Search

The local climate zone (LCZ) scheme is of great value for urban heat island (UHI) effect studies by providing a standard classification framework to describe the local physical structure at a global scale. In recent years, with the rapid development of satellite imaging techniques, both multi-spectral (MS) and synthetic aperture radar (SAR) data have been widely used in LCZ classification tasks. However, the fusion of MS and SAR data still faces the challenges of the different imaging mechanisms and the feature heterogeneity. In this study, to fully exploit and utilize the features of SAR and MS data, a data-grouping method was firstly proposed to divide multi-source data into several band groups according to the spectral characteristics of different bands. Then, a novel network architecture, namely Multi-source data Fusion Network for Local Climate Zone (MsF-LCZ-Net), was introduced to achieve high-precision LCZ classification, which contains a multi-branch CNN for multi-modal feature extraction and fusion, followed by a classifier for LCZ prediction. In the proposed multi-branch structure, a split–fusion-aggregate strategy was adopted to capture multi-level information and enhance the feature representation. In addition, a self channel attention (SCA) block was introduced to establish long-range spatial and inter-channel dependencies, which made the network pay more attention to informative features. Experiments were conducted on the So2Sat LCZ42 dataset, and the results show the superiority of our proposed method when compared with state-of-the-art methods. Moreover, the LCZ maps of three main cities in China were generated and analyzed to demonstrate the effectiveness of our proposed method. [ABSTRACT FROM AUTHOR]

Regional landslide identification is important for the risk management of landslide hazards. The traditional methods of regional landslide identification were mainly conducted by a human being. In previous studies, automatic landslide recognition mainly focused on new landslides distinct from the environment induced by rainfall or earthquake, using the image classification method and semantic segmentation method of deep learning. However, there is a lack of research on the automatic recognition of old loess landslides, which are difficult to distinguish from the environment. Therefore, this study uses the object detection method of deep learning to identify old loess landslides with Google Earth images. At first, a database of loess historical landslide samples was established for deep learning based on Google Earth images. A total of 6111 landslides were interpreted in three landslide areas in Gansu Province, China. Second, three object detection algorithms including the one-stage algorithm RetinaNet and YOLO v3 and the two-stage algorithm Mask R-CNN, were chosen for automatic landslide identification. Mask R-CNN achieved the greatest accuracy, with an AP of 18.9% and F1-score of 55.31%. Among the three landslide areas, the order of identification accuracy from high to low was Site 1, Site 2, and Site 3, with the F1-scores of 62.05%, 61.04% and 50.88%, respectively, which were positively related to their recognition difficulty. The research results proved that the object detection method can be employed for the automatic identification of loess landslides based on Google Earth images. [ABSTRACT FROM AUTHOR]

Vegetation maps are important sources of information for biodiversity conservation, ecological studies, vegetation management and restoration, and national strategic decision making. The current Vegetation Map of China (1:1000000) was generated by a team of more than 250 scientists in an effort that lasted over 20 years starting in the 1980s. However, the vegetation distribution of China has experienced drastic changes during the rapid development of China in the last three decades, and it urgently needs to be updated to better represent the distribution of current vegetation types. Here, we describe the process of updating the Vegetation Map of China (1:1000000) generated in the 1980s using a "crowdsourcing-change detection-classification-expert knowledge" vegetation mapping strategy. A total of 203,024 field samples were collected, and 50 taxonomists were involved in the updating process. The resulting updated map has 12 vegetation type groups, 55 vegetation types/subtypes, and 866 vegetation formation/sub-formation types. The overall accuracy and kappa coefficient of the updated map are 64.8% and 0.52 at the vegetation type group level, 61% and 0.55 at the vegetation type/subtype level and 40% and 0.38 at the vegetation formation/sub-formation level. When compared to the original map, the updated map showed that 3.3 million km2 of vegetated areas of China have changed their vegetation type group during the past three decades due to anthropogenic activities and climatic change. We expect this updated map to benefit the understanding and management of China's terrestrial ecosystems. [ABSTRACT FROM AUTHOR]

• Evaluated the performance of TLS data on deriving degraded grassland AGB. • The SMR model yields the highest accuracy for predicting AGB (R2 = 0.84). • Canopy cover and minimum height are the two best predictors for estimating AGB. • AGB prediction accuracy increases with the TLS point density. Aboveground biomass (AGB) is an important indicator for grassland ecosystem assessment, management and utilization. Remote sensing technologies have driven the development of grassland AGB estimation from labor-intensive to highly-efficient. However, optical image-based remote sensing methods are fraught with uncertainty issues due to the saturation effects. In this study, we evaluated the capability of the emerging terrestrial laser scanning (TLS) technique in estimating grassland AGB in the northern agro-pastoral ecotone of China. Seven variables (i.e., canopy cover, canopy volume, mean height, maximum height, minimum height, range of height, and standard deviation of height) were extracted from the TLS data of 30 plots across the northern agro-pastoral ecotone of China, and were used to build regression models with field measured AGB using four regression methods, which are simple regression (SR) model, stepwise multiple regression (SMR) model, random forest (RF) model and artificial neural network (ANN) model. The results demonstrate that TLS is a feasible technique for extracting grassland structural parameters. Mean grass height and canopy cover obtained from TLS data have good correspondence with field measurements (R2 > 0.7, p -values < 0.001). Among the four regression models, the SMR model yields the highest prediction accuracy (R2 = 0.84, RMSE = 48.89 g/m2), followed by the RF model (R2 = 0.78, RMSE = 68.72 g/m2), the SR model (R2 = 0.80, RMSE = 86.4 g/m2), and the ANN model (R2 = 0.73, RMSE = 101.40 g/m2). Minimum grass height and canopy coverage are the two most important variables influencing the prediction accuracy of the SMR model, and the prediction accuracy of the SMR model increases with the increase of point density. The results of this study can provide guidance for choosing the optimal model and data collection method for estimating degraded grassland AGB using TLS in agro-pastoral ecotone. [ABSTRACT FROM AUTHOR]