A5085580167- Land Use and Ecosystem Services
- Spatial and Panel Data Analysis
- Geochemistry and Geologic Mapping
- Remote-Sensing Image Classification
- Marine and coastal ecosystems
- Geological and Geochemical Analysis
- Oceanographic and Atmospheric Processes
- Advanced Image Fusion Techniques
- Regional Economic and Spatial Analysis
- Atmospheric and Environmental Gas Dynamics
- Marine and fisheries research
- Air Quality and Health Impacts
- Air Quality Monitoring and Forecasting
- Hydrocarbon exploration and reservoir analysis
- Mineral Processing and Grinding
- Cryospheric studies and observations
- Climate variability and models
- Soil Geostatistics and Mapping
- Climate change and permafrost
- Remote Sensing in Agriculture
- Geological and Geophysical Studies
- Geographic Information Systems Studies
- Advanced Neural Network Applications
- Advanced Image Processing Techniques
- Image and Signal Denoising Methods
Zhejiang University
2017-2025
Capital Medical University
2022-2024
Chinese University of Hong Kong
2021
Kunming University of Science and Technology
2020
Zhejiang Provincial Institute of Communications Planning,Design & Research
2019
University of Hong Kong
1997
Geographically weighted regression (GWR) is a classic and widely used approach to model spatial non-stationarity. However, the makes no precise expressions of its weighting kernels insufficient estimate complex geographical processes. To resolve these problems, we proposed geographically neural network (GNNWR) that combines ordinary least squares (OLS) networks non-stationarity based on concept similar GWR. Specifically, designed spatially (SWNN) represent nonstationary weight matrix in...
Geographically weighted regression (GWR) and geographically temporally (GTWR) are classic methods for estimating non-stationary relationships. Although these have been widely used in geographical modeling spatiotemporal analysis, they face challenges adequately expressing space-time proximity constructing a kernel with optimal weights. This probably results an insufficient estimation of non-stationarity. To address complex non-linear interactions between time space, neural network (STPNN) is...
The accurate assessment of large-scale and complex coastal waters is a grand challenge due to the spatial nonstationarity nonlinearity involved in integrating remote sensing situ data. We developed water quality method based on newly proposed geographically neural network weighted regression (GNNWR) model address that obtained highly realistic distribution basis comprehensive index Chinese Water Quality Classification Standards. Using geostationary ocean color imager (GOCI) data observations...
Estimating crop yield in large areas is essential for ensuring food security and sustainable development. Accounting variations the temporal cumulative growth of crops across regions (i.e., spatial heterogeneity growth) can improve accuracy estimation areas. However, current learning methods have limitations such as cutting off inherent correlations among regions, difficulty obtaining accurate prior knowledge, high subjectivity. Therefore, this study proposed a novel deep adaptive model...
Geographically weighted regression (GWR) is a classical method of modeling spatially non-stationary relationships. The geographically neural network (GNNWR) model solves the problem inaccurate construction spatial weight kernels using network. However, when distribution observations uneven, proximity expression in input GWR and GNNWR models does not fully represent impact whole research space on estimating point. Therefore, we established global grid (GSPG) to express each point proposed...
Spatial downscaling is an important approach to obtain high-resolution land surface temperature (LST) for thermal environment research. However, existing methods are unable sufficiently address both spatial heterogeneity and complex nonlinearity, especially in scenes (<120 m), accordingly limit the representation of regional details accuracy inversion. In this study, by integrating normalized difference vegetation index (NDVI), building (NDBI), digital elevation model (DEM), slope data, a...
The spatial attention mechanism has been frequently employed for the semantic segmentation of remote sensing images, given its renowned capability to model long-range dependencies. As images often exhibit intricate backgrounds, significant intraclass variability, and a foreground-background imbalance, mechanism-based methods somehow tend introduce an extensive amount background context through intensive affinity operations, causing unsatisfactory outcomes. While several class-aware attempt...
Historical news media reports serve as a vital data source for understanding the risk of urban ground collapse (UGC) events. At present, application large language models (LLMs) offers unprecedented opportunities to effectively extract UGC events and their spatiotemporal information from vast amount data. Therefore, this study proposes an LLM-based inventory construction framework consisting three steps: crawling, event recognition, attribute extraction. Focusing on Zhejiang province, China,...
Porphyry copper ore is a vital strategic mineral resource. It often associated with significant hydrothermal alteration, which alters the original mineralogical properties of rock. Extracting alteration information from remote sensing data crucial for porphyry exploration. However, current method extracting ASTER does not consider influence disturbing factors, such as topography, and ignores weak report surface minerals, has limitations. Therefore, this paper selects Gondwana region East...
Accurate prediction of mineral resources is imperative to meet the energy demands modern society. Nonetheless, this task often difficult due estimation bias and limited interpretability conventional statistical techniques machine learning methods. To address these shortcomings, we propose a novel geospatial artificial intelligence approach, denoted as geographically neural network-weighted logistic regression, for prospectivity mapping. This model integrates spatial patterns networks,...
Accurate crop classification using remote sensing imagery with limited labeled data remains a challenging yet highly valuable task in practical applications. Recently, self-supervised contrastive learning has shown considerable potential generating discriminative and generalized features from unlabeled images. Nevertheless, due to the inherent complexity of planting structures growth patterns, existing methods struggle fully capture distinct spatial spectral characteristics various crops. To...