The rapid development in deep learning and computer vision has introduced new opportunities and paradigms for building extraction from remote sensing images. In this paper, we propose a novel fully convolutional network (FCN), in which a spatial residual inception (SRI) module is proposed to capture and aggregate multi-scale contexts for semantic understanding by successively fusing multi-level features. The proposed SRI-Net is capable of accurately detecting large buildings that might be easily omitted while retaining global morphological characteristics and local details. On the other hand, to improve computational efficiency, depthwise separable convolutions and convolution factorization are introduced to significantly decrease the number of model parameters. The proposed model is evaluated on the Inria Aerial Image Labeling Dataset and the Wuhan University (WHU) Aerial Building Dataset. The experimental results show that the proposed methods exhibit significant improvements compared with several state-of-the-art FCNs, including SegNet, U-Net, RefineNet, and DeepLab v3+. The proposed model shows promising potential for building detection from remote sensing images on a large scale.

The rapid development of information technology and location techniques not only leads to an increasing growth of massive geospatial big data but also raises the attention of using these data to complement with remote sensing images. Many efforts have been made to utilize geospatial big data to identify human activity patterns and carry out urban and environmental researches, integrating with remote sensing images. Nonetheless, there are still many issues, including the representativeness and locality of geospatial big data, as well as the fusion methods, remain to be further explored. In this article, we first reviewed the innovation and proceedings of data mining and analyzing techniques, as well as remote sensing applications driven by geospatial big data. Besides, two popular concepts, namely, “Social Sensing” and “Urban Computing,” were briefly introduced. Then, we highlighted the role of geospatial big data in mining human activity dynamics and socioeconomic characteristics, and the feasibility of combining with remote sensing data for various studies. Lastly, we presented some empirical case studies on the confluence of remote sensing and geospatial big data in land use extraction, environmental and disaster monitoring, as well as socioeconomic dynamics sensing. The provided examples and discussion demonstrated the high efficiency and complementarity of the integration of remote sensing and geospatial big data, which benefits decision making from multiple perspectives and scales.

Water footprint (WF) is a measure of the actual appropriation of water resources. WF accounting can provide a scientific basis for the managements of water resources. In this study, a multi-regional input-output model is employed to measure the quantity of blue WF (WF) and inter-provincial virtual water (VW) flows in China for the years of 2007 and 2010. The results show that: (1) China’s total WF increased from 205.42 billion m³ in 2007 to 229.34 billion m³ in 2010. Approximately 42% of the WF was attributed to VW embodied in inter-provincial trade. Xinjiang is the largest province of VW export, whereas Shanghai had the largest net VW inflows. (2) From 2007 to 2010, the share of the agricultural sector in the entire VW trade declined, but was still as high as 82.78%, followed by the industrial sector. (3) The north-to-south and south-to-south patterns were witnessed in the domestic VW flows. The provincial WF variations are found to be affected by the per capital GDP, total water resources, per capita water resources, and urban population. (4) By linking VW with an integrated WAVE+ (water accounting and vulnerability evaluation) factor, it was found that virtual scarce water (VSW) was mainly exported by the provinces in northern China. At the national level, the amounts of VSW inflows were consistently greater than those of VSW outflows for both years, 2007 and 2010, implying an increased pressure on the provinces with water deprivation issues. Overall, these results can provide a basis for refining the spatiotemporal allocation of water resources and mitigating the conflict between water supply and demand in China.

China has been the largest carbon emitter in the world since 2007 and is thus confronted with huge emission reduction pressures. The regional differences in socio-economic development lead to complex inter-provincial carbon transfer in China, which hinders the determination of the emission reduction responsibilities for the various provinces. Based on the latest multi-regional input-output data, this study analyzes the carbon footprint, inter-provincial carbon transfer, and the corresponding variations of 30 provinces in China from 2007 to 2010. The results show that the domestic carbon footprint increased from 4578 Mt in 2007 to 6252 Mt in 2010. Provinces with high carbon footprints were mainly found in central China, such as Shandong, Jiangsu, and Henan. Carbon footprints of the developed coastal provinces were greater than those of less developed provinces in Northwestern China. Per capita GDP (Gross Domestic Product) was positively correlated to the per capita carbon footprint, indicating a positive relationship between the economic development level and corresponding carbon emissions. Provincial carbon inflows were found to have increased steadily (ranging between 32% and 41%) from 2007 to 2010. The increases in direct carbon emissions varied largely among different provinces, ranging from below 30% in the developed provinces to more than 60% in the moderately developed provinces (e.g., Sichuan and Chongqing). The embodied carbon transferred from moderately developed or remote provinces to those developed ones. In other words, the carbon emission pressures of the developed provinces were shifted to the less developed provinces. The major paths of carbon flow include the transfers from Hebei to Jiangsu (32.07 Mt), Hebei to Beijing (26.78 Mt), Hebei to Zhejiang (25.60 Mt), and Liaoning to Jilin (27.60 Mt).