Multisource remote sensing images (RSIs) can capture the complementary information of ground objects for use in semantic segmentation. However, there can be inconsistency and interference noise among the multimodal data from different sensors. Therefore, it is a challenge to effectively reduce the differences and noise between the different modalities and fully utilize their complementary features. In this article, we propose a universal cross-fusion transformer framework (CFFormer) for the semantic segmentation of multisource RSIs, adopting a parallel dual-stream structure to extract features separately from the different modalities. We introduce a feature correction module (FCM) that corrects the features of the current modality by combining features from the other modalities in both the spatial and channel dimensions. In the feature fusion module (FFM), we employ a multihead cross-attention mechanism to interact globally and fuse features from the different modalities, enabling the comprehensive utilization of the complementary information in multisource RSIs. Finally, comparative experiments demonstrate that the proposed CFFormer framework not only achieves state-of-the-art (SOTA) accuracy but also exhibits outstanding robustness when compared to the current advanced networks for semantic segmentation of multisource RSIs. Specifically, CFFormer achieves a mean intersection over union (mIoU) of 58% and an overall accuracy (OA) of 85.35% on the WHU-OPT-SAR dataset, outperforming the second-ranked network by 4.71% and 1.74%, respectively. On the Vaihingen and Potsdam datasets, CFFormer also achieves the best results, with mIoU and OA values of 84.31%/91.88% and 88.62%/92.64%, respectively. The source code is available athttps://github.com/masurq/CFFormer.

Synthetic Aperture Radar (SAR) is vital for coastal wetland mapping due to its cloud and vegetation penetration capabilities. However, using SAR imagery in wetland mapping is challenged by similar backscatter coefficients and speckle noise. To address these issues, a patch-based cascade forest (PBC) method is proposed in this paper, which combines multi-temporal, full-polarization GF-3 SAR data to map typical ground objects in the Yellow River Delta wetlands. Comparative experiments are conducted using different temporal datasets, and the proposed method is compared with Random Forest (RF), Support Vector Machines (SVM), eXtreme Gradient Boosting (XGB), and multi-Grained Cascade Forest (gc-Forest). The results show that the multi-temporal patch-based cascade forest method, achieving an Overall Accuracy (OA) of 89.46% and a Kappa coefficient of 0.872, significantly outperforms single-temporal methods and other machine learning algorithms, which proved that our method fit for long-term, high-quality wetland classification.

The amplitude and phase of synthetic aperture radar (SAR) backscatter coefficients are sensitive to surface dielectric constant, canopy structure, and surface roughness. Particularly, quad-pol SAR observations which provide HH, VV, and HV/VV polarimetry information are more beneficial for crop monitoring compared with single and dual-pol data. Considering the difference of penetrability of horizontal and vertical signals over oriented canopy structure, we explores the potential of co-pol and cross-pol phase difference, like ∆ϕHH−VVand ∆ϕHH−HVand ∆ϕVV−VH, for crop phenology and structure monitoring. Besides, time-series variation of phase difference is supposed to be able to track crop growth. This paper introduces the form of the co-pol and cross-pol phase difference derived from Sinclare scattering matrix (S2×2) and the covariance matrix (C3×3). Subsequently, the mechanism of phase difference relating with the scattering phase center, the oriented crop structure, and the penetration depths of different polarization are provided. Time-series L-band quad-pol UAVSAR data which are collected from June 22 to July 17, 2012 in Winnipeg, Canada are used for the analysis of temporal and spacial phase difference characteristics over canola, soybean, and wheat fields. Results indicate significant potential of co-pol and cross-pol phase differences in crop structure and phenology monitoring. Especially, the evolution of time-series ∆ϕHH−VVis more consistent with taht of LAI measurements compared with the cross-pol phase difference.

Synthetic Aperture Radar (SAR) exhibits the capacity for comprehensive and continuous Earth observation, regardless of weather conditions and diurnal variations. Contemporary methodologies for rice field identification utilizing SAR rely upon the entirety of the rice growth cycle data, posing challenges in discerning rice cultivation within the ongoing cultivation cycle. In addressing this challenge, the present study introduces a novel metric, namely the 3-Sigmoid index (SSSI), designed to quantify the early season probability of land parcels planted rice. SSSI fully uses the crucial feature of intensity variation from low to high backscatter based on time series SAR data as paddy fields progress in their growth stages, so that we can identify in-season rice early. The method was validated in two experimental areas, and the experimental results indicate that the approach exhibits heigh accuracy in early-stage identification. Moreover, the method demonstrated successful recognition of rice fields during the tillering phase and even prior to it. In addition, the SSSI is independence from requisite prior knowledge, reference samples, and a plethora of pre-established parameters. This characteristic underscores its potential for widespread and extensive practical applications, particularly in regions characterized by persistent cloud cover, where optical remote sensing data is frequently inaccessible.

Monitoring rice growth from the space has aroused the interests of researchers from all over the world. Previous researches about rice phenology focused on dividing paddy rice into different phenological intervals, without knowing the specific date when a certain phenological stage starts. In this paper, different characteristics of the start of three important phenological stages are analyzed. A set of C-band SAR images acquired by SENTINEL-1 has been used to retrieve the start of three key phenological stages (leaf development, stem elongation and inflorescence emergence) of paddy rice. A dynamic time warping method was applied for the alignment of SAR time series curves, which are constructed using backscattering coefficients of different polarimetric channels (VH, VV and VH/VV ratio). Results show that all three kinds of time series curves show great potential in detecting start of leaf development of rice and VH/VV curves perform better in recognizing start of inflorescence emergence.

The natural surface soil is often mixed with a lot of sand and gravel. In this paper, a new soil moisture inversion method for gravel areas from polarimetric SAR (PolSAR) data is proposed. First, the backscattering of gravel areas is divided into two parts: surface scattering and volume scattering, which are obtained by the polarimetric decomposition method. For the volume scattering part, the Dense Medium Radiative Transfer (DMRT) model is used to obtain the soil moisture, and the Advanced Integral Equation Model (AIEM) and the Oh model are used for the surface scattering part. Finally, the weighted sum of the inversion results of the two parts is taken as the final inversion result. The accuracy of the proposed method was evaluated by field soil moisture data from Wuhai city, Inner Mongolia and ALOS-2 PolSAR data.

A modified model-based polarimetric decomposition method, considering both the surface and dihedral scattering depolarization, is proposed for soil moisture retrieval. In the parameter solution, it works combining at least two polarimetric SAR images acquired simultaneously in different incidence angles. Moreover, it needs not to decide whether dihedral or surface scatter is the dominant contribution. The experiments to demonstrate the potential of the proposed approach is carried out using L-band polarimetric UAVSAR multi-incidence data in Winnipeg, Canada. The scattering mechanism of forest, grass land, urban, and barren area are analyzed compared with Yamaguchi three-component decomposition results. The performance of the soil moisture estimation algorithm is also assessed by comparing the retrieval results with in situ measurements.