development of higher resolution SAR flood mapping that accurately identifies flood features at all scales can be incorporated into operational flood forecasting tools, improving response and resilience to large flood events. Here, we present a comparison of several methods for characterizing flood inundation using a combination of synthetic aperture radar (SAR) remote sensing data and machine learning methods. We implement two applications with SAR GRD data, an amplitude thresholding technique applied, for the first time, to Sentinel-1A/B SAR data, and a machine learning technique, DeepLabv3+. We also apply DeepLabv3+ to a false color RGB characterization of dual polarization SAR data. Analyses at 10 m pixel spacing are performed for the major flood event associated with Hurricane Harvey and associated inundation in Houston, TX in August of 2017. We compare these results with high-resolution aerial optical images over this time period, acquired by the NOAA Remote Sensing Division. We compare the results with NDWI produced from Sentinel-2 images, also at 10 m pixel spacing, and statistical testing suggests that the amplitude thresholding technique is the most effective, although the machine learning analysis is successful at reproducing the inundation shape and extent. These results demonstrate the effectiveness of flood inundation mapping at unprecedented resolutions and its potential for use in operational emergency hazard response to large flood events.

Improved k-NN Mapping of Forest Attributes in Northern Canada Using Spaceborne L-Band SAR, Multispectral and LiDAR Data

Satellite forest inventories are the only feasible way to map Canada’s vast, remote forest regions, such as those in the Northwest Territories (NWT). A method used to create such inventories is the k-nearest neighbour (k-NN) algorithm, which spatially extends information from forest inventory (FI) plots to the entire forest land base using wall-to-wall features typically derived from Landsat data. However, the benefits of integrating L-band synthetic aperture radar (SAR) data, strongly correlated to forest biomass, have not been assessed for Canadian northern boreal forests. Here we describe an optimized multivariate k-NN implementation of a 151,700 km2 area in southern NWT that included ca. 2007 Landsat and dual-polarized Phased Array type L-band SAR (PALSAR) data on board the Advanced Land Observing Satellite (ALOS). Five forest attributes were mapped at 30 m cells: stand height, crown closure, stand/total volume and aboveground biomass (AGB). We assessed accuracy gains compared to Landsat-based maps. To circumvent the scarcity of FI plots, we used 3600 footprints from the Geoscience Laser Altimeter System (GLAS) as surrogate FI plots, where forest attributes were estimated using Light Detection and Ranging (LiDAR) metrics as predictors. After optimization, k-NN predicted forest attribute values for each pixel as the average of the 4 nearest (k = 4) surrogate FI plots within the Euclidian space of 9 best features (selected among 6 PALSAR, 10 Landsat, and 6 environmental features). Accuracy comparisons were based on 31 National Forest Inventory ground plots and over 1 million airborne LiDAR plots. Maps that included PALSAR HV backscatter resulted in forest attribute predictions with higher goodness of fit (adj. R2), lower percent mean error (ME%), and percent root mean square error (RMSE%), and lower underestimation for larger attribute values. Predictions were most accurate for conifer stand height (RMSE% = 32.1%, adj. R2 = 0.58) and AGB (RMSE% = 47.8%, adj. R2 = 0.74), which is much more abundant in the area than mixedwood or broadleaf. Our study demonstrates that optimizing k-NN parameters and feature space, including PALSAR, Landsat, and environmental variables, is a viable approach for inventory mapping of the northern boreal forest regions of Canada.

An Improved Sea Ice Classification Algorithm with Gaofen-3 Dual-Polarization SAR Data Based on Deep Convolutional Neural Networks

The distribution of sea ice is one of the major safety hazards for sea navigation. As human activities in polar regions become more frequent, monitoring and forecasting of sea ice are of great significance. In this paper, we use SAR data from the C-band synthetic aperture radar (SAR) Gaofen-3 satellite in the dual-polarization (VV, VH) fine strip II (FSII) mode of operation to study the Arctic sea ice classification in winter. SAR data we use were taken in the western Arctic Ocean from January to February 2020. We classify the sea ice into four categories, namely new ice (NI), thin first-year ice (tI), thick first-year ice (TI), and old ice (OI), by referring to the ice maps provided by the Canadian Ice Service (CIS). Then, we use the deep learning model MobileNetV3 as the backbone network, input samples of different sizes, and combine the backbone network with multiscale feature fusion methods to build a deep learning model called Multiscale MobileNet (MSMN). Dual-polarization SAR data are used to synthesize pseudocolor images and produce samples of sizes 16 × 16 × 3, 32 × 32 × 3, and 64 × 64 × 3 as input. Ultimately, MSMN can reach over 95% classification accuracy on testing SAR sea ice images. The classification results using only VV polarization or VH polarization data are tested, and it is found that using dual-polarization data could improve the classification accuracy by 10.05% and 9.35%, respectively. When other classification models are trained using the training data from this paper for comparison, the accuracy of MSMN is 4.86% and 1.84% higher on average than that of the model built using convolutional neural networks (CNNs) and ResNet18 model, respectively.

