Developing a common globally applicable method for optical remote sensing of ecosystem light use efficiency

This study examines the use of spectral reflectance to determine ecosystem photosynthetic light use efficiency (LUE) during mid-growing season for 32 globally distributed flux tower sites. Surface reflectance for 133 spectral bands were extracted for areas around the flux towers from imagery collected by the Hyperion instrument on the Earth Observing 1 satellite. The average reflectance spectra were matched with LUE derived from CO2 flux data collected using eddy covariance techniques from the La Thuile Fluxnet Synthesis Dataset, resulting in 79 observations collected between 2000 and 2007. LUE was calculated from daily gross ecosystem production (GEP) and incident photosynthetically active radiation (PAR) from the flux towers and MODIS fraction of absorbed PAR (fPAR). An examination of all possible two band normalized difference vegetation indices found that no index performed better than R2 of 0.55, although 60 of the bands were included in normalized difference indices with a R2 > 0.45. A partial least squares regression (PLSR) using all spectral bands produced a R2 of 0.81, suggesting the use of this approach to develop a globally-applicable retrieval method for LUE using spectral imagery from future missions flying imaging spectrometers, such as the NASA Surface Biology and Geology mission. •Examined multispectral and hyperspectral approaches to link spectral reflectance to photosynthetic light use efficiency•Utilized over 70 Hyperion hyperspectral images of 32 different and widely dispersed sites•Combined spectral imagery with data from the FLUXNET global network of eddy covariance flux towers•Partial least squares regression applied to 133 spectral bands described over 80% of the variance in light use efficiency.