Knowledge of the precise water depth in shallow areas of the ocean is of great significance to the safe navigation of ships and hydrographic surveying. Compared with traditional bathymetry, satellite remote sensing for water depth determination makes it possible to cover large areas by dynamic observation. In this paper, we conducted an optically shallow water bathymetric inversion study using a Stumpf empirical model, random forest model, neural network model, and support vector machine model based on Sentinel-2 satellite images and Ganquan Dao measured bathymetry data. We compared and analyzed the inversion results based on the empirical model and different machine learning models. The results show that the Stumpf empirical and machine learning models are capable of inverting optically shallow water depth. Moreover, the machine learning models had better fitting ability than the Stumpf empirical model with a sufficient number of samples, especially when the water depth was greater than 15 m. In addition, the random forest model had the highest overall accuracy among these models, with a root mean square error (RMSE) of 1.41 m and a regression coefficient (R²) of 0.96 for the test data.

Building function type is an important parameter for urban planning and disaster management. However, existing identification methods do not always correctly recognize all building functions because of missing point of interest (POI) data in private areas. In this study, we proposed a hierarchical data-mining model to identify building function types using accessible auxiliary data, which was then applied to a case study. Residential building property was assessed to address missing residential POIs. The building functions were assigned to one of five different types, or a mixed-function type. Standard deviation and mean values extracted from remotely sensed images, distances to major roads, and building shape parameters were used to infer the function types of buildings without assigned function types. The proposed model was able to identify 65% of buildings not previously assigned as residential through the POI, with an overall accuracy of 87%. In addition, all buildings were successfully assigned a function type of residential, commercial, office, warehouse, public service, or mixed-function, with an overall accuracy of 85% for unclassified buildings. Our results demonstrated that missing POI data in private areas could be addressed by integration with multisource data using a simple method.

Land use/cover change (LUCC) is one of the major factors influencing the storage of ecosystem carbon. The carbon storage in Qinghai-Tibet Plateau, the world’s highest plateau, is affected by a combination of many factors. Using MCD12Q1 land classification data, aboveground biomass, belowground biomass, soil carbon and humus carbon data, as well as field sampling data for parameters verification, we applied the InVEST model to simulate the ecosystem carbon storage and the impacts of driving factors. The field survey samples were used to test the regression accuracy, and the results confirmed that the model performance was reasonable and acceptable. The main conclusions of this study are as follows: From 2001 to 2010, carbon storage in the Qinghai-Tibet Plateau increased by 10.39 billion t when assuming that the carbon density in each land cover type was constant. Changes of the land cover types caused carbon storage to increase by 116 million t, which contributed 13.82% of the dynamic carbon storage. Consequently, changes in carbon density accounted for 86.18% of the carbon storage change. In addition, we investigated the soil organic matter and aboveground biomass characteristics between 2012 and 2014 and found that the influences of fencing and dung on carbon storage were positive.

This study explored the model of urban impervious surface (IS) density, land surface temperature (LST), and comprehensive ecological evaluation index (CEEI) from urban centers to suburbs. The interrelationships between these parameters in Guangzhou from 1987 to 2019 were analyzed using time-series Landsat-5 TM (Thematic Mapper), Landsat-8 OLI (Operational Land Imager), and TIRS (Thermal Infrared Sensor) images. The urban IS densities were calculated in concentric rings using time-series IS fractions, which were used to construct an inverse S-shaped urban IS density function to depict changes in urban form and the spatio-temporal dynamics of urban expansion from the urban center to the suburbs. The results indicated that Guangzhou experienced expansive urban growth, with the patterns of urban spatial structure changing from a single-center to a multi-center structure over the past 32 years. Next, the normalized LST and CEEI in each concentric ring were calculated, and their variation trends from the urban center to the suburbs were modeled using linear and nonlinear functions, respectively. The results showed that the normalized LST had a gradual decreasing trend from the urban center to the suburbs, while the CEEI showed a significant increasing trend. During the 32-year rapid urban development, the normalized LST difference between the urban center and suburbs increased gradually with time, and the CEEI significantly decreased. This indicated that rapid urbanization significantly expanded the impervious surface areas in Guangzhou, leading to an increase in the LST difference between urban centers and suburbs and a deterioration in ecological quality. Finally, the potential interrelationships among urban IS density, normalized LST, and CEEI were also explored using different models. This study revealed that rapid urbanization has produced geographical convergence between several ISs, which may increase the risk of the urban heat island effect and degradation of ecological quality.

The capsule network (Caps) is a novel type of neural network that has great potential for the classification of hyperspectral remote sensing. However, the Caps suffers from the issue of gradient vanishing. To solve this problem, a powered activation regularization based adaptive capsule network (PAR-ACaps) was proposed for hyperspectral remote sensing classification, in which an adaptive routing algorithm without iteration was applied to amplify the gradient, and the powered activation regularization method was used to learn the sparser and more discriminative representation. The classification performance of PAR-ACaps was evaluated using two public hyperspectral remote sensing datasets, i.e., the Pavia University (PU) and Salinas (SA) datasets. The average overall classification accuracy (OA) of PAR-ACaps with shallower architecture was measured and compared with those of the benchmarks, including random forest (RF), support vector machine (SVM), 1-dimensional convolutional neural network (1DCNN), two-dimensional convolutional neural network (CNN), three-dimensional convolutional neural network (3DCNN), Caps, and the original adaptive capsule network (ACaps) with comparable network architectures. The OA of PAR-ACaps for PU and SA datasets was 99.51% and 94.52%, respectively, which was higher than those of benchmarks. Moreover, the classification performance of PAR-ACaps with relatively deeper neural architecture (four and six convolutional layers in the feature extraction stage) was also evaluated to demonstrate the effectiveness of gradient amplification. As shown in the experimental results, the classification performance of PAR-ACaps with relatively deeper neural architecture for PU and SA datasets was also superior to 1DCNN, CNN, 3DCNN, Caps, and ACaps with comparable neural architectures. Additionally, the training time consumed by PAR-ACaps was significantly lower than that of Caps. The proposed PAR-ACaps is, therefore, recommended as an effective alternative for hyperspectral remote sensing classification.