With the rapid advancements in deep learning technology, the field of remote sensing change detection (RSCD) has witnessed significant improvements and innovations. In this context, bitemporal image processing, using features directly extracted by the backbone for subsequent fusion operations, may be obstructed by external environmental factors, potentially limiting the effective capture of complex feature variations. Moreover, overlooking local features during the fusion of bitemporal features can significantly affect the final detection results. As a result, achieving accurate change detection (CD) still encounters various challenges. To tackle these issues, this paper proposes a CD network (AEGL-Net) with Adaptive Multiscale Enhancement (AME) and Global-Local Feature Fusion (GLFF) modules. First, AME enhances features at each stage of backbone extraction through an adaptive strategy, balancing the enhancement of semantic information and texture details. Then, GLFF is used to fuse the bitemporal image features, which enhances the modeling of global dependencies while also fusing shared and context-aware weights to enhance the local features. Finally, the merged features are fed into the decoder to generate precise change maps. Experiments conducted with four open RSCD datasets (LEVIR-CD, S2Looking, SYSU-CD, and UAV-CD) demonstrate that our proposed AEGL-Net outperforms ten state-of-the-art models in the RSCD field. Our code is available at https://github.com/yikuizhai/AEGL-Net.

Although deep learning has advanced remote sensing change detection (RSCD), most methods rely solely on image modality, limiting feature representation, change pattern modeling, and generalization—especially under illumination and noise disturbances. To address this, we propose MMChange, a multimodal RSCD method that combines image and text modalities to enhance accuracy and robustness. An Image Feature Refinement (IFR) module is introduced to highlight key regions and suppress environmental noise. To overcome the semantic limitations of image features, we employ a vision-language model (VLM) to generate semantic descriptions of bi-temporal images. A Textual Difference Enhancement (TDE) module then captures fine-grained semantic shifts, guiding the model toward meaningful changes. To bridge the heterogeneity between modalities, we design an Image-Text Feature Fusion (ITFF) module that enables deep cross-modal integration. Extensive experiments on LEVIR-CD, WHU-CD, and SYSU-CD demonstrate that MMChange consistently surpasses state-of-the-art methods across multiple metrics, validating its effectiveness for multimodal RSCD. Code is available at: https://github.com/yikuizhai/MMChange.

Remote sensing technologies, such as aerial photography and satellite remote sensing, play a crucial role in assessing and monitoring the changes in land use and land cover (LULC) for large areas. Deep learning methods are recently becoming popular for LULC classification and have maintained promising performance in many applications. However, challenges remain in LULC classification for the complex semantic appearance in high-resolution remote sensing images. In this paper, we proposed a novel framework that cascades two deep learning models i.e., ResNet and SegNet for land use and land cover classification. The advantages of strong semantic feature extraction (ResNet) and efficient boundary delineation (SegNet) capabilities of these two models can be combined to derive highly accurate land use and land cover classification results. Our method performed well on the recently developed multisource Dense-Pixel Annotation Dataset Globe230K.

The rise of multimodal big-earth data and deep learning in recent years has promoted the development of multimodal image registration. Registration of optical and synthetic aperture radar (SAR) is significantly enhanced by the geometrical differences between the two modalities, apart from the unimodal image registration. Despite the widespread use of deep learning, however, some shadows still exist in this direction, such as insufficient training data and mismatching in local regions. In this work, the potential of reinforcement learning is explored in optical and SAR image registration. Specifically, reinforcement learning explores the possible displacement directions and magnitudes of the four predefined corner points in a specific search space, to solve the affine or homograph matrix based on the displacements of the corner points, and finally obtain the alignment results of different modes. Results show that with reinforcement learning, there is improvements in the visualization of image registration.

The Earth observation era has bestowed dividends upon supervised land cover classification based on deep learning and optical data. However, limitations, such as insufficient spectral information and reduced quality during inclement weather for optical data, coupled with the need for extensive labeled samples, impede accurate classification. This letter harnesses multimodal images with deep contrastive learning to reduce reliance on labeled data and classify land covers. By employing a well-designed contrastive learning method with triangular similarity loss, our model can learn effective multimodal features without labeled samples. Moreover, the learned features are fused at the early feature level and used for the downstream classification task with fewer labeled samples. Experimental results demonstrate the benefits of incorporating multiple modalities, highlighting the potential of combining multimodal image analysis and contrastive learning for land cover classification with limited labeled samples.

