The semantic segmentation methods based on deep neural networks predominantly employ supervised learning, relying heavily on the quantity and quality of annotated samples. Due to the complexity of high-resolution remote sensing imagery, obtaining sufficient and precise pixel-level labeled data is highly challenging. This letter introduces a novel self-supervised learning method using a pretrained denoising diffusion probabilistic model (DDPM) to leverage semantic information from large-scale unlabeled remote sensing imageries. Building on this, a multistage fusion scheme between pretrained features and high-resolution features is proposed, enabling the network to learn more effective strategies to leverage prior information provided by the pretrained model while preserving the rich semantic details of high-resolution images. Experimental results on two remote sensing semantic segmentation datasets show that the proposed Diff-HRNet outperforms all compared methods, demonstrating the potential of pretrained diffusion models in extracting crucial feature representations for semantic segmentation tasks.

A key issue of existing deep-learning-based object detection methods in remote sensing images is that they often struggle to differentiate the background and small object regions due to multi-level downsampling operations therein. Downsampling operations help extract high-level semantic features but result in excessive loss of spatial features of small objects. In this paper, we propose a new small object detector using multispectral remote sensing images, named content-aware downsampling-based detection network (CADDN), where we newly design a content-aware downsampling-based module (CADM). Unlike conventional downsampling operations that apply uniform downsampling parameters across the entire feature map, CADM adaptively assigns higher weights to feature elements that are critical for distinguishing objects from the background, and this assignment is guided by the contextual awareness of object locations during the downsampling process. Experiments based on multispectral remote sensing images with small ships and vehicles demonstrate that CADM can accurately identify and preserve the locations of important object-related features, and CADDN correspondingly achieves superior small object detection performance than state-of-the-art methods.

Forests are the most extensive terrestrial ecosystems in the world. Their vertical structure information not only reflects the spatial structure characteristics of the forest but also the physiological and ecological processes. The existing studies on forest vertical structure through remote sensing often encounter challenges such as high costs, difficulties in acquiring effective data, or complexities in data processing. In this study, a new technology for detecting the vertical structure of vegetation is proposed and validated, which can obtain the vertical structure of vegetation by single flight. In order to adopt the new technology, a radar system was developed, incorporating a modular design scheme that emphasizes high integration and lightweight characteristics, and it was deployed on an unmanned aerial vehicle (UAV) flight platform. It consists of a main control unit, a signal processing unit, and a data recording unit, which weighs only 0.92 kg in total. Meanwhile, a novel algorithm is proposed to reconstruct the vertical structure of targets, effectively addressing critical challenges in UAV radar signal processing, such as strong system coupled signal, significant noise in radar signals, and high sidelobes in images. In order to validate the capability of the new technology, UAV flight experiments, as well as in situ observation, were carried out in typical vegetation areas. The proposed algorithm was used to process the acquired radar echoes, yielding a root-mean-square error (RMSE) of 1.33 m for the vegetation height compared to the ground synchronous measurement.

Quad-pol synthetic aperture radar (SAR) is one of the most effective approaches for dry and wet snow identification data, but its applicability is limited by the high cost of quad-pol SAR data. Dual-pol SAR such as Sentinel-1 has larger spatial coverage, longer time sequences, and freely accessible data, but there is still a highly uncertainty in dual-pol SAR to distinguish dry and wet snow due to limited polarimetric information. In this study, a pixel neighborhood-based snow identification algorithm was developed and verified using dual-pol C-band SAR data in Northern Xinjiang, China. A total of six decomposed parameters were obtained to characterize the polarimetric information of dual-pol SAR data by modifying the H-$\alpha $decomposition applicable to dual-pol SAR data. In the case of limited training samples, polarimetric features that were most sensitive to snow identification were selected as the optimal features for support vector machine (SVM), and the result derived from SVM was employed as the initial labels of Markov random field (MRF) model to separate dry and wet snow using iterative conditional mode (ICM). Then, the proposed algorithm, dual-pol SVM-MRF (DSVM-MRF), was validated and compared with previously published methods. The results show that the DSVM-MRF acquires the superior snow recognition with the overall accuracy (OA) and Kappa coefficient of 84.5% and 0.58%, respectively.

