TAVIs: Topographically Adjusted Vegetation Index for a Reliable Proxy of Gross Primary Productivity in Mountain Ecosystems

作者：