Hyperspectral images (HSIs) provide exceptional spatial and spectral resolution of a scene, crucial for various remote sensing applications. However, the high dimensionality, presence of noise and outliers, and the need for precise labels of HSIs present significant challenges to the analysis of HSIs, motivating the development of performant HSI clustering algorithms. This paper introduces a novel unsupervised HSI clustering algorithm—Superpixel-based and Spatially-regularized Diffusion Learning (S2DL)—which addresses these challenges by incorporating rich spatial information encoded in HSIs into diffusion geometry-based clustering. S2DL employs the Entropy Rate Superpixel (ERS) segmentation technique to partition an image into superpixels, then constructs a spatially-regularized diffusion graph using the most representative high-density pixels. This approach reduces computational burden while preserving accuracy. Cluster modes, serving as exemplars for underlying cluster structure, are identified as the highest-density pixels farthest in diffusion distance from other highest-density pixels. These modes guide the labeling of the remaining representative pixels from ERS superpixels. Finally, majority voting is applied to the labels assigned within each superpixel to propagate labels to the rest of the image. This spatial-spectral approach simultaneously simplifies graph construction, reduces computational cost, and improves clustering performance. S2DL’s performance is illustrated with extensive experiments on four publicly available, real-world HSIs: Indian Pines, Salinas, Salinas A, and WHU-Hi. Additionally, we apply S2DL to landscape-scale, unsupervised mangrove species mapping in the Mai Po Nature Reserve, Hong Kong, using a Gaofen-5 HSI. The success of S2DL in these diverse numerical experiments indicates its efficacy on a wide range of important unsupervised remote sensing analysis tasks.

With the advancement of deep learning (DL) technologies, remarkable progress has been achieved in change detection (CD). Existing DL-based methods primarily focus on the discrepancy in bitemporal images, while overlooking the commonality in bitemporal images. However, one of the reasons hindering the improvement of CD performance is the inadequate utilization of image information. To address the above issue, we propose a Deeply Supervised Hybrid Feature Aggregation Network (DS-HyFA-Net). This network predicts changes by integrating the distinctness and the commonality in bitemporal images. Specifically, the DS-HyFA-Net primarily consists of a set of encoders and a Hybrid Feature Aggregation (HyFA) module. It uses a Siamese encoder (or Encoder I) and a specialized encoder (or Encoder II) to extract distinct and common features (CFs) in bitemporal images, respectively. The HyFA module efficiently aggregates distinct and common features (or hybrid features) and generates a change map using a predictor. In addition, a common feature learning strategy (CFLS) is introduced, based on deeply supervised (DS) techniques, to guide Encoder II in learning CFs. Experimental results on three well-recognized datasets demonstrate the effectiveness of the innovative DS-HyFA-Net, achieving F1-Scores of 93.33% on WHU-CD, 90.98% on LEVIR-CD, and 81.14% on SYSU-CD. Our code is available athttps://github.com/yikuizhai/DS-HyFA-Net.

Change detection (CD) is a process of extracting changes on the Earth’s surface from bitemporal images. Current CD methods that use high-resolution remote sensing images require extensive computational resources and are vulnerable to the presence of irrelevant noises in the images. In addressing these challenges, a comparison-based attention Siamese network (CAS-Net) is proposed. The network utilizes contrastive attention modules (CAMs) for feature fusion and employs a classifier to determine similarities and differences of bitemporal image patches. It simplifies pixel-level CDs by comparing image patches. As such, the influences of image background noises on change predictions are reduced. Along with the CAS-Net, an unmanned aerial vehicle (UAV) similarity detection (UAV-SD) dataset is built using high-resolution remote sensing images. This dataset, serving as a benchmark for CD, comprises 10000 pairs of UAV images with a size of$256 \times 256$. Experiments of the CAS-Net on the UAV-SD dataset demonstrate that the CAS-Net is superior to other baseline CD networks. The CAS-Net detection accuracy is 93.1% on the UAV-SD dataset. The code and the dataset can be found athttps://github.com/WenbaLi/CAS-Net.

