Abstract
Local climate zone (LCZ) has become a new standard classification scheme in urban landscapes and showed great potential in urban climate research. Traditional classifiers and ordinary neural networks only consider the spectral or local spatial features of the pixel, ignoring the effect of nonlocal information on the LCZ classification. The graph convolutional network (GCN) has been used to exploit the relationship between adjacent and global land covers owing to the ability to conduct flexible convolution over graphs. In this work, we integrated a convolutional neural network and two GCNs into an end-to-end hybrid framework and generated LCZs directly from the original images. Local-, regional-, and global-level features were extracted and grouped complementarily to foster better performance. Experiments were conducted in six cities around the world to verify the effectiveness of our method. Results showed that the average classification accuracy of the six cities reached 0.956 and performed better than any other comparable model. Ablation experiments also demonstrated the mutual promotion of the different modules. Finally, the small sample experiment provided a practical reference for the LCZ classification in the absence of samples in future.
Highlights
CITIES are currently home to half of the world’s population, and all of the projected 1.1 billion global population growth will be virtually in urban areas over the 15 years [1]
Many classification models are available, such as the national land cover dataset (NLCD) product of the United States, which has 20 land cover types with only four urban classes based on urban building density [3]; the European CORINE (Co-ORdinated INformation on the Environment) land cover database, which contains 44 land cover categories with only three urban areas [4]; and the Global Land Cover product (GLC30) [5] with 10 classes and only “artificial surface” referring to all artificial impervious surfaces
The results show that the network with regional graph convolutional network (RGCN) or global GCN (GGCN) provides a significant improvement in OA compared with the performance of baseline, which demonstrates the effectiveness of the graph convolutional network (GCN) module
Summary
CITIES are currently home to half of the world’s population, and all of the projected 1.1 billion global population growth will be virtually in urban areas over the 15 years [1]. Many classification models are available, such as the national land cover dataset (NLCD) product of the United States, which has 20 land cover types with only four urban classes based on urban building density [3]; the European CORINE (Co-ORdinated INformation on the Environment) land cover database, which contains 44 land cover categories with only three urban areas [4]; and the Global Land Cover product (GLC30) [5] with 10 classes and only “artificial surface” referring to all artificial impervious surfaces These classification schemes only have rough city types and do not provide the necessary detail. The local climate zone scheme was proposed to fulfill all the requirements [6]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
