Drolet, GG, Huemmrich, KF, Hall, FG, Middleton, EM, Black, TA, Barr, AG, Margolis, HA (2005). A MODIS-derived photochemical reflectance index to detect inter-annual variations in the photosynthetic light-use efficiency of a boreal deciduous forest. REMOTE SENSING OF ENVIRONMENT, 98(3-Feb), 212-224.
The relationships between ecosystem-level light use efficiency (LUE) obtained from an eddy covariance flux tower and MODIS-derived values of a scaled Photochemical Reflectance Index (sPRI) were investigated for a boreal aspen stand (Populus tremuloides Michx.) in Saskatchewan, Canada. Using MODIS ocean band 11 at 531 nm, we tried to detect variations in canopy reflectance related to the xanthophyll cycle. We tested several other MODIS bands as the reference band because the 570 nm reference band that had been determined to be optimum for calculating PRI in earlier studies is not available on MODIS. While LUE varied greatly within the 2001, 2002, and 2003 growing seasons, the Normalized Difference Vegetation Index (NDVI) calculated from tower sensors remained stable. LUE had a negative exponential relationship with vapor pressure deficit and air temperature, and was at a maximum when absorbed PAR was <200 mu mol m(-2) s(-1). The range and magnitude of tower-based LUE values were smaller on clear days, when MODIS acquisitions were possible, than they were overall. Furthermore, the orbital parameters of the Terra and Aqua satellites restricted MODIS acquisitions to a 2.5-h period in early afternoon at out study site when LUE values were typically lower than they were earlier in the day. Strong correlations between MODIS-sPRI and LUE were found only for backscatter reflectance scenes when band 13 (667 nm) was used as the reference band. The correlations were higher for sPRI calculated from top of atmosphere reflectances than for surface reflectances (r(2) = 0.76 and 0.53, respectively). The absolute backscatter reflectance of bands 11, 12 (551 nm), and 13 all decreased with increasing LUE. The decrease in band 13 suggests that the correlation between sPRI and LUE that we observed was caused by reductions in canopy chlorophyll content from 2001 to 2003 and/or by increased visibility of brighter non-photosynthetic material. Regional values of sPRI from 260 deciduous forest pixels in the 10,000 km(2) Vicinity of the tower for two contrasting days, one in 2001 and one in 2003, were consistent with that observed for the flux tower footprint. (C) 2005 Elsevier Inc. All rights reserved.