Remote sensing change detection (RSCD) aims to identify changes within bi-temporal registered images. However, existing deep learning (DL)-based RSCD networks often suffer from large numbers of parameters, high computational complexity, and low inference speed, making it challenging to achieve efficient inference in real-world deployments. In addition, current models lack robust feature-fitting capabilities, necessitating the development of an efficient and powerful RSCD model to address this issue. Therefore, we propose a novel RSCD network named efficient adjacent feature harmonizer network (EAFH-Net) with fast computational speed and lightweight design. It is based on MobileNetV2, considering that change maps of different sizes contain temporal information of bitemporal features and spatial information at various scales, we introduce a multiscale feature neighbor fusion module (MFNFM) to address the lack of interaction between sophisticated-level and elementary-level features, and spatial and channel feature harmonizer module (SCFHM) to harmonize the spatiotemporal information of the change maps. Moreover, data-driven DL algorithms face another challenge due to insufficient granularity and the need for more practical datasets. Therefore, we present unmanned aerial vehicle (UAV)-CD+, a dataset comprising 2002 pairs of bi-temporal UAV low-altitude images, each sized at$1024\times 1024$. We performed experiments on three publicly accessible datasets in conjunction with UAV-CD+, comparing the results with other state-of-the-art (SOTA) methods. EAFH-Net attains the utmost precision, obtaining 91.74% on LEVIR-CD, 84.28% on SYSU-CD, 95.07% on WHU-CD, 79.12% on CLCD, and 70.12% on UAV-CD+. We have our model code available at the following link:https://github.com/yikuizhai/UCSFH-Net.

Accurate and timely monitoring of the functional urban impervious surfaces (FUIS), e.g., ports, roads, residential and non-residential buildings, is vital but challenging for coastal cities due to their diverse land covers and complicated weather. Synthetic aperture radar (SAR), with its all-weather working capability, provides a unique opportunity to monitor coastal cities promptly and regularly. This study develops a scattering texture hierarchical fusion network (STHFN) to integrate SAR scattering characteristics for more accurate recognition of FUIS based on the scattering texture index (STI). Experimental results verify the effectiveness of STHFN with up to 8% improvement in the accuracy of FUIS classification, demonstrating the promising application of STHFN in utilizing SAR backscattering features for coastal urban monitoring.

The rapid urbanization process over metropolitan cities often came with the dramatic urban land transformation of different natural land cover types into urban impervious surfaces (UIS) or UIS back to natural land covers. However, most UIS mapping studies usually focus on inter-year variation, suggesting that the urbanization process is irreversible. The rapid intra-year urban reconstruction and urban renewal are under-explored. This study develops a new algorithm for intra-year urban renewal using all available Sentinel-1 and Sentinel-2. Time series SAR and optical data are utilized for intra-year UIS change detection, i.e., to detect the changes from UIS to other land cover types. Results showed that the urban renewal process occurred within the metropolitan area, i.e., UIS changed into the bare land or vegetation. The overall accuracy and kappa of urban renewal change detection are 72.79% and 0.4599, respectively.

Urban land cover (ULC) is a fundamental indicator of urbanization, while cloud cover hinders accurate and timely ULC monitoring. The fusion of synthetic aperture radar (SAR) and cloud-free optical data has shown good performance in previous studies, while there is a lack of investigation in cloud-prone areas where optical data is contaminated by clouds. This study proposes a cloud-oriented framework for fusing the two data sources for ULC classification in cloud-prone areas. For alleviating cloud interference, the framework proposes a cloud probability weighting strategy and a pixel-wise cloud dictionary learning algorithm considering the interference difference in different cloud probability levels. Experiments show that all algorithms using fused data improve the overall accuracy (OA) of above 6% and 20% compared with using single SAR and optical data, respectively. Compared with traditional SVM, RF, and dictionary learning methods which ignore cloud interference and directly concatenate optical and SAR features, the proposed method shows a significant improvement of 3% in OA. It improves almost all land covers' producer accuracy (PA) and user accuracy (UA), up to 9%. Further experiments with three cloud probability level samples find that the higher the cloud probability, the lower the classification accuracy of the sample. At each probability level, the proposed pixel-wise cloud dictionary learning method improves more than 2% in OA, improves up to 4% to 10% in PA and UA.

