Improving Estimates of Aboveground Biomass as Trajectory Indicators in Tidal Wetlands using NAIP Imagery

In an effort to improve carbon accounting in tidal marshes, a key "blue carbon" ecosystem, we developed a method for integrating high resolution imagery into biomass estimates and trajectory indicators of land cover change. Currently, NOAA's Coastal Change Analysis Program (C-CAP) provides tidal marsh cover classes nationally based on 30m Landsat imagery. In order to improve wetland biomass estimates derived from C-CAP land cover and Landsat reflectance data, we developed a methodology to better define biomass extents using 1m National Agriculture Imagery Program (NAIP) imagery. For each of six sentinel estuary sites around the U.S., we performed an object-based segmentation on NAIP imagery. Three classes, `green vegetation', `non- vegetation', and `water' were identified using four NAIP bands (red, green, blue and near infrared), the normalized difference vegetation index (NDVI), and normalized difference water index (NDWI), in a rule-based classification. We performed an accuracy assessment at each site with results ranging from 79% to 92% overall accuracy. Using a fishnet grid of Landsat pixel extents, combined with our 1m classification, we calculated the proportion of green vegetation, non-vegetation, and water per 30m pixel. The proportion of green vegetation was used to scale biomass measurements to improve carbon stock estimates. In addition, we explored the use of multi-temporal NAIP-based fractional cover estimates as trajectory indicators for active marsh loss and restoration as defined by the C-CAP land cover transitions. With NAIP imagery collected ever two years, these methods provide a feasible approach for U.S. blue carbon accounting over time.