Urbanization in China is progressing rapidly and continuously, especially in the newly developed metropolitan areas. The Google Earth Engine (GEE) is a powerful tool that can be used to efficiently investigate these changes using a large repository of available optical imagery. This work examined land-cover changes in the central region of the lower Yangtze River and exemplifies the application of GEE using the random forest classification algorithm on Landsat dense stacks spanning the 30 years from 1987 to 2017. Based on the obtained time-series land-cover classification results, the spatiotemporal land-use/cover changes were analyzed, as well as the main factors driving the changes in different land-cover categories. The results show that: (1) The obtained land datasets were reliable and highly accurate, with an overall accuracy ranging from 88% to 92%. (2) Over the past 30 years, built-up areas have continued to expand, increasing from 537.9 km² to 1500.5 km², and the total area occupied by built-up regions has expanded by 178.9% to occupy an additional 962.7 km². The surface water area first decreased, then increased, and generally showed an increasing trend, expanding by 17.9%, with an area increase of approximately 131 km². Barren areas accounted for 6.6% of the total area in the period 2015–2017, which was an increase of 94.8% relative to the period 1987–1989. The expansion of the built-up area was accompanied by an overall 25.6% (1305.7 km²) reduction in vegetation. (3) The complexity of the key factors driving the changes in the regional surface water extent was made apparent, mainly including the changes in runoff of the Yangtze River and the construction of various water conservancy projects. The effects of increasing the urban population and expanding industrial development were the main factors driving the expansion of urban built-up areas and the significant reduction in vegetation. The advantages and limitations arising from land-cover mapping by using the Google Earth Engine are also discussed.

Railway turnout system is a key infrastructure to railway safety and efficiency. However, it is prone to failure in the field. Therefore, many railway departments have adopted a monitoring system to monitor the operation status of turnouts. With monitoring data collected, many researchers have proposed different fault-diagnosis methods. However, many of the existing methods cannot realize real-time updating or deal with new fault types. This paper—based on imbalanced data—proposes a Bayes-based online turnout fault-diagnosis method, which realizes incremental learning and scalable fault recognition. First, the basic conceptions of the turnout system are introduced. Next, the feature extraction and processing of the imbalanced monitoring data are introduced. Then, an online diagnosis method based on Bayesian incremental learning and scalable fault recognition is proposed, followed by the experiment with filed data from Guangzhou Railway. The results show that the scalable fault-recognition method can reach an accuracy of 99.11%, and the training time of the Bayesian incremental learning model reduces 29.97% without decreasing the accuracy, which demonstrates the high accuracy, adaptability and efficiency of the proposed model, of great significance for labor-saving, timely maintenance and further, safety and efficiency of railway transportation.

In this article, a novel superresolution mapping (SRM) method based on pixel-, subpixel-, and superpixel-scale spatial dependence (PSSSD) with pansharpening technique is proposed. First, an original coarse resolution remote sensing image and a high-resolution panchromatic image are fused by the pansharpening technique to produce a pansharpened result. A segmentation image with numerous superpixels, which represent irregular objects, is generated by adaptively segmenting the pansharpened result. Second, the class proportions of pixels, subpixels, and superpixels are, respectively, obtained from the original image, the pansharpened image, and the segmentation image. Pixel-scale spatial dependence is obtained using a pixel spatial attraction model. Subpixel-scale spatial dependence is derived using a novel subpixel spatial attraction model. Superpixel-scale spatial dependence is obtained through an extended random walker algorithm. Third, the three-scale spatial dependence is integrated into the PSSSD. Finally, a class allocation method based on object units is used to obtain the SRM results according to the PSSSD. Experimental results for three remote sensing images show that the proposed PSSSD outperforms the existing state-of-the-art SRM methods.