Application of the photosynthetic light-use efficiency model in a northern Great Plains grassland

The light-use efficiency (LUE) model of photosynthesis is widely used to estimate ecosystem photosynthesis and net primary production from remote sensing measurements. The fraction of absorbed photosynthetically active radiation (fAPAR) is a dominant term in this model, and it is fundamentally important for model calculations of ecosystem productivity across large areas. The LUE term is sometimes considered a constant, but may be best represented as a variable scalar under stress conditions. The main objective of this study was to better understand factors influencing fAPAR, its relationship with seasonal variation in canopy greenness (Normalized Difference Vegetation Index (NDVI)), and the consequences of potential seasonal changes in the NDVI-fAPAR relationship for LUE model calculations of ecosystem photosynthesis in a semi-arid grassland. We used two approaches to determine fAPAR, (i) direct incoming and outgoing radiation measurements above and below the canopy, and (ii) an inversion approach based on incident photosynthetically active radiation and the light response curve of net ecosystem productivity measured by eddy covariance at low light levels. The two approaches resulted in fAPAR values that were very strongly correlated during the initial development of the canopy until peak leaf area index (LAI) was reached. During this time, a strong linear relationship also occurred between fAPAR and NDVI calculated from spectral reflectance measurements of the grassland canopy. After peak LAI, there was hysteresis in the NDVI–fAPAR relationship, and the two fAPAR estimates diverged. Light-use efficiency model calculations of ecosystem photosynthesis made using fAPAR values were strongly correlated with chamber CO2 exchange measurements during the initial development of the canopy leaf area. After peak LAI, a stress function, based on either soil moisture or vapour pressure deficit (VPD) measurements, was necessary to reduce quantum yield and model calculations of ecosystem photosynthesis during periods of relatively low soil moisture and higher VPD later in the growing season. Both stress functions were similarly effective in improving the correlation between modeled and measured ecosystem photosynthesis values, and indicated reduced LUE under late season conditions. Modulating LUE based on the Photochemical Reflectance Index or the Water Band Index (both proposed as possible indicators of LUE) was not effective to improve the correlation between modeled and measured ecosystem photosynthesis values in this ecosystem.