Remote Sensing of Bioindicators for Forest Health Assessment

The impacts of tropospheric ozone on forest health in Mediterranean type climates in California, USA and Catalonia, Spain were investigated using a combination of remote sensing, Geographic Information System (GIS), and field studies focused on sensitive bioindicator conifer species and ambient ozone monitoring. For the field validation of impacts of tropospheric ozone on conifer health, the Ozone Injury Index (OII) was applied to the bioindicator species Pinus ponderosa, Pinus jeffreyi, and Pinus uncinata. Combining these three tools, it was possible to build meaningful ecological models covering large areas to enhance our understanding of the biotic and abiotic interactions which affect forest health. Regression models predicting ozone injury improved considerably when incorporating ozone exposure with GIS related to plant water status, including water availability and water usage, as a proxies for estimating the stomatal conductance and ozone uptake R2=0.35, p = 0.016 in Catalonia, R2=0.36, p < 0.001 in Yosemite and R2=0.33, p = 0.007 in Sequoia/Kings Canyon National Parks in California). Individual OII components in Catalonia were modeled with improved success compared to the original full OII, in particular visible chlorotic mottling (R2=0.60, p < 0.001). The visual chlorotic mottling component of the OII was the most strongly correlated to remote sensing indices, in particular the photochemical reflectance index (PRI; R2=0.28, p=0.0044 for OIIVI-amount and R 2=0.33 and p=0.0016 for OIIVI -severity). Regression models assessing ozone injury to conifers using imaging spectroscopy techniques also improved when incorporating the GIS proxies of stomatal conductance (R 2=0.59, p<0.0001 for OII in California and R2=0.68, p<0.0001 for OIIVI in Catalonia). Finally, taking advantage of a time series of ambient ozone monitoring in Catalonia, it was found that all models improved when incorporating the cumulative exposure to ozone over a period of three years (R2=0.56, p<0.0001 with imaging spectroscopy indices alone and R2=0.77, p<0.0001 with GIS added) and that it was possible to model the three year average ambient ozone using a modified version of the OII (P<0.0001, R2=0.53, RMSE=2.73 with only the OII subcomponents VI-Severity and FWHORL and P<0.0001, R2 = 0.90, RMSE = 1.35 with GIS).