We developed and tested a method for mapping shade cast on water channels by riparian vegetation using imagery sources that are affordable and regularly available across Oregon. We validated the optical imagery-based results against lidar-based shade estimates that, while more accurate, are not viable for ongoing statewide monitoring due to their expense. This work provides insights toward determining a viable strategy for statewide monitoring of riparian vegetation condition, which could in turn support a data-driven prioritization and assessment framework to increase the efficiency, effectiveness and accountability of riparian restoration efforts.

We created a model to predict shade from 1-foot Oregon Statewide Imagery Program (OSIP) imagery and 10-meter Sentinel-2 satellite imagery, trained from shade estimated using solar-path modeling applied to lidar data collected over three disparate study areas. The model explains nearly 77% of the variation in lidar-derived shade across the study areas. NAIP-based models significantly outperformed Sentinel-2 models; we found that multi-scale image textural information derived from NAIP was important in creating accurate shade estimates. Maps of shade from the optical-based model were created over the entire Johnson Creek watershed in metropolitan Portland; visual inspection of the results shows a very high correspondence to photo-interpreted NAIP imagery, including accurate response to subtle and fine-scale variation in conditions.