Research
My research integrates multi-source Earth observation, long-term satellite time series, spatial analysis, and GeoAI to understand the dynamics of coastal and wetland ecosystems across local to national scales.
I develop interpretable, transferable, and operational GeoAI methods for long-term ecosystem mapping, change detection, and restoration monitoring. My current research includes intelligent identification of coastal wetlands, monitoring ecosystem evolution, tracking Spartina alterniflora invasion, removal, and reinvasion, and large-scale mapping driven by multi-source remote sensing data.
Research Themes
Coastal Ecosystem Mapping
Salt marshes, tidal flats, coastlines, mangroves, and other intertidal ecosystems.
Wetland Dynamics and Ecological Assessment
Long-term ecosystem change, anthropogenic disturbance, restoration, and biological invasion.
Interpretable GeoAI
Knowledge-guided classification, machine learning, deep learning, foundation models, and uncertainty-aware model fusion.
Multi-source Earth Observation
Integration of SAR, optical imagery, ocean-colour data, meteorological observations, and socioeconomic information.
Featured Ongoing Research
National-scale monitoring of tidal-flat dynamics
This research develops an efficient and reproducible framework for mapping tidal flats along the coast of China using long-term Landsat observations. The workflow automatically identifies low-tide imagery and integrates knowledge-based rules to reduce interference from clouds, waves, and tidal variability.
The resulting time-series dataset reveals major tidal-flat dynamics in the Liaohe River Delta, Yellow River Delta, Yangtze River Delta, Jiangsu coast, and other rapidly changing coastal regions. The work supports national-scale wetland monitoring, restoration assessment, and sustainable coastal management.
Completed Research Projects
National-scale mapping of coastal salt marshes
This study developed a knowledge-based classification framework using time-series Sentinel-1 SAR observations to produce a 10-m map of coastal salt marshes across China. Annual SAR composite features were used to reduce the effects of tidal fluctuations and seasonal variability.
The study mapped approximately 127,477 ha of salt marshes and distinguished four dominant vegetation communities with an overall accuracy of 87.3%. The resulting dataset provides a national baseline for wetland conservation, restoration, and invasive-species management.
Salt marsh · Sentinel-1 · SAR · Time series · National mapping
View Paper
Monitoring suspended sediment through a complete tidal cycle
This study used high-frequency GOCI observations to investigate tidal-driven variations in suspended sediment concentration in Hangzhou Bay. An empirical remote-sensing model was developed using field measurements and hourly satellite imagery.
The results characterized the spatial and temporal response of suspended sediment to different tidal stages and demonstrated the value of geostationary ocean-colour imagery for monitoring highly dynamic and turbid coastal waters.
Suspended sediment · GOCI · Hangzhou Bay · Tidal dynamics
View Paper
Remote sensing of photovoltaic infrastructure in coastal China
This research combined Sentinel-1 SAR and Sentinel-2 optical imagery with morphological, spectral, and spatial features to map photovoltaic power stations across China’s coastal provinces.
The random-forest framework achieved an overall accuracy of 96.9% and identified approximately 837.3 km² of photovoltaic installations. The subsequent spatial analysis examined their relationships with coastal proximity, solar resources, and regional electricity demand.
Photovoltaics · Sentinel-1/2 · Random forest · Coastal provinces
View Paper
Long-term ecological assessment of China’s Ramsar sites
This study assessed ecological change across China’s Ramsar sites using 36 years of dense Landsat observations. Seven indicators representing vegetation condition, hydrological dynamics, and anthropogenic disturbance were integrated into a national ecological assessment framework.
More than half of the evaluated sites showed long-term ecological degradation, while many sites exhibited signs of recent recovery. The study demonstrates how long-term satellite records can support systematic wetland conservation assessment.
Ramsar sites · Landsat · Ecological assessment · Change detection
View Paper
Economic drivers of coastal-wetland change
This research investigated the relationship between socioeconomic development and coastal-wetland transitions along the East China Sea from 1985 to 2020. Remote-sensing observations, socioeconomic indicators, and the Environmental Kuznets Curve framework were combined to identify long-term patterns and potential drivers.
The results reveal distinct responses among natural and artificial wetlands and highlight the influence of reclamation, urbanization, regional development, and environmental policy on coastal landscapes.
Coastal wetlands · Socioeconomic drivers · Reclamation · East China Sea
View Paper
GeoAI recognition of salt-marsh fairy circles
This study developed a generalizable GeoAI framework for recognizing and counting salt-marsh fairy circles from conventional high-resolution RGB satellite imagery.
The framework combines zero-shot Segment Anything Model segmentation, tailored spatial and geometric features, ensemble and deep-learning models, and imbalance-aware Bayesian probability updating. It enables the monitoring of rare ecological patterns under limited training data and severe class imbalance.
Salt-marsh fairy circles · GeoAI · SAM · Bayesian fusion · Deep learning
View Paper
Remote sensing of agricultural disruption in conflict zones
This research integrated Sentinel-1 and Sentinel-2 time-series observations, crop phenology, machine learning, and transfer learning to map sunflower cultivation and estimate production changes in eastern Ukraine.
The framework provides a scalable approach for assessing agricultural disruption in regions where field observations are unavailable. It links spatial patterns of crop loss with conflict intensity, population displacement, and damage to agricultural infrastructure.
Agricultural systems · Conflict · Sunflower · Time-series mapping
View Paper
Rainy-season onset over the Qinghai–Tibet Plateau
This study investigated the spatial and temporal variability of rainy-season onset over the Qinghai–Tibet Plateau using daily precipitation observations from 106 meteorological stations between 1971 and 2015.
The analysis identified a clear southeast-to-northwest progression of rainy-season onset and examined its long-term trends and regional differences under the combined influence of topography and atmospheric circulation.
Rainy-season onset · Qinghai–Tibet Plateau · Climate variability
View Paper