While forests across the world experience unprecedented rates of environmental change, our ability to monitor and react to these changes is also increasing rapidly. Airborne remote sensing offers the unique opportunity to observe every tree canopy within a research plot, however, the reflectance data are strictly empirical and require calibration with field measurements if they are to be useful in diagnosing tree nutrient status. Here we tested the ability of the Airborne Observatory Platform of the National Ecological Observatory Network to detect the response of northern hardwood forests to experimental nutrient addition and we compared crown-level hyperspectral profiles to traditional leaf chemistry analyses. Experimental nutrient addition involved 8 years of 30 kg N/ha/yr as NH4NO3, 10 kg P/ha/yr as NaH2PO4, a combined N+P treatment at the same rates, and a one-time application of 1150 kg/ha Ca as CaSiO2 in 2011. Within three mature hardwood stands, we matched individually tagged trees with a geo-referenced aerial photograph. We selected fully lit pixels from each tree crown and calculated the Photochemical Reflective Index (PRI) to infer light use efficiency and photoinhibition. We also examined how nutrient addition alters canopy reflectance and whether this can be used to remotely sense tree nutrient status.
Results/Conclusions
Preliminary comparisons of six vegetation indices (normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), soil-adjusted vegetation index (SAVI), normalized difference nitrogen index (NDNI), normalized difference lignin index (NDLI), and photochemical reflective index (PRI)) suggest that PRI is sensitive to nutrient addition, with less negative PRI values observed for trees that received N than those that did not. This remotely sensed nutrient treatment response is novel because experimental manipulations are rare in remote sensing studies of forested ecosystems, even though they offer stronger interpretations. Reflectance intensity values for 426 spectral bands with 1-m spatial resolution were used to further explore spectral properties associated with nutrient addition. Trees that received N or P had marginally higher reflectance values than trees that did not receive nutrient treatment. This research may improve aerial forest inventory and forest health monitoring, thereby increasing our ability to track individual trees in response to environmental changes. Future investigations will examine sections of spectral profiles for signs of nutrient limitation and assess relationships between tree growth and canopy reflectance.