The RADARSAT Constellation Mission Core Applications: First Results

The Canadian RADARSAT Constellation Mission (RCM) has passed its early operation phase with the performance evaluation being currently active. This evaluation aims to confirm that the innovative design of the mission’s synthetic aperture radar (SAR) meets the expectations of intended users. In this study, we provide an overview of initial results obtained for three high-priority applications; flood mapping, sea ice analysis, and wetland classification. In our study, the focus is on results obtained using not only linear polarization, but also the adopted Compact Polarimetric (CP) architecture in RCM. Our study shows a promising level of agreement between RCM and RADARSAT-2 performance in flood mapping using dual-polarized HH-HV SAR data over Red River, Manitoba, suggesting smooth continuity between the two satellite missions for operational flood mapping. Visual analysis of coincident RCM CP and RADARSAT-2 dual-polarized HH-HV SAR imagery over the Resolute Passage, Canadian Central Arctic, highlighted an improved contrast between sea ice classes in dry ice winter conditions. A statistical analysis using selected sea ice samples confirmed the increased contrast between thin and both rough and deformed ice in CP SAR. This finding is expected to enhance Canadian Ice Service’s (CIS) operational visual analysis of sea ice in RCM SAR imagery for ice chart production. Object-oriented classification of a wetland area in Newfoundland and Labrador by fusion of RCM dual-polarized VV-VH data and Sentinel-2 optical imagery revealed promising classification results, with an overall accuracy of 91.1% and a kappa coefficient of 0.87. Marsh presented the highest user’s and producer’s accuracies (87.77% and 82.08%, respectively) compared to fog, fen, and swamp.

Seasonal snow cover classification based on SAR imagery and topographic data

ABSTRACT Melting snow is a sensitive indicator of regional climate change and snowmelt runoff. Optical remote sensing offers an effective tool to delineate the extent of snow cover, but its application is limited by sky conditions and illumination. In contrast, synthetic aperture radar (SAR) has the advantages of high penetration and can obtain the bulk properties of snowpack, such as snow wetness, in addition to the snow-covered area (SCA). The widely used algorithm based on the multitemporal backscattering coefficient can delineate wet snow and further estimate dry snow cover by empirical rules. However, employing a fixed threshold on wet snow extraction and only considering altitude in dry snow estimation lead to the misclassification of snow cover. In this study, an algorithm was developed for dry and wet snow discrimination based on C-band dual-polarization SAR data. First, the adaptive threshold obtained by the Otsu thresholding method (OTSU) was implemented on the ratio image to improve wet snow extraction, and then topographic information, including elevation, slope and aspect, was added to further infer dry snow cover. Snow field observations were carried out in March 2021 and February 2017 to evaluate the effectiveness of the presented method in typical snow regions with different snowpack characteristics. Compared with using a single equation with a fixed threshold (57.9% and 56.5%, respectively), the overall accuracy of the novel approach in prairie and alpine snow areas was 84.2% and 73.9%, respectively. The results show that considering the adaptive threshold and topography offers superior performance in identifying dry and wet snow.

Wetland Classification in Poyang Lake Using Dual-polarization Synthetic Aperture Radar Data with Feature Combination

Wetland Classification in Poyang Lake Using Dual-polarization Synthetic Aperture Radar Data with Feature Combination

Sea Spike Suppression Method Based on Optimum Polarization Ratio in Airborne SAR Images.

In the condition of ocean observation for high-resolution airborne synthetic aperture radar (SAR), sea spikes will cause serious interference to SAR image interpretation and marine target detection. In order to improve the ability of target detection, it is necessary to suppress sea spikes in SAR images. However, there is no report on sea spike suppression methods in SAR images. As a step forward, a sea spike suppression method based on optimum polarization ratio in airborne SAR images is proposed in this paper. This method is only applicable to the situation where VV and HH dual-polarized SAR data containing sea spikes are acquired at the same time. By calculating the optimum polarization ratio, this method further obtains the difference image of the panoramic area accomplishing sea spike suppression. This method is applied to a field airborne X-band SAR data, including ocean waves, oil spills and ships. The results show that the sea spikes are well suppressed, the contrast of ocean waves and the contrast of oil spills are improved, and the false alarm rate of ship detection is reduced. The discussions on these results demonstrate that the proposed method can effectively suppress sea spikes and improve the interpretability of SAR images.

