Soil moisture (SM) data can provide guidance for decision-makers in fields such as drought monitoring and irrigation management. Soil Moisture Active Passive (SMAP) satellite offers sufficient spatial resolution for global-scale applications, but its utility is limited in regional areas due to its lower spatial resolution. To address this issue, this study proposed a downscaling framework based on the Stacking strategy. The framework integrated extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), and categorical boosting (CatBoost) to generate 1 km resolution SM data using 15 high-resolution factors derived from multi-source datasets. In particular, to test the influence of terrain partitioning on downscaling results, Anhui Province, which has diverse terrain features, was selected as the study area. The results indicated that the performance of the three base models varied, and the developed Stacking strategy maximized the potential of each model with encouraging downscaling results. Specifically, we found that: (1) The Stacking model achieved the highest accuracy in all regions, and the performance order of the base models was: XGBoost > CatBoost > LightGBM. (2) Compared with the measured SM at 87 sites, the downscaled SM outperformed other 1 km SM products as well as the downscaled SM without partitioning, with an average ubRMSE of 0.040 m³/m³. (3) The downscaled SM responded positively to rainfall events and mitigated the systematic bias of SMAP. It also preserved the spatial trend of the original SMAP, with higher levels in the humid region and relatively lower levels in the semi-humid region. Overall, this study provided a new strategy for soil moisture downscaling and revealed some interesting findings related to the effectiveness of the Stacking model and the impact of terrain partitioning on downscaling accuracy.

Timely and accurate soil moisture information is of great importance in agricultural monitoring. The Gaofen-3 (GF-3) satellite, the first C-band multi-polarization synthetic-aperture radar (SAR) satellite in China, provides valuable data sources for soil moisture monitoring. In this study, a soil moisture retrieval algorithm was developed for the GF-3 satellite based on a backscattering coefficient simulation database. We adopted eight optical vegetation indices to determine the relationships between these indices and vegetation water content (VWC) by combining Landsat-8 data and field measurements. A backscattering coefficient database was built using an advanced integral equation model (AIEM). The effects of vegetation on backscattering coefficients were corrected using the water cloud model (WCM) to obtain the bare soil backscattering coefficient ( ${\sigma }_{soil}^{°}$ ). Then, soil moisture retrievals were obtained at HH, VV and HH+VV combination respectively by minimizing the observed bare soil backscattering coefficient ( ${\sigma }_{soil}^{°}$ ) and the AIEM-simulated backscattering coefficient ( ${\sigma }_{\mathit{\text{soil-simu}}}^{°}$ ). Finally, the proposed algorithm was validated in agriculture region of wheat and corn in China using ground soil moisture measurements. The results showed that the normalized difference infrared index (NDII) had the best fit with measured VWC values (R = 0.885) among the eight vegetation water indices; thus, it was adopted to correct the effects of vegetation. The proposed algorithm using GF-3 satellite data performed well in soil moisture retrieval, and the scheme combining HH and VV polarization exhibited the highest accuracy, with a root mean square error (RMSE) of 0.044 m³m⁻³, followed by HH polarization (RMSE = 0.049 m³m⁻³) and VV polarization (RMSE = 0.053 m³m⁻³). Therefore, the proposed algorithm has good potential to operationally estimate soil moisture from the new GF-3 satellite data.

The soil dielectric constant model is essential for retrieving soil properties based on microwave remote sensing. However, the existing saline soil dielectric constant models perform poorly in simulating the dielectric constant of soil with high water content and salinity. In this study, the Wang Yueru (WYR) saline soil dielectric constant model, which was demonstrated to perform well in describing the effect of salinity and moisture on the dielectric constant, was validated based on experimental measurements of soil samples under different water content and salinity degrees. Furthermore, we adjusted the model form, refitted the empirical coefficient in the model, and finally acquired a two-stage model for simulating the soil dielectric constant. The enhanced model was validated under different soil moisture and salinity ranges using experimental measurements of soil samples. Compared to the original model, the proposed model exhibits a larger improvement in simulating the soil dielectric constant, and the RMSE of the simulated results dramatically decreased from 7.3 to 1.6, especially for soil with high salinity and water content. On this basis, a model suitable for L-band microwave was established. This model is of great significance for studying soil dielectric characteristics and retrieving soil parameters based on L-band data. Furthermore, this model can be used to retrieve soil salinity and water content using microwave remote sensing under a broadened application situation, such as in saline-alkali soils, wetlands, and salt marshes.

