Applications of hyperspectral remote sensing and GIS for assessing forest health and air pollution

The objective of this project is the assessment of air pollution impacts on conifer health in the Sierra Nevada of California, USA and the Pyrenees of Catalonia, Spain using remote sensing indices of forest health in conjunction with GIS analyses of the variability various stressors across natural landscape gradients. The Ozone Injury Index (OII) field metric applied to P. ponderosa and P. jeffreyi in the USA and adapted to P. uncinata in Spain included chlorotic mottling, needle retention, needle length, and crown depth. Species-level classifications of AVIRIS and CASI hyperspectral imagery were all near 80% overall accuracy for the target bioindicator species. Combining remote sensing indices with GIS variables related to microsite ozone uptake variability produced improved regressions for Catalonia (R
²
=0.68, p<;0.0001) and California (R
²
=0.56, p<;0.0001). Multiple regression models for the ozone injury visual component (VI) alone performed much better than the full OII in Catalonia combining the remote sensing index PRI and a three year average of ambient ozone (R
²
=0.56, p<;0.0001) and better still when including GIS variables (R
²
=0.77, p<;0.0001).