Mapping the land cover/use type of urban area surface plays a vital role in many remote sensing applications. The performance of classification is inevitably limited by the finite amount of information available from a single data source, the restricted atmosphere condition and the complex landscape of the urban areas. Even when multiple sources of data are used, the fusion strategy is relatively homogeneous. In this paper, we aim to explore the potential of transformer based fusion method in mapping the urban regions with optical and synthetic aperture radar images. Specifically, we propose a hybrid fusion transformer network that simultaneously implements multi-source data fusion at both the feature and the decision levels. The experiments are conducted on the high resolution multiple remote sensing images, and the results show that the hybrid fusion based on transformer can achieve 82.17% in overall accuracy (OA) and 76.91 % in kappa coefficient. Moreover, compared with convolution neural network based methods, the transformer based methods are on average 2% higher in OA and 3.6% higher in kappa coefficient.

Urban impervious surfaces (UIS) indicate the environmental and socioeconomic influences of rapid urbanization. Synthetic aperture radar (SAR) reflects the scattering behaviors of different land covers while multispectral data demonstrate their physicochemical properties. Numerous studies reported that the incorporation of SAR and optical data supplement each other for better extracting UIS, nevertheless, the shadow and layover effects remain unclear, especially in very high-resolution observations. This study analyzed the shadow and layover influences from both optical and SAR data for fine resolution UIS estimation. Given the SAR shadow and layover distribution, we proposed a multisource shadow-based fuzzy set (MSFS) approach for fusing optical and SAR in optical shadow areas using decision fusion. SAR layovers showed effectiveness in UIS extraction. MSFS delivered 3% and 7% improvement in overall accuracy compared with SVM and RF using feature fusion respectively.

Accurate mapping of mangroves species is essential for mangrove management, and deep learning of hyperspectral images (HSIs) shows a great advantage in classification with the fine spectrum. However, the sparely available annotations of HSIs are key challenges for accurate mapping using deep learning, especially for mangrove species within small patches. In this work, a high spatial resolution HSI is synthesized using the method of hyperspectral-multispectral image fusion with spectral variability, providing augmented samples as well as spatial information of mangroves. Secondly, the latest 3D convolutional neural network (3DCNN) was investigated to explore spatial and spectral information for mangrove species mapping. Compared to Gaofen 5 using conventional machine learning methods, the synthetic image provides manyfold samples and higher accuracy for mangrove species mapping using 3DCNNs. This work is expected to improve the situation of sample shortage and spatial information deficiency for mangrove species mapping using deep learning with HSIs.

Sonneratia have posed a threat to native mangrove species in Hong Kong. Early detection of individual Sonneratia when they are introduced and naturalized before invasion is essential for native mangrove species protection, especially for Sonneratia with a strong ability of propagation. This letter aims to provide an effective way to the accurate detection of individual Sonneratia. Specifically, using very high spatial resolution remotely sensed data, we adapt the RetinaNet, incorporating multiscale features for sapling detection and convolutional neural networks for detecting the Sonneratia distributed scatteredly among native species. The Sonneratia were detected with a higher mean average precision (mAP) 0.50 of 0.3891 with a precision of 0.5465 than that from the deformable part model. In addition, 3678 Sonneratia were detected at early stage. This letter can support the government for mangrove forest management and offer a scientific guidance for adequate response to the species invasion, like annual removal of Sonneratia, and then reduce the consumption of labor and time over a large scale. In addition, it can provide a quantitative survey for Sonneratia management.

High-resolution urban impervious surface (UIS) is essential for social and environmental analysis. However, shadows have become a major challenge to the accurate UIS mapping in high-resolution optical images, as the low reflectance usually leads to misclassification of shadows as roads or waters. To solve this problem, we proposed a shadow free multisource stack sparse autoencoder (ShdFree-MS-SSAE) for urban shadow detection and compensation. Multisource data, including optical, SAR and LiDAR were used for the occlusion information recover. First, MS-SSAE was proposed for urban land cover classification, including shadow and non-shadow area. Then, shadow area in optical data was enhanced with a linear compensation method. Finally, MS-SSAE was applied to classify the enhanced shadow area and the non-shadow area. The results demonstrated that ShdFree-MS-SSAE framework was effective for UIS mapping, with an average improvement of 10%.

