Development of a topographic-corrected temperature and greenness model (TG) for improving GPP estimation over mountainous areas

摘要

The temperature and greenness model (TG) demonstrates that the combination of enhanced vegetation index and Moderate Resolution Imaging Spectroradiometer (MODIS) land surface temperature (LST) is feasible in obtaining gross primary productivity (GPP) at the landscape, regional, and global scales. However, the input LST data of TG is always available at a coarse resolution (similar to 1 km), averaging a relatively large portion of the topographic characteristics. Hence, GPP simulated using the coarse spatial resolution LST data would suffer from unavoidable bias over mountainous areas. Considering the above limitation, this work proposed a mountainous temperature and greenness model (MTG) through integrating an elevation-corrected factor and a radiation-corrected factor with the current TG model. The proposed MTG model was validated at sixteen eddy covariance (EC) sites with apparent topography in the carbon footprint areas. Results showed that MTG-simulated GPP presented a better agreement with EC GPP than TG-simulated GPP, characterized by an increase of 0.06 in R-2 and a decrease of 5.43 gC m(-2) 8d(-1) in root mean square error, suggesting that the MTG model had a better feasibility of capturing the GPP variations over mountainous areas than the TG model. The standard deviation of MTG-simulated GPP at the sixteen study sites varied between 3.29 and 22.79 gC m(-2) 8d(-1), highlighting the importance of considering topography within coarse pixels when obtaining GPP estimates over mountainous areas. Furthermore, results also indicated that the MTG-simulated GPP showed obvious responses to topography, suggesting that the MTG model could adequately characterize the topographic effects on plant photosynthesis. More specifically, MTG-simulated GPP increased when slope increased in the sunlit terrains, while it was found to have a lower value when slope increased in the shaded terrains. Our study suggests that incorporating topography information into current GPP models is a practical approach to improve GPP estimates over mountainous areas.