Using hyperspectral and thermal imagery to monitor stress of Southern California plant species during the 2013–2015 drought

Abstract

From 2012 to 2015, California experienced the most severe drought since 1895, causing natural vegetation throughout the state to become water-stressed. With many areas in California being inaccessible and having extremely rugged terrain, remote sensing provides a means for monitoring plant stress across a broad landscape. Airborne hyperspectral and thermal imaging captured the drought in the spring, summer, and fall seasons of 2013 – 2015 across 11,640 km² of Southern California. Here we provide a large-scale analysis of plant species’ annual and seasonal temperature variability throughout this prolonged drought. We calculated the Temperature Condition Index (TCI) using airborne thermal imagery and a plant species classification map derived from airborne hyperspectral imagery to track response in three dominant species (e.g., Mediterranean grasses and forbs, chamise, and coast live oak) that have different stress adaptation strategies. The annual grasses and forbs showed strong seasonal changes in TCI, which corresponded to the typical green-up, peak biomass in summer, and senescence in the fall. They also had the strongest change in TCI values as the drought progressed from 2013 to 2015, with the months of April and August showing the most pronounced changes. The deeper rooted, native chamise evergreen shrub and coast live oak evergreen, broadleaf tree showed a more minor shift in seasonal and yearly patterns of TCI, but even these very well adapted species showed an increased amount of TCI stress as the drought progressed from 2013 to 2015. Across the study area and image dates, TCI stress was not evenly distributed, and in August 2015 almost the entire region experienced elevated TCI stress. To better understand the environment’s effect on plant stress, we relate topographic attributes to plant stress. Higher TCI values correlated with south or south-southwest facing slopes, while other topographic attributes were weakly correlated with TCI. An increase in elevation had a strong correlation with a decrease in TCI stress, but this relationship weakened as the drought progressed. The synergistic capabilities of hyperspectral and thermal imagery demonstrate that we can monitor the dynamic nature of plant species’ stress temporally and spatially. This work supports improved monitoring of natural landscapes and informing management possibilities, especially for areas prone to continued drought and high risk of wildfires.