Generating time series of PRI across the Amazon basin using hyperspectral Hyperion data

Models of carbon cycling in tropical forests rely on estimates of fAPAR and LUE that are based on field experiments or remote sensing. However, much uncertainty surrounds such models on a regional scale because of unresolved questions over ecosystem response to climate variability. The photochemical reflectance index (PRI) is a normalized difference index that is sensitive to short-term changes in photosynthetic activity, unlike other more commonly used indices like NDVI. This is because PRI is indicative of the epoxidation state of xanthophyll pigments involved in non-photochemical quenching. Hence, an understanding of how PRI varies spatially and temporally would be a useful benchmark against which to evaluate model output and measurements of solar-induced fluorescence (SIF). Computing PRI requires two narrow spectral bands at 531 and 570 nm, and therefore cannot be quantified using broad-band optical sensors. The Hyperion instrument has generated Earth observations since 2000 and provides a unique opportunity to quantify PRI over large spatial and temporal scales. Here we describe a quantitative framework to correct striping, spectral keystone, and spectral smile inherent in the Hyperion data record. We show how these data can be used to generate time series of PRI to facilitate deeper understanding of photosynthetic activity in the Amazon basin.