Vegetation index sensitivity analysis for wetland vegetation phenology using Sentinel-2 Data

Abstract:

Wetlands are a vitally important resource to both humans and wildlife. Billions of dollars are spent annually on wetland restoration. Tracking wetland vegetation phenology using satellite remote sensing provides a cost-effective way to monitor wetland development and maturity and support adaptive management. Wetland phenology should be characterized in a way that preserves the distinctions between plant functional types because vegetation indexes may perform differently across types. Other factors including history of restoration and environmental conditions can also influence the effectiveness of these indexes. Here, we evaluate the potential of Sentinel-2 data to monitor wetland phenology stratified by plant functional types. We compare six different vegetation indexes across six different restoration sites located in the Sacramento – San Joaquin Delta (California, USA) and track these index trajectories from 2016 to 2020. We performed a sensitivity analysis using Sentinel-2 imagery paired with temporally-aligned plant functional type classifications (emergent, submerged, and floating macrophytes) to create a vegetation index time-series at each study site to characterize how index selection may impact phenological assessment across different environmental conditions and vegetation compositions. In addition to improving characterization of wetland development and maturity, this research supports agencies responsible for wetland restoration and management by providing a user-friendly and extensible Google Earth Engine graphical and scripting interface to increase accessibility to remote sensing analysis and support the monitoring of these restoration sites in the years to come.