Distribution of snow and its melting is a critical factor affecting local weather, avalanche and flood forecasting, livelihood of people residing, and hydropower production. Most of the existing dry and wet snow identification methods were based on expensive quad-pol SAR with finite generalizability, while dual-pol SAR with larger coverage, longer time series and open availability has more advantages. In this study, an unsupervised algorithm for dry and wet snow discrimination, NSAE-WFCM, is proposed based on a variety of polarimetric features derived from H-α decomposition in dual-pol mode using C-band Sentinel-1 SAR data. NSAE-WFCM constructs a deep training network using the pixel neighborhood-based sparse autoencoder (NSAE) to optimize polarimetric parameters, and inputs reconstructed features with different weights into feature-weighted fuzzy C-means clustering (WFCM) to distinguish dry and wet snow for each underlying surface. Ground observation was carried out during the snow melting period of March 2021 in Altay, China, to validate dual-pol NSAE-WFCM method with an overall accuracy and kappa coefficient of 88.8% and 0.68, respectively. The results show that NSAE-WFCM’s accuracy is similar to that of the quad-pol SAR-based dry and wet snow result (90.0%), and significantly better than that of previously published approaches extended to dual-pol SAR, such as SVM (76.7%), H-α-Wishart (65.5%), SPAN-based threshold method (51.7%), and wet snow-based method (43.1%). Therefore, the NSAE-WFCM algorithm improves the ability to classify wet and dry snow based on dual-pol polarimetric features, overcomes the high dependence of existing methods on quad-pol SAR data, and reduces manual interpretation by using unsupervised clustering.

Snow density is an important variable in snowpack research. The comprehensive applicability evaluation of the snow density datasets is a prerequisite of these datasets for their applications in hydrology processes and climate change, as well as in snow equivalent water retrieval algorithms. In this letter, the applicability of three snow density datasets, including European ReAnalysis (ERA)-Interim, ERA5, and the newly released ERA5-Land datasets, was first assessed using two ground evaluation datasets with different land covers from seven snow survey courses and four densely sampled networks in China. The results show that the ERA-Interim dataset significantly overestimates snow density during the entire snow season, with an overall root mean square error (RMSE) larger than 112 kg/m3, and lacks temporal dynamics. The ERA5 and ERA5-Land datasets are generally in good agreement with the ground measurements in China. The averaged RMSEs of the ERA5 dataset are 56.2 kg/m3 against snow course sites and 28.3 kg/m3 versus the densely sampled measurements, and those of the ERA5-Land dataset are 56.6 and 28.4 kg/m3, respectively. However, the ERA5 and ERA5-Land datasets still underestimate snow density over time, especially for the middle and late snow seasons. These new findings are expected to provide valuable feedback to model developers to further enhance the accuracy of snow density datasets.

Sensitivity analysis (SA) of model parameters is of great importance for understanding, development, and application of models. However, the influence of snow microstructure variability on snow water equivalent retrieval from passive microwave measurements is still unclear. This article explores the parameter sensitivity of the microwave emission model of layered snowpacks (MEMLS) with improved born approximation (IBA) by using a quantitative global SA method, the extended Fourier amplitude sensitivity test (EFAST) algorithm. A deep analysis is conducted, including the sensitivity of passive microwave emission to snow parameters, the sensitivity variation analysis for different snow conditions, and the temporal properties of the parameter sensitivity. The results show the exponential correlation length, snow depth, and snow density are the three most sensitive parameters for snow without salt in the MEMLS model for the brightness temperature gradient at 18.7 and 36.5 GHz. For snow with a small salt content, the exponential correlation length, snow depth, snow temperature, and snow density are the four most sensitive parameters. Second, snow parameter variability highly affects the microwave radiation. The sensitivity values of microwave brightness temperature to snow depth gradually increase when the exponential correlation length is less than 0.25 mm and then slightly decreases with the increase of exponential correlation length and decreases along with the increase of snow density. Finally, our analysis highlights the importance to include the snow density, especially for deep snow depth, in the combination of sensitive factors in future multiparameter retrievals.

Accurate monitoring of the global snow cover is important in understanding the impact of the climate on snow cover. Google Earth Engine (GEE) is a cloud computing platform dedicated to satellite imagery and other Earth observation data. Taking the Landsat TM data as true data, this paper evaluated and analyzed MODIS snow cover products by GEE in snow seasons. Using GEE JavaScript program, we obtain that error rate of MODIS snow cover product, the missing rate during the snow season with finally 88.94% of overall average accuracy in the test sites. The results showed that the GEE platform can be used to assess accurately the snow products with high efficiency.