Zonation of mangrove species is the predictable and discrete ordering of mangrove species caused by a unique, intertidal environment. Mangrove zonation pattern is formed by complex abiotic and biotic environmental factors and, at the same time significantly influences the associate flora and fauna communities even the whole coastal ecosystem and local carbon cycle. In this study, high resolution time-series satellite images were employed to investigate the characteristics and inner structures of mangrove zonation using landscape metrics during the past decade in the Deep Bay area, Hong Kong SAR and Shenzhen, China. Both native and exotic mangrove species were discriminated and analyzed. The results of this study shown that native mangrove stands in the Deep Bay showed relatively clearer zonation during the past decade, presenting a sequence of the species in this tide-dominated shore with higher aggregation and reunion degree. In comparison, the distribution pattern of exotic species is in higher degree of fragmentation and less connectivity. The patch-based landscape metrics were proven to be effective methods in measuring and describing the spatial distribution pattern of mangrove zonation based on high resolution satellite images.

The land resource is becoming scarcer and scarcer for a rapidly developing city. Thus, the land price assessment is important for the government to auction the land appropriately. In the paper, we introduced the deep neural network to evaluate the land price, taking the Shenzhen city in China as a case. Firstly, twenty influencing factors and land price data were gathered. Then, Shenzhen city was segmented into many grids with a size of 300 × 300 m. Secondly, the land price of each grid was derived with Kriging approach based upon the samples of land price. And the twenty influencing factors was quantified. Thirdly, the land price data and influencing factors were partitioned into training and testing datasets with the ratio of 8:1, and the training data were utilized to train the deep neural network based on regression analysis and classification with different hidden layers. Finally, the results were analyzed, and the deep neural network with the highest accuracy was selected as the optimum model. Therefore, our proposed method is an efficient approach to evaluate the land price with deep neural network.

In recent years, the big data industry chain has become more mature. Analyzing and managing cities by utilizing various big data in cities has become a hot research topic. Urban functional regions discovering is one of the important applications. The mainstream in urban functional regions discovering are probabilistic topic models, such as latent Dirichlet allocation (LDA) based topic model, which seeing the regions as documents and their functions are their topics. These methods require feature engineering by hand, which will construct features of limited expressiveness. To overcome these methods' shortcomings, we introduced a deep learning topic model called document neural autoregressive distribution estimation (DocNADE) into urban functional regions mining. And we did an experiment to test its effect. The experimental result shows that this DocNADE framework has achieved a considerable result in urban function inference compared with Dirichlet Multinomial Regression (DMR) based topic model which is a state of the art of urban functional regions discovering.

Development of a city cultivates regions with different functions such as working areas and entertainment venues. People in a city usually travel among these regions in certain movement patterns. Identifying those regions will facilitate government management and promote further development of the city. In this paper, we proposed a framework to identify urban functional regions in Chengdu city based upon mobility pattern and point of interest (POIs) information extracted from mobile check-ins data. Firstly, unlike GPS trajectories, location check-ins were discontinuous. Thus, the typical mobility patterns of location check-ins was mined. Secondly, an arrival/departure matrix based on the typical mobility patterns was constructed to obtain the topics of regions by clustering POIs. Because we considered a region's function as our topics, we transferred the problem into a topic modeling problem, and applied an improved probabilistic topic model to infer functions of the regions. We evaluated our approach with 227,428 check-ins in Chengdu collected from Sina Weibo from April 12 2012 to February 16 2013. The results showed that our method outperformed baseline methods solely clustering POIs.

Not only the urbanization development but also its ecological process lays emphasis on the Impervious Surface Area (ISA) extraction, whereas, the ISA extraction from high-resolution images is challenging for both the phenomenon of the mixed pixels and shadow effects. To solve the problem, a Multi-Source Dictionary Sparse Representation Classification (MSD-SRC) method using WorldView-2 and TerraSAR-X dataset is proposed. First, it uses multi-source data and fuzzy samples by Low Pass Filtering (LPF) to solve the problem of road and building misclassification; second, learning Multi-Source Dictionary for non-shadow and shadow classes, then using discriminative sparse coding method for classification, therefore to reduce shadow effects and improve the ISA extraction accuracy. Experimental results demonstrated the effectiveness of the proposed method.

