Remote sensing of crop light use efficiency using photochemical reflectance index

Abstract

The photochemical reflectance index (PRI) has been suggested as an indicator of light use efficiency (LUE) for various plant functional types at both leaf and canopy scales. However, the application of PRI for the estimation of LUE has been always contaminated by canopy structures and directional effects associated with changes in illumination and viewing angles. The most important variable that controls the structural associated characteristics is the leaf area index (LAI), which is a major determinant of the absorbed photosynthetic active radiation (APAR) by a canopy. Using ground measured PRI and LUE over three growing seasons on winter wheat in China during 2005–2007, we found that canopy LAI overall explained 66% (p < 0.001) variance of PRI, indicating the essential of removing influences of external structural factors on PRI. Suggested by this, we defined the structural-related signal in PRI (sPRI) as a function of LAI and consequently, the residual PRI (rPRI=PRI-sPRI) would be independent on canopy structural characteristics. Our results showed that a mixed non-linear model using rPRI and variety and sampling time as fixed and random effects, respectively, had an improved accuracy in the estimation of LUE over PRI with increased R2 for both the overall dataset and data of each year. This finding is also supported when validated at flux tower site that the MODIS-derived daily rPRI can provide improved modeling of LUE (R2=0.49, p<0.001) compared with PRI (R2=0.23, p=0.011). These findings support the structural dependence of PRI and provide a solution for the removal of the structural signal.