From local to regional compound flood mapping with deep learning and data fusion techniques

Compound flooding (CF), as a result of oceanic, hydrological, meteorological and anthropogenic drivers, is often studied with hydrodynamic models that combine either successive or concurrent processes to simulate inundation dynamics. In recent years, convolutional neural networks (CNNs) and data fusion (DF) techniques have emerged as effective alternatives for post-flood mapping and supported current efforts of complex physical and dynamical modeling. Yet, those techniques have not been explored for large-scale (regional) compound flood mapping. Here, we evaluate the performance of a CNN & DF framework for generating CF maps along the southeast Atlantic coast of the U.S. as a result of Hurricane Matthew (October 2016). The framework fuses multispectral imagery from Landsat analysis ready data (ARD), dual-polarized synthetic aperture radar data (SAR), and coastal digital elevation models (DEMs) to produce flood maps at moderate (30 m) spatial resolution. The highest overall accuracy (97%) and f1-scores of permanent water/floodwater (99/100%) are achieved when ARD, SAR and DEM datasets are readily available and fused. Moreover, the resulting CF maps agree well (80%) with hindcast flood guidance maps of the Coastal Emergency Risk Assessment and can effectively match post-flood high water marks of the U.S. Geological Survey distributed in coastal counties. We ultimately evaluate the framework with different DF alternatives and highlight their usefulness for large-scale compound flood mapping as well as calibration of hydrodynamic models. The cost-effective approach proposed here enables efficient estimation of exposure to compound coastal flooding and is particularly useful in data scarce regions.

A joint change detection method on complex-valued polarimetric synthetic aperture radar images based on feature fusion and similarity learning

ABSTRACT Existing deep learning-based change detection methods in the field of polarimetric synthetic aperture radar (POLSAR) usually directly deal with intensity images. Methods can be easily transferred from optical image processing to synthetic aperture radar (SAR) image processing. However, the polarization information, which is critical in POLSAR data, has been discarded under normal situations. This paper introduces a novel joint change detection network based on similarity learning for coregistered complex-valued SAR data. A pseudo-siamese network takes both amplitude information and polarization information of POLSAR data as the input. The fusion of low-level, middle-level and high-level features enables the network to keep high-resolution and have strong representation ability during training procedures. Our novel sub-networks, which we term C2-Net and Intensity-Net, deal with the covariance matrix of complex SAR data and amplitude SAR data, respectively. The Intensity-Net works as a typical classification network and detects targets directly. The C2-Net attempts to find the relationship between two SAR data patches. An improved cosine similarity function is used to measure the similarity between two generated feature vectors in C2-Net. Output probability vectors of the two sub-networks are combined for final change detection. The two sub-networks are trained jointly and simultaneously. Control experiments show that proposed improvements are working. Experimental results on dual-polarization complex-valued SAR data from Sentinel-1 demonstrate the effectiveness of the proposed method.

Crop Mapping Using Sentinel Full-Year Dual-Polarized SAR Data and a CPU-Optimized Convolutional Neural Network With Two Sampling Strategies

Although optical remote sensing can capture the Earth's environment with visible and infrared sensors, it is limited by weather conditions. Often, only a few sets of cloud-free optical imagery are available in cloudy regions, where many agricultural towns are located. On the other hand, radar remote sensing can capture imagery under cloudy conditions. In this study, we examined the capability of Sentinel-1 multitemporal dual-polarized synthetic aperture radar (SAR) imagery in a whole year from Google Earth Engine in crop mapping in two study sites in Chongqing, China, and Landivisiau, France. Results show that it is possible to produce better crop classification maps using multitemporal SAR imagery, but the performance is limited by local terrain. Flat agricultural regions, such as Western Europe, are expected to benefit from the multitemporal SAR information. Mountain agricultural regions, such as Southwestern China, will encounter difficulties due to the undulate terrain. We also tested two sampling strategies, i.e., random sampling and regional sampling, and observed high variation in overall accuracy: the former led to a higher accuracy. The gap is caused by the diversity of training sets examined using tSNE visualization. The importance of SAR channels in each month are correlated with their entropy. Data from the growing season are important in distinguishing crop types. The 3-D convolutional neural network (CNN) achieved similar results under a huge computation cost compared with 2-D CNNs. Based on the experiments, we recommend to use a lightweight 2-D CNN that can run on the CPU for real-world crop mapping with SAR data.

Monitoring of rice lodging using Sentinel-1 data

Synthetic Aperture Radar (SAR) can monitor rice regardless of time and weather condition, and the ability of crop lodge monitoring has been demonstrated by previous studies. However, there have no study about monitoring of rice lodging using satellite-based SAR. In this study, we extracted backscatter coefficient (BC) and H/A/Alpha polarimetric parameters from dual-polarized Sentinel-1 SAR data, and a lodging sensitivity factor γ was constructed for selecting optimal radar feature parameter (ORFP) which is highly sensitivity to lodging rice. Then, the decision tree classification method was used with multiple ORFPs to classify the healthy and lodging rice plots. We firstly acquired shape of rice region using Sentinel-2 image and Maximum likelihood classification method for eliminating the influence of other features before classification of lodging rice. The result showed that the overall accuracy of 87.18% is achieved with the combination of ORFPs and decision tree.