Deep learning-based hyperspectral image (HSI) classification has attracted more and more attention because of its excellent classification ability. Generally, the outstanding performance of these methods mainly depends on a large number of labeled samples. Therefore, it still remains an ongoing challenge how to integrate spatial structure information into these frameworks to classify the HSI with limited training samples. In this study, an effective spectral-spatial HSI classification scheme is proposed based on superpixel pooling convolutional neural network with transfer learning (SP-CNN). The suggested method includes three stages. The first part consists of convolution and pooling operation, which is a down-sampling process to extract the main spectral features of an HSI. The second part is composed of up-sampling and superpixel (homogeneous regions with adaptive shape and size) pooling to explore the spatial structure information of an HSI. Finally, the hyperspectral data with each superpixel as a basic input rather than a pixel are fed to fully connected neural network. In this method, the spectral and spatial information is effectively fused by using superpixel pooling technique. The use of popular transfer learning technology in the proposed classification framework significantly improves the training efficiency of SP-CNN. To evaluate the effectiveness of the SP-CNN, extensive experiments were conducted on three common real HSI datasets acquired from different sensors. With 30 labeled pixels per class, the overall classification accuracy provided by this method on three benchmarks all exceeded 93%, which was at least 4.55% higher than that of several state-of-the-art approaches. Experimental and comparative results prove that the proposed algorithm can effectively classify the HSI with limited training labels.

Due to the differences in land cover and natural surroundings within cities, residents in various regions face different thermal risks. Therefore, this study combined multi-source data to analyze the relationship between urban heat risk and local climate zones (LCZ). We found that in downtown Shenyang, the building-type LCZ was mainly found in urban centers, while the natural- type LCZ was mainly found in suburbs. Heat risk was highest in urban centers, gradually decreasing along the suburban direction. The thermal risk indices of the building-type LCZs were significantly higher than those of the natural types. Among the building types of LCZs, LCZ 8 (open middle high-rise) had the highest average thermal risk index (0.48), followed by LCZ 3 (0.46). Among the natural types of LCZs, LCZ E (bare rock and paved) and LCZ F (bare soil and sand) had the highest thermal risk indices, reaching 0.31 and 0.29, respectively. This study evaluated the thermal risk of the Shenyang central urban area from the perspective of LCZs and combined it with high-resolution remote sensing data to provide a reference for thermal risk mitigation in future urban planning.

Several studies have demonstrated that the urban heat island (UHI) impact is significantly influenced by changes in urban land cover (LC). In this study, we examined the dynamic change in LC in Shijiazhuang City and its impact on UHI intensity using multi-source data, including Landsat and LC data. The results showed that (1) the 15-year span exhibited a substantial growth in built-up area from 14.541% to 19.718%, whereas cropland and grassland dropped by 3.342% and 3.625%, respectively. (2) The mean land surface temperature (LST) in the study area increased significantly. From a point-to-plane distribution, the high-temperature zone (>32°C) grew, and the direction of growth was consistent with urbanization. (3) The average temperature of the barren areas was the highest, followed by the built-up areas. (4) The normalized difference vegetation index (NDVI) and LST have a strong negative association, showing that LST rises with vegetation loss. Our research results provide a scientific basis for deeper understanding of the mechanisms underlying UHI creation, as well as for developing logical land use strategies and evaluating the effects of urbanization on local climates.

Population is one of the key problematic factors that are restricting China’s economic and social development. Previous studies have used nighttime light (NTL) imagery to calculate population density. This study analyzes the spatiotemporal evolution of the population in Northeast China based on linear regression analyses of NPP-VIIRS NTL imagery and statistical population data from 36 cities in Northeast China from 2012 to 2017. Based on a comparison of the estimation results in different years, we observed the following. (1) The population of Northeast China showed an overall decreasing trend from 2012–2017, with population changes of +31,600, −960,800, −359,800, −188,000, and −1,127,600 in the respective years. (2) With the overall population loss trend in Northeast China, the population increased in only three cities, namely, Shenyang, Dalian, and Panjin, with an average increase during the six-year period of 24,200, 6,500, and 2,000 people, respectively. (3) The four major urban agglomerations in Northeast China (the Harbin-Daqing-Qiqihar Industrial Corridor, Changjitu Pilot Zone, Liaoning Coastal Economic Belt, and Shenyang Economic Zone) have annual populations far exceeding 4 million people. A correct appreciation of the population dynamics is vital to resource management and comprehensive management efforts. Making full use of natural resources and regional advantages could effectively improve and potentially solve the urban population loss problem and would be of great innovative significance for supporting the realization of the Millennium Development Goals.

Although superpixel segmentation provides a powerful tool for hyperspectral image (HSI) classification, it is still a challenging problem to classify an HSI at superpixel level because of the characteristics of adaptive size and shape of superpixels. Furthermore, these characteristics of superpixels along with the appearance of noisy pixels makes it difficult to appropriately measure the similarity between two superpixels. Under the assumption that pixels within a superpixel belong to the same class with a high probability, this paper proposes a novel spectral–spatial HSI classification method at superpixel level (SSC-SL). Firstly, a simple linear iterative clustering (SLIC) algorithm is improved by introducing a new similarity and a ranking technique. The improved SLIC, specifically designed for HSI, can straightly segment HSI with arbitrary dimensionality into superpixels, without consulting principal component analysis beforehand. In addition, a superpixel-to-superpixel similarity is newly introduced. The defined similarity is independent of the shape of superpixel, and the influence of noisy pixels on the similarity is weakened. Finally, the classification task is accomplished by labeling each unlabeled superpixel according to the nearest labeled superpixel. In the proposed superpixel-level classification scheme, each superpixel is regarded as a sample. This obviously greatly reduces the data volume to be classified. The experimental results on three real hyperspectral datasets demonstrate the superiority of the proposed spectral–spatial classification method over several comparative state-of-the-art classification approaches, in terms of classification accuracy.