In the field of remote sensing, data acquired from a single sensor usually can't meet the needs of some special applications, because the information extracted from the data are often incomplete and limited. Data fusing can solve this problem, but it will lead to the redundant information. In this paper, we proposed a novel manifold learning approach to perform dimensionality reduction for the fusing optical and SAR data. And three typical manifold learning models, namely, ISOMAP, local linear embedding (LLE) and principle component analysis (PCA), were utilized to test the robustness of our method by comparing with the land cover classification results. Our experimental results showed that our proposed method obtained the best land cover classification results among these approaches for the fusing optical and SAR data.

Incorporating optical and SAR data to estimate impervious surface is useful but challenging due to their different geometric imaging mechanism. The recent development of deep convolutional networks (DCN) opens a promising opportunity. In this study, the typical DCN, AlexNet, was modified to estimate the impervious surface from optical and SAR data. GoogLeNet and the Support Vector Machine (SVM) were employed for comparison. Experimental results indicated the effectiveness of AlexNet with an accuracy of over 99%, outperforming both GoogLeNet and SVM. Furthermore, 60~80% of training samples outperformed the results from the whole training set under certain number of epochs, indicating that large number of training samples may not necessarily produce better results, depending on other factors (e.g. number of epochs). Generally, AlexNet was able to fuse the optical and SAR data and improved the accuracy of estimating impervious surface by about 2% compared with that using optical data alone.

Multisource remote sensing data have been widely used to improve land-cover classifications. The combination of synthetic aperture radar (SAR) and optical imagery can detect different land-cover types, and the use of genetic algorithms (GAs) and support vector machines (SVMs) can lead to improved classifications. Moreover, SVM kernel parameters and feature selection affect the classification accuracy. Thus, a GA was implemented for feature selection and parameter optimization. In this letter, a GA-SVM algorithm was proposed as a method of classifying multifrequency RADARSAT-2 (RS2) SAR images and Thaichote (THEOS) multispectral images. The results of the GA-SVM algorithm were compared with those of the grid search algorithm, a traditional method of parameter searching. The results showed that the GA-SVM algorithm outperformed the grid search approach and provided higher classification accuracy using fewer input features. The images obtained by fusing RS2 data and THEOS data provided high classification accuracy at over 95%. The results showed improved classification accuracy and demonstrated the advantages of using the GA-SVM algorithm, which provided the best accuracy using fewer features.

Under the indoor building environment, when the fires and other accidents occur, how to effectively organize the masses evacuation and fire rescue, is closely related to the safety of people's lives and property and has become a critical problem of public concern. This paper presents an improved ant colony algorithm (ACO) to solve the problem of how to optimize the evacuation route and rescue route when an accident occurs. According to the key factors affecting people emergency evacuation, such as indoor building environment, fire and its combustion products, problem of path's optimal selection, etc., we propose an emergency evacuation model, based on the model it can give an optimal evacuation route for the mass and an optimal rescue route for the firefighters. We also analyzes the search results, it shows that the search results is robust and reasonable.

In China, drought is one of the major environmental disasters, which bring great harm to the people. The middle reach of Yangtze River is the most important base to produce grains in China. Influenced by the summer monsoon, the drought occurs frequently. In our paper, the NDVI and LST from MODIS data were utilized to calculate the TVDI (Temperature Vegetation Dryness Index), which were used to monitor the drought of the study area. Meteorological drought indices were calculated from 10-day precipitation, temperature and evaporation data of 94 meteorological stations, including precipitation standardized variables, dryness and relative moisture index were used to analyze the degree of drought and the area of drought. The results showed that TVDI is significantly related to soil moisture.

Land use/land cover (LULC) classification using optical and synthetic aperture radar (SAR) remote sensing images is becoming increasingly important to produce more accurate LULC products. As an important step, feature normalization techniques have been studied by the areas of pattern recognition. Nevertheless, because of the totally different imaging mechanisms of optical and SAR sensors, most of the existing normalization approaches are not suitable for optical and SAR data fusion. Moreover, whether normalization is a significant step remains unclear regarding optical and SAR fusion. Taking the Satellite Pour l'Observation de la Terre (SPOT-5) and the Advanced Land Observing Satellite (ALOS)/Phased Array type L-band SAR (PALSAR) (HH and HV polarizations) as the optical and SAR data, this letter aims to evaluate the impact of feature normalization. Experimental results indicated that feature normalization is not necessarily significant depending on fusion methods. For instance, distribution-dependent classifiers (e.g., a maximum likelihood classifier) are independent of feature normalization; thus, it has no impact on the results when using these classifiers. Moreover, advanced classifiers (e.g., a support vector machine) with built-in normalization are also not influenced by feature normalization. In contrast, a minimum distance classifier and an artificial neural network (ANN) depend on the input values of optical and SAR features and thus can be influenced by feature normalization. However, our experiments showed a fluctuation in classification accuracy using an ANN with normalized features. Therefore, more experiments are required to investigate the optimal normalization approaches for the optical and SAR images when using an ANN as the fusion method.

