Research
Boya (Paul) Zhang studies how water and ecosystems change across inland and coastal environments using satellite Earth observation, hydrology, and artificial intelligence.
Understanding a Changing Water World from Space
Water connects wetlands, agricultural landscapes, coasts, and the ocean. Yet floods, water-level change, and water hidden beneath vegetation can be difficult to observe at the scales where decisions are made. My research combines satellite Earth observation, hydrology, field knowledge, and AI to reveal how water changes across space and time—and how ecosystems respond to climate extremes, sea-level rise, and water management.
Research Manuscripts
Spatiotemporal variations of wetland backscatter: The role of water depth and vegetation characteristics in Sentinel-1 dual-polarization SAR observations
By challenging the long-standing assumption that Synthetic Aperture Radar (SAR) backscatter is insensitive to hydrological processes beneath dense vegetation, Dr. Zhang established field-defining SAR-based frameworks that characterize interactions among microwave signals, vegetation structure, and underlying water surfaces. Using the South Florida Everglades as a model system, his research demonstrated clear linear relationships between SAR backscatter and water levels, and those relationships vary across vegetation types. This work has been widely cited and provides an important foundation for flood mapping and estimation of water level dynamics using SAR observations.
Space-based detection of significant water-depth increase induced by Hurricane Irma in the Everglades wetlands using Sentinel-1 SAR backscatter observations
As extreme weather events such as hurricanes become more frequent and intense, there is an increasing need for rapid detection of wetland areas experiencing significant water level rise. Using the South Florida Everglades and Category-3 Hurricane Irma as a case study, this work demonstrates that SAR backscatter can reliably detect water level changes and track post-hurricane recovery to pre-event baseline conditions. SAR acquisitions on the day of landfall provide timely and accurate characterization of areas with substantial inundation and hydrological disturbance.
Automated Reference Points Selection for InSAR Time Series Analysis on Segmented Wetlands
California’s Central Valley contains extensive, highly fragmented wetlands that provide critical habitat for migratory birds along the Pacific Flyway. Because these wetlands are actively managed and have complex water-level dynamics, assuming a constant water level can lead to substantial inaccuracies in estimating water storage. This study developed an automated InSAR time-series approach that identifies optimal reference points in heterogeneous wetlands, where low-coherence areas such as open water and dense vegetation make conventional reference-point selection challenging. The method achieved water-level estimates with an RMSE as low as 8 mm.
Exemplary publications:
Space-Based Mapping of Pre- and Post-Hurricane Mangrove Canopy Heights Using Machine Learning with Multi-Sensor Observations
As climate change intensifies hurricanes, coastal mangroves face increasing threats from storm damage, erosion, and canopy loss. Using Category 3 Hurricane Irma as a case study, this study integrated airborne LiDAR, spaceborne SAR, and optical imagery to assess mangrove canopy height before and after the storm. Developed spatial and temporal transfer-learning frameworks that integrate multi-sensor satellite observations and machine learning to enable landscape-scale canopy height mapping and cross-event prediction of post-hurricane forest structure from pre-Irma observations, achieving RMSEs of 1.9 m and 1.7 m, respectively.
Exemplary publications:
When salt meets restoration: shifts in salinity levels reshape coastal wetlands under rising seas
Sea-level rise and saltwater intrusion are reshaping freshwater marshes in the Southeastern Saline Everglades (SESE). Using 41 years of Landsat imagery and transfer learning, this study mapped changes in vegetation and wetland cover. We found that marsh degradation has expanded over the past four decades as salinity increased. Areas with greater freshwater inputs showed less degradation and greater expansion of salt-tolerant wetlands, while areas with moderate to high salinity experienced more severe marsh loss and retreat of saline wetlands. The greatest changes occurred in areas where infrastructure has reduced freshwater flow.