Automated methods to extract buildings from very high resolution (VHR) remote sensing data have many applications in a wide range of fields. Many convolutional neural network (CNN) based methods have been proposed and have achieved significant advances in the building extraction task. In order to refine predictions, a lot of recent approaches fuse features from earlier layers of CNNs to introduce abundant spatial information, which is known as skip connection. However, this strategy of reusing earlier features directly without processing could reduce the performance of the network. To address this problem, we propose a novel fully convolutional network (FCN) that adopts attention based re-weighting to extract buildings from aerial imagery. Specifically, we consider the semantic gap between features from different stages and leverage the attention mechanism to bridge the gap prior to the fusion of features. The inferred attention weights along spatial and channel-wise dimensions make the low level feature maps adaptive to high level feature maps in a target-oriented manner. Experimental results on three publicly available aerial imagery datasets show that the proposed model (RFA-UNet) achieves comparable and improved performance compared to other state-of-the-art models for building extraction.

Coastal aquaculture plays an important role in the provision of seafood, the sustainable development of regional and global economy, and the protection of coastal ecosystems. Inappropriate planning of disordered and intensive coastal aquaculture may cause serious environmental problems and socioeconomic losses. Precise delineation and classification of different kinds of aquaculture areas are vital for coastal management. It is difficult to extract coastal aquaculture areas using the conventional spectrum, shape, or texture information. Here, we proposed an object-based method combining multi-scale segmentation and object-based neighbor features to delineate existing coastal aquaculture areas. We adopted the multi-scale segmentation to generate semantically meaningful image objects for different land cover classes, and then utilized the object-based neighbor features for classification. Our results show that the proposed approach effectively identified different types of coastal aquaculture areas, with 96% overall accuracy. It also performed much better than other conventional methods (e.g., single-scale based classification with conventional features) with higher classification accuracy. Our results also suggest that the multi-scale segmentation and neighbor features can obviously improve the classification performance for the extraction of cage culture areas and raft culture areas, respectively. Our developed approach lays a solid foundation for intelligent monitoring and management of coastal ecosystems.

Marine aquaculture plays an important role in seafood supplement, economic development, and coastal ecosystem service provision. The precise delineation of marine aquaculture areas from high spatial resolution (HSR) imagery is vital for the sustainable development and management of coastal marine resources. However, various sizes and detailed structures of marine objects make it difficult for accurate mapping from HSR images by using conventional methods. Therefore, this study attempts to extract marine aquaculture areas by using an automatic labeling method based on the convolutional neural network (CNN), i.e., an end-to-end hierarchical cascade network (HCNet). Specifically, for marine objects of various sizes, we propose to improve the classification performance by utilizing multi-scale contextual information. Technically, based on the output of a CNN encoder, we employ atrous convolutions to capture multi-scale contextual information and aggregate them in a hierarchical cascade way. Meanwhile, for marine objects with detailed structures, we propose to refine the detailed information gradually by using a series of long-span connections with fine resolution features from the shallow layers. In addition, to decrease the semantic gaps between features in different levels, we propose to refine the feature space (i.e., channel and spatial dimensions) using an attention-based module. Experimental results show that our proposed HCNet can effectively identify and distinguish different kinds of marine aquaculture, with 98% of overall accuracy. It also achieves better classification performance compared with object-based support vector machine and state-of-the-art CNN-based methods, such as FCN-32s, U-Net, and DeeplabV2. Our developed method lays a solid foundation for the intelligent monitoring and management of coastal marine resources.

The scientific evaluation of landscape fragmentation and connectivity is important for habitat conservation. It is strongly influenced by the spatial resolution of source maps, particularly in urban environments. However, there is limited comprehensive investigation of the spatial grain effect on urban habitat and few in-depth analysis across different urban gradients. In this paper, we scrutinize the spatial grain effects of urban green space (UGS) cover maps (derived from remote sensing imagery and survey data) with respect to evaluating habitat fragmentation and connectivity, comparing among different urban gradient scenarios (downtown, urban periphery, and suburban area) in Hangzhou, a megacity in China. The fragmentation was detected from three indices, including Entropy, Contagion, and Hypsometry. Then morphological spatial pattern analysis (MSPA) was applied for the landscape element identification. The possibility of connectivity (PC) and patch importance (dPC) were proposed for measuring the landscape connectivity based on Cores and Bridges from MSPA results. The results indicate that the farther the location is from downtown, the less sensitive the landscape element proportion to the spatial resolution. Among the three fragmentation indices, the overall hypsometry index has the lowest sensitivity to the spatial resolution, which implies this index’s broader application value. Considering connectivity, high spatial resolution maps are appropriate for analyzing highly heterogeneous urban areas, while medium spatial resolution maps are more applicable to urban periphery and suburban area with larger UGS patches and less fragmentation. This study suggests that the spatial resolution of UGS maps substantially influence habitat fragmentation and connectivity, which is critical for decision making in urban planning and management.

Cultivated land in China has faced severe challenges in recent years due to rapid urbanization. In 1997, the “cultivated land requisition-compensation balance” policy was implemented by the government to maintain the quantity and quality of cultivated land. Previous studies mainly focused on the quantity changes of cultivated land. In this study, from a quality perspective, we characterized the occupation and compensation of cultivated land in Wenzhou City utilizing remote sensing and geographic information systems during 2005–2010 and 2010–2014. Our results indicated that although the quantity balance of cultivated land has been achieved in Wenzhou, there was a trend of consuming prime cultivated land for urbanization while compensating with less productive land. It was also found that topography, water resources, and accessibility play important roles in cultivated land changes, with urbanization occurring on the eastern coastal plain where high quality lands are prevalent. Less than 60% of the gained lands were under cultivation, with the majority of reclaimed land from forests and coastal areas and located in remote regions. Therefore, we suggest that a “cultivated land protection red line” policy should be implemented to protect the best cultivated lands, while preventing random land reclamation to secure agricultural and environmental sustainability.

Light intensity primarily drives plant growth and morphogenesis, whereas the ecological impact of light intensity on the phyllosphere (leaf surface and endosphere) microbiome is poorly understood. In this study, garden lettuce (Lactuca sativa L.) plants were grown under low, medium, and high light intensities. High light intensity remarkably induced the leaf contents of soluble proteins and chlorophylls, whereas it reduced the contents of leaf nitrate. In comparison, medium light intensity exhibited the highest contents of soluble sugar, cellulose, and free amino acids. Meanwhile, light intensity resulted in significant changes in the composition of functional genes but not in the taxonomic compositions of the prokaryotic community (bacteria and archaea) in the phyllosphere. Notably, garden lettuce plants under high light intensity treatment harbored more sulfur-cycling mdh and carbon-cycling glyA genes than under low light intensity, both of which were among the 20 most abundant prokaryotic genes in the leaf phyllosphere. Furthermore, the correlations between prokaryotic functional genes and lettuce leaf metabolite groups were examined to disclose their interactions under varying light intensities. The relative abundance of the mdh gene was positively correlated with leaf total chlorophyll content but negatively correlated with leaf nitrate content. In comparison, the relative abundance of the glyA gene was positively correlated with leaf total chlorophyll and carotenoids. Overall, this study revealed that the functional composition of the phyllosphere prokaryotic community and leaf metabolite groups were tightly linked in response to changing light intensities. These findings provided novel insights into the interactions between plants and prokaryotic microbes in indoor farming systems, which will help optimize environmental management in indoor farms and harness beneficial plant–microbe relationships for crop production.