Geladandong Ice Field is one of the largest ice fields in central Qinghai-Tibetan Plateau, also the water source of Yangtze River, Selin Co Lake and Chibozhang Co Lake. In this study, we aim to monitor the glacier flow rates and its change in 1990s and 2000s by Differential SAR Interferometry and offset tracking. We obtained SAR images acquired by ERS-1/2 and Envisat/ASAR in 1990s and 2000s then processed them with offset-tracking method, also validated with D-InSAR method. The result indicates that most glaciers in Geladandong Ice Fields kept stable flow rates at 15-30m/a during 1990s and 2000s. Some glaciers are identified as surge type. We selected two glaciers and studied the time series of flow velocity profiles. During the surging period, flow rate could be 3 to 10 time as normal years. These two glaciers forwarded their terminus during their surging period by Landsat optical monitoring. We concluded that during 1990s and 2000s most glaciers kept a stable flow velocity and some glacier surged during the study period therefore terminus forwarding cannot equal to mass gaining.

Rocky desertification is a type of land desertification. It comes from the fragile ecological and geological environment, where the human activity is very strong and the land productivity is degraded severely. As a natural disaster, rocky desertification is very destructive, and it is very difficult to be recovered. The karst region of China in southwest is the world's concentrated karsts region. It is also one of the largest contiguous karsts regions. The karst region is also the most typical ecological fragile regions in China. We utilized the ETM+ images in 2000 to study the rocky desertification of the north regions in Guangxi province in the past ten years. Firstly, the rocky exponential model was established to extract rocky desertification information of the region. Then, the RGB image was composited to interpret and obtain the rocky desertification. Our experiment showed that rocky desertification of the karst region can be classified into no rocky desertification, moderate desertification, and severe rocky desertification.

Synthetic Aperture Radar (SAR) data has been reported to be able to provide complementary information towards optical remote sensing data for improving the urban impervious surface estimation. However, most existing researches were focused on using only single polarization SAR data. This study presents a preliminary experiment on the combined use of multispectral optical data and dual polarization SAR data for impervious surfaces estimation. Experimental results using SPOT-5 and ALOS PALSAR images showed a consistent result compared with our previous result using single polarization SAR data. Comparison results showed that not every polarimetric feature was able to provide positive effect to the impervious surfaces estimation. Compared with using only optical and SAR data, the separated HH and HV polarization data provided a positive effect to the result by improving the accuracy. The incorporation of both Entropy and Alpha features were also able to improve the accuracy. However, the HH/HV ratio and the separated use of Entropy did not provide positive results. A combination of all the features turned out to obtain the highest accuracy compared with using a subset of the features.

An improved reconstruction algorithm is proposed for compact polarimetric (CP) synthetic aperture radar (SAR) on oil spill detection. Based on the differences in statistical behavior between open and oil-covered sea surfaces, the proposed algorithm can iteratively reconstruct quad-pol SAR images from CP SAR data. During the experiment, it outperformed two existing compact SAR reconstruction algorithms in terms of both statistical and information theoretical analysis.

Accurate land covers classification is challenging in urban areas due to the diversity of urban land covers. This study presents a classification strategy with combined optical and Synthetic Aperture Radar (SAR) images using Random Forest (RF). Optimization of RF is conducted, indicating the optimal number of decision trees is 10 and the optimal number of features is 4 for splitting each tree node. The overall accuracy (OA) and Kappa coefficient are used to assess the classification. Result shows that classification with combined optical and SAR images (OA: 69.08%; Kappa: 0.6288) is higher than that with single optical image (OA: 81.43%; Kappa: 0.7770). Benefits of the combined use of optical and SAR images mainly come from reducing the confusions between water and shade, and between bare soil and dark impervious surfaces.