Monitoring Systematic Dryland Ecosystem Optical Properties with a Semi-automated Robotic Tram System at Jornada Experimental Range

Dryland systems are crucial for determining interannual variation of the global carbon (C) cycle. Variation in C sink strength and plant productivity is driven by large interannual variability in rainfall. A strong link between plant productivity and C sink strength is promising for using remotely sensed indices to scale dryland C dynamics across space and time. However, it is unclear how individual species responses affect landscape-level remotely sensed indices. Here, we use high resolution, plot level, hyper-spectral measurements to understand how species-specific plant responses to abiotic conditions drive vegetation indices at the land-scape scale. Within the Jornada Experimental Range in the Chihuahua Desert, land surface optical properties have been observed since 2009. Hyperspectral reflectance data were semi-automatically gathered in the visible to near-infrared range with ancillary red, green, blue plot level images, air temperature, precipitation and sky conditions across a 110 meter transect using a robotic tram system. These data were used to explore the spatio-temporal variation of Vegetation Indices (eg: NDVI, PRI, EVI) for Creosote, Mesquite, and bare ground. We found unique spectral signatures for bare ground and vegetation. Bare ground looks distinct from vegetation and the two vegetation types differ seasonally. Bare ground reflectance has an almost linear increase from 400-900nm and the magnitude of reflectance rises in all wavelengths when the ground was drier. Green vegetation shows peaks at visible region and near infrared region giving a signature that is distinct from bare ground. The signature was similar for Creosote in summer and winter, because it is evergreen. In contrast, the signature of deciduous Mesquite varied seasonally. In summer Mesquite and Creosote had similar spectral signatures while in winter Mesquite was more similar to bare ground. This study shows that ground based remote sensing approaches can provide detailed spectral profiles of landcover types in dryland systems. Better understanding of species-specific variation in spectral indices can help interpret landscape-scale remote sensing (eg: cameras and satellites) and create a better understanding for how plant-responses drive C dynamics when modeling land-atmosphere carbon exchange.