Using topographic and remotely sensed variables to assess ozone injury to conifers in the Sierra Nevada (USA) and Catalonia (Spain)

The capacity to remotely identify impacts of ozone on conifers in California, USA and Catalonia, Spain was investigated using remote sensing and terrain-driven GIS analyses related to plant water relations and ozone uptake. The Ozone Injury Index (OII) field metric applied to Pinus ponderosa and Pinus jeffreyi in the USA and adapted to Pinus uncinata in Spain included visible chlorotic mottling, needle retention, needle length, and crown depth. Species classifications of AVIRIS and CASI hyperspectral imagery all approached 80% overall accuracy for the target bioindicator species. Remote sensing vegetation indices correlated best with longer-wavelength SWIR indices from the AVIRIS data in California, with the exception of the Photosynthetic Reflectance Index (PRI) correlation with the OII Visual Component (OIIVI), which was also the highest direct correlation in Catalonia. In Catalonia, the OIIVI alone and its subparts correlated better with the CASI data than with the full OII, namely the PRI (R2 = 0.28, p = 0.0044 for OIIVI-amount and R2 = 0.33 and p = 0.0016 for OIIVI-severity). Stepwise regression models of ozone injury developed using remote sensing indices combined with terrain-derived GIS variables were significant for OII in California (R2 = 0.59, p < 0.0001) and in Catalonia (R2 = 0.68, p < 0.0001 for OIIVI). Multiple regression models of ozone injury including a three year average of O3 exposure were significant both with imaging spectroscopy indices alone (R2 = 0.56, p < 0.0001) and with topographic variables added (R2 = 0.77, p < 0.0001) in Catalonia. Applying the multivariate models to image classifications could provide useful maps useful for ozone impact monitoring but requires further validation before being considered operational.