Comparing Temporal Variations in LUE and GPP across Evergreen and Deciduous Forest Types

Abstract:

Estimating gross primary production (GPP) is an important goal of global change research. However, the relationship between GPP and its environmental drivers is highly complex and as a result, accurate modeling of GPP is difficult. One possible technique to help constrain the uncertainties is by using remote sensing data to try and determine the factors driving GPP directly from satellite imagery. In this study, we used GPP from flux data (GPP_EC) and meteorological observations of a deciduous (SOA) and a coniferous evergreen forest (DF-49) to optimize light use efficiency of sunlit (LUE_sun) and shaded (LUE_shaded) canopies. We based our analysis on the two-leave light use efficiency model (TL-LUE) at daily, 8 day, and 16 day scales by using the Markov chain Monte Carlo (MCMC). The photochemical reflectance index (PRI) of sunlit (PRI_sun) and shaded (PRI_shaded) leaves was calculated from spectral observations and related to tower based GPP at the three temporal scales. We found that the coefficient of determination (R²) between PRI_sun and LUE_sun, as well as PRI_shaded and LUE_shaded at the evergreen forest was lower than that at the deciduous forest. The modeled GPP was closely to the GPP_EC at the three temporal scales. The R² between the GPP_EC and modeled daily GPP was the highest when using daily measures of LUE, and lowest when uisng16-day LUE_sun and LUE_shaded. The results indicated that LUE is an important parameter when modeling instantaneous GPP and the short term variations of it. The results help to obtain a better understanding of how many satellite observations are needed to reliably constrain existing GPP models from remote sensing data.