Gradient Nearest Neighbor (GNN) Mapping as Applied to California Oak Woodland

Authors: David Bell, Matt Gregory and Hans-Erik Andersen, USDA Forest Service, Pacific Southwest Research Station

To effectively manage and conserve California oak woodlands, it is important understand ecosystem status and trends. By integrating USDA Forest Service Forest Inventory and Analysis field measurements and multispectral Landsat imagery, gradient nearest neighbor (GNN) imputation mapping provides 30-m resolution, annual (1990-2017) maps of forest vegetation provide useful information for ecosystem monitoring. Here, we explore how GNN maps can be used to monitor California oak woodland status and trends. To understand how to appropriately use GNN to monitor oak woodlands, we examine GNN accuracy and appropriate scale for estimating the area of oak woodlands. We then assess trends in oak woodland area over a 27 year period to identify areas of expansion and decline. Finally, we highlight how maps and other auxiliary data can be used to understand changes in oak woodland spatial patterns within individual landscapes associated with disturbances, such as wildfire. This approach to exploring map accuracy, quantifying landscape-level trends, and diagnosing disturbance impacts can support California oak woodland conservation and management by highlighting landscapes where oak woodlands are at risk or prioritize restoration efforts.