Developing imaging tools to better measure vegetation distribution and health

One of our main foci is the development of tower-based imaging systems to better quantify and understand vegetation health. Terrestrial ecology has historically been a data limited discipline. Key ecological measurements, such as plant species composition and growth, are usually made by hand. Ecologists have traditionally spent many days in the field manually measuring ecosystem attributes and recording their observations in log books. A great deal of effort is required to characterize just a few ecological aspects of a small patch of land. Terrestrial ecology is reaching a crossroads: it remains a data limited discipline but there is hope for a quantum leap forward in our ability to monitor and characterize ecosystems. This opportunity comes from the revolution in consumer electronics, homeland security and surveillance, computers and data storage, image and data analysis software, and telecommunications. The challenge is to adopt these technologies to measure important ecological properties, such as plant physiology, vegetation health, vegetation structure, species composition, and invasive species presence.

We have developed and deployed an imaging system at our eddy covariance site in Pinyon-Juniper woodland to quantify the species-level patterns of stress and mortality over time, and also to learn how to better interpret the Landsat record. The imaging system combines a four channel spectrometer with cameras that are sensitive to Visible, Near Infrared (NIR), Shortwave Infrared (SWIR), and Thermal radiation; these cameras include filters that mimic the spectral sensitivity of several Landsat bands. The cameras and spectrometer foreoptic are positioned on a pan-tilt mount on the tower that scans a 300o x 90o area every hour and allows us to collect images of hundreds of distinct plants. The imaging system is being used to test several approaches that have been proposed to detect vegetation stress, mortality, and species composition. We are exploring the potential to detect stomatal closure and stress by: a) increased canopy temperature with decreased evaporative cooling, b) Photochemical Reflectance Index (PRI), c) Fraunhofer line fluorescence, and d) water band indices. Similarly, we are exploring the potential to detect plant mortality by: a) NIR reflectance, b) SWIR reflectance, and c) radiance temperature with soil exposure, and to identify plant species by: a) differential phenological and interannual patterns, b) spectral reflectance, and c) BRDF and the effect of solar angle.

Rapid Characterization of Vegetation Structure with a Microsoft Kinect Sensor Sensors (1424-8220) 13:2, 2013 Azzari, George; Goulden, Michael; Rusu, Radu