Multi-scale functional mapping of tidal marsh vegetation for restoration monitoring

Nearly half of the world's natural wetlands have been destroyed or degraded, and in recent years, there have been significant endeavors to restore wetland habitat throughout the world. Detailed mapping of restoring wetlands can offer valuable information about changes in vegetation and geomorphology, which can inform the restoration process and ultimately help to improve chances of restoration success. I studied six tidal marshes in the San Francisco Estuary, CA, US, between 2003 and 2004 in order to develop techniques for mapping tidal marshes at multiple scales by incorporating specific restoration objectives for improved longer term monitoring. I explored a "pixel-based" remote sensing image analysis method for mapping vegetation in restored and natural tidal marshes, describing the benefits and limitations of this type of approach (Chapter 2). I also performed a multi-scale analysis of vegetation pattern metrics for a recently restored tidal marsh in order to target the metrics that are consistent across scales and will be robust measures of marsh vegetation change (Chapter 3). Finally, I performed an "object-based" image analysis using the same remotely sensed imagery, which maps vegetation type and specific wetland functions at multiple scales (Chapter 4). The combined results of my work highlight important trends and management implications for monitoring wetland restoration using remote sensing, and will better enable restoration ecologists to use remote sensing for tidal marsh monitoring. Several findings important for tidal marsh restoration monitoring were made. Overall results showed that pixel-based methods are effective at quantifying landscape changes in composition and diversity in recently restored marshes, but are limited in their use for quantifying smaller, more fine-scale changes. While pattern metrics can highlight small but important changes in vegetation composition and configuration across years, scientists should exercise caution when using metrics in their studies or to validate restoration management decisions, and multi-scale analyses should be performed before metrics are used in restoration science for important management decisions. Lastly, restoration objectives, ecosystem function, and scale can each be integrated into monitoring techniques using remote sensing for improved restoration monitoring.