Due to the coarse spatial resolution of currently available microwave (mostly passive) soil moisture (SM) products, it is difficult to apply these SM data in watersheds or at local scales. To this end, a number of downscaling approaches have been developed to improve the spatial resolution of microwave SM products. Specifically, the optical-/thermal-based downscaling methods are most widely used in recent decades. However, such methods normally rely on instantaneous optical/thermal land surface parameters, which are commonly inapplicable under cloudy conditions. The purpose of this study is to develop a new downscaling method based on the temporal variation in geostationary satellite-derived land surface temperature and net surface shortwave radiation. The proposed method has a certain potential to improve data availability under cloudy conditions, because geostationary satellites are capable of providing land surface parameters at high temporal resolution. The proposed method was tested over the REMEDHUS network in Spain. The scaling strategy of cumulative distribution function matching was used to remove systematic differences in spatial mismatch between satellite pixels and in situ SM measurements. Results indicate that the downscaled SM agrees well with in situ measurements and has comparable accuracy with the original microwave SM product. The overall root mean square errors with the in situ measurements for the original microwave SM and the downscaled SM are 0.054 and 0.057 $\text{m}^{3}/\text{m}^{3}$ , respectively. This method not only has a successful attempt to downscale microwave SM data using temporal information but also has the potential to avoid the failure of traditional instantaneous observations-based downscaling procedure due to clouds.

Accurately mapping the surface water bodies through remote sensing technology is of great significance for water resources management, flood monitoring, and drought monitoring. At present, many scholars at home and abroad carry out research on deep learning image recognition algorithms based on convolutional neural networks, and a variety of variant-based convolutional neural networks are proposed to be applied to extract water bodies from remote sensing images. However, due to the low depth of convolutional layers employed and underutilization of water spectral feature information, most of the water body extraction methods based on convolutional neural networks (CNNs) for remote sensing images are limited in accuracy. In this study, we propose a novel surface water automatic extraction method based on the convolutional neural network (CRAUnet++) for Sentinel-2 images. The proposed method includes three parts: (1) substituting the feature extractor of the original Unet++ with ResNet34 to enhance the network’s complexity by increasing its depth; (2) Embedding the Spatial and Channel ‘Squeeze and Excitation’ (SCSE) module into the up-sampling stage of the network to suppress background features and amplify water body features; (3) adding the vegetation red edge-based water index (RWI) into the input data to maximize the utilization of water body spectral information of Sentinel-2 images without increasing the data processing time. To verify the performance and accuracy of the proposed algorithm, the ablation experiment under four different strategies and comparison experiment with different algorithms of RWI, FCN, SegNet, Unet, and DeepLab v3+ were conducted on Sentinel-2 images of the Poyang Lake. The experimental result shows that the precision, recall, F1, and IoU of CRAUnet++ are 95.99%, 96.41%, 96.19%, and 92.67%, respectively. CRAUnet++ has a good performance in extracting various types of water bodies and suppressing noises because it introduces SCSE attention mechanisms and combines surface water spectral features from RWI, exceeding that of the other five algorithms. The result demonstrates that CRAUnet++ has high validity and reliability in extracting surface water bodies based on Sentinel-2 images.

Snow depth is a critical climate indicator and a key parameter for Arctic sea ice retrieval. In this study, we retrieve pan-Arctic snow depth from 2012 to 2021 by integrating satellite altimetry, passive microwave brightness temperatures, and multi-source ground/airborne data. We employ four machine learning models—Light Gradient Boosting Machine (LightGBM), Multiple Linear Regression (MLR), Random Forest (RF), and Long Short-Term Memory (LSTM)—to leverage the complementary strengths of altimetry and microwave datasets while evaluating the performance of different machine learning (ML) architectures. Through permutation feature importance analysis, we identified that the 89 GHz polarization ratio has a significantly greater influence on snow depth retrieval over multi-year ice compared to that over first-year ice. Validation against Operation IceBridge and MOSAiC measurements reveals complementary strengths of snow retrieval among the models. The MLR model achieves the highest overall snow depth accuracy (root-means-square-error = 7.19 cm, correlation = 0.67 against OIB), while the LSTM demonstrates minimal mean bias of snow depth between satellite-based and in situ observations (1.98 cm against OIB; 0.30 cm against MOSAiC). All ML models exhibit robust generalization capabilities. Our retrieved snow depth products improved sea ice thickness estimation significantly, reducing bias between satellite-based and a standard climatology-based ice thickness product by nearly an order of magnitude. Our long-term snow products offer users a reliable, high-accuracy dataset for advancing Arctic energy budget modeling and sea ice studies.