Urban building segmentation is a prevalent research domain for very high resolution (VHR) remote sensing; however, various appearances and complicated background of VHR remote sensing imagery make accurate semantic segmentation of urban buildings a challenge in relevant applications. Following the basic architecture of U-Net, an end-to-end deep convolutional neural network (denoted as DeepResUnet) was proposed, which can effectively perform urban building segmentation at pixel scale from VHR imagery and generate accurate segmentation results. The method contains two sub-networks: One is a cascade down-sampling network for extracting feature maps of buildings from the VHR image, and the other is an up-sampling network for reconstructing those extracted feature maps back to the same size of the input VHR image. The deep residual learning approach was adopted to facilitate training in order to alleviate the degradation problem that often occurred in the model training process. The proposed DeepResUnet was tested with aerial images with a spatial resolution of 0.075 m and was compared in performance under the exact same conditions with six other state-of-the-art networks—FCN-8s, SegNet, DeconvNet, U-Net, ResUNet and DeepUNet. Results of extensive experiments indicated that the proposed DeepResUnet outperformed the other six existing networks in semantic segmentation of urban buildings in terms of visual and quantitative evaluation, especially in labeling irregular-shape and small-size buildings with higher accuracy and entirety. Compared with the U-Net, the F1 score, Kappa coefficient and overall accuracy of DeepResUnet were improved by 3.52%, 4.67% and 1.72%, respectively. Moreover, the proposed DeepResUnet required much fewer parameters than the U-Net, highlighting its significant improvement among U-Net applications. Nevertheless, the inference time of DeepResUnet is slightly longer than that of the U-Net, which is subject to further improvement.

Accurate landslide detection and mapping are essential for land use planning, management/assessment, and geo-disaster risk mitigation as well as post disaster reconstructions. Till now, visual interpretation and field survey are still the most widely adopted techniques for landslide mapping, which are often criticized labor-intensive, time consuming and costly. With the rapid advancement of artificial intelligence, deep-learning-based approach for landslide detection and mapping has drawn great attention for its significant advantages over the traditional techniques. However, lack of sufficient training samples has constrained the application of deep-learning-based approach in landslide detection from satellite images for a long time. The present study aimed to examine the feasibility of a new deep-learning-based approach to intelligently detect and map earthquake-triggered landslides from single-temporal RapidEye satellite images. Specifically, the proposed approach consists of three steps. First of all, a standard data pre-processing workflow to automatically generate training samples was designed and some data augmentation strategies were implemented to alleviate the lack of training samples. Then, a cascaded end-to-end deep learning network, namely LandsNet, was constructed to learn various features of landslides. Finally, the identified landslide maps were further optimized with morphological processing. Experiments in two spatially independent earthquake-affected regions showed our proposed approach yielded the best F1 value of about 86.89%, which was about 7% and 8% higher than that obtained by ResUNet and DeepUNet, respectively. Comparative studies on the feasibility and robustness of the proposed approach with ResUNet and DeepUNet demonstrated its strong application potentials in the emergency response of natural disasters.

Influences of the increasing pressure of climate change and anthropogenic activities on wetlands ecosystems and agriculture are significant around the world. This paper assessed the spatiotemporal land use and land cover changes (LULCC), especially for conversion from marshland to other LULC types (e.g., croplands) over the Songnen and Sanjiang Plain (SNP and SJP), northeast China, during the past 35 years (1980–2015). The relative role of human activities and climatic changes in terms of their impacts on wetlands and agriculture dynamics were quantitatively distinguished and evaluated in different periods based on a seven-stage LULC dataset. Our results indicated that human activities, such as population expansion and socioeconomic development, and institutional policies related to wetlands and agriculture were the main driving forces for LULCC of the SJP and SNP during the past decades, while increasing contributions of climatic changes were also found. Furthermore, as few studies have identified which geographic regions are most at risk, how the future climate changes will spatially and temporally impact wetlands and agriculture, i.e., the suitability of wetlands and agriculture distributions under different future climate change scenarios, were predicted and analyzed using a habitat distribution model (Maxent) at the pixel-scale. The present findings can provide valuable references for policy makers on regional sustainability for food security, water resource rational management, agricultural planning and wetland protection as well as restoration of the region.

Owning to the nature of flood events, near-real-time flood detection and mapping is essential for disaster prevention, relief, and mitigation. In recent years, the rapid advancement of deep learning has brought endless possibilities to the field of flood detection. However, deep learning relies heavily on training samples and the availability of high-quality flood datasets is rather limited. The present study collected 16 flood events in the Yangtze River Basin and divided them into three categories for different purpose: training, testing, and application. An efficient methodology of dataset-generation for training, testing, and application was proposed. Eight flood events were used to generate strong label datasets with 5296 tiles as flood training samples along with two testing datasets. The performances of several classic convolutional neural network models were evaluated with those obtained datasets, and the results suggested that the efficiencies and accuracies of convolutional neural network models were obviously higher than that of the threshold method. The effects of VH polarization, VV polarization, and the involvement of auxiliary DEM on flood detection were investigated, which indicated that VH polarization was more conducive to flood detection, while the involvement of DEM has a limited effect on flood detection in the Yangtze River Basin. Convolutional neural network trained by strong datasets were used in near-real-time flood detection and mapping for the remaining eight flood events, and weak label datasets were generated to expand the flood training samples to evaluate the possible effects on deep learning models in terms of flood detection and mapping. The experiments obtained conclusions consistent with those previously made on experiments with strong datasets.