A timely and accurate crop type mapping is very significant, and a prerequisite for agricultural regions and ensuring global food security. The combination of remotely sensed optical and radar datasets presents an opportunity for acquiring crop information at relative spatial resolution and temporal resolution adequately to capture the growth profiles of various crop species. In this paper, we employed Sentinel-1A (S-1) and Sentinel-2A (S-2) data acquired between the end of June and early September 2016, on a semi-arid area in northern Nigeria. A different set of (VV and VH) SAR and optical (SI and SB) images, illustrating crop phenological development stage, were employed as inputs to the two machines learning Random Forest (RF) and Support Vector Machine (SVM) algorithms to automatically map maize fields. Significant increases in overall classification were shown when the multi-temporal spectral indices (SI) and spectral band (SB) datasets were added with the different integration of SAR datasets (i.e., VV and VH). The best overall accuracy (OA) for maize (96.93%) was derived by using RF classification algorithms with SI-SB-SAR datasets, although the SI datasets for RF and SB datasets for SVM also produced high overall maize classification accuracies, of 97.04% and 97.44%. The outcomes indicate the robustness of the RF or SVM methods to produce high-resolution maps of maize for subsequent application from agronomists, policy planners, and the government, because such information is lacking in our study area.

Administering an urban boundary (UB) is increasingly important for curbing disorderly urban land expansion. The traditionally manual digitalization is time-consuming, and it is difficult to connect UB in the urban fringe due to the fragmented urban pattern in daytime data. Nighttime light (NTL) data is a powerful tool used to map the urban extent, but both the blooming effect and the coarse spatial resolution make the urban product unable to meet the requirements of high-precision urban study. In this study, precise UB is extracted by a practical and effective method using NTL data and Landsat 8 data. Hangzhou, a megacity experiencing rapid urban sprawl, was selected to test the proposed method. Firstly, the rough UB was identified by the search mode of the concentric zones model (CZM) and the variance-based approach. Secondly, a buffer area was constructed to encompass the precise UB that is near the rough UB within a certain distance. Finally, the edge detection method was adopted to obtain the precise UB with a spatial resolution of 30 m. The experimental results show that a good performance was achieved and that it solved the largest disadvantage of the NTL data-blooming effect. The findings indicated that cities with a similar level of socio-economic status can be processed together when applied to larger-scale applications.

Forest tree species information plays an important role in ecology and forest management, and deep learning has been used widely for remote sensing image classification in recent years. However, forest tree species classification using remote sensing images is still a difficult task. Since there is no benchmark dataset for forest tree species, a forest tree species dataset (FTSD) was built in this paper to fill the gap based on the Sentinel-2 images. The FTSD contained nine kinds of forest tree species in Qingyuan County with 8,815 images, each with a resolution of 64 × 64 pixels. The images were produced by combining forest management inventory data and Sentinel-2 images, which were acquired with less than 20% clouds from 1 April to 31 October, including the years 2017, 2018, 2019, 2020, and 2021. Then, the images were preprocessed and downloaded from Google Earth Engine (GEE). Four different band combinations were compared in the paper. Moreover, a Principal Component Analysis (PCA) and Normalized Difference Vegetation Index (NDVI) were also calculated using the GEE. Deep learning algorithms including DenseNet, EfficientNet, MobileNet, ResNet, and ShuffleNet were trained and validated in the FTSD. RGB images with red, green, and blue (PC1, PC2, and NDVI) obtained the highest validation accuracy in four band combinations. ResNet obtained the highest validation accuracy in all algorithms after 500 epochs were trained in the FTSD, which reached 84.91%. As a famous and widely used remote sensing classification satellite imagery dataset, NWPU RESISC-45 was also trained and validated in the paper. ResNet achieved a high validation accuracy of 87.90% after training 100 epochs in NWPU RESISC-45. The paper shows in forest tree species classification based on remote sensing images and deep learning that (1) PCA and NDVI can be combined to improve the accuracy of classification; (2) ResNet is more suitable than other deep learning algorithms including DenseNet, EfficientNet, MobileNet, and ShuffleNet in remote sensing classification; and (3) being too shallow or deep in ResNet does not perform better in the FTSD, that is, 50 layers are better than 34 and 101 layers.

Address matching, which aims to match an input descriptive address with a standard address in an address database, is a key technology for achieving data spatialization. The construction of today’s smart cities depends heavily on the precise matching of Chinese addresses. Existing methods that rely on rules or text similarity struggle when dealing with nonstandard address data. Deep-learning-based methods often require extracting address semantics for embedded representation, which not only complicates the matching process, but also affects the understanding of address semantics. Inspired by deep transfer learning, we introduce an address matching approach based on a pretraining fine-tuning model to identify semantic similarities between various addresses. We first pretrain the address corpus to enable the address semantic model (abbreviated as ASM) to learn address contexts unsupervised. We then build a labelled address matching dataset using an address-specific geographical feature, allowing the matching problem to be converted into a binary classification prediction problem. Finally, we fine-tune the ASM using the address matching dataset and compare the output with several popular address matching methods. The results demonstrate that our model achieves the best performance, with precision, recall, and an F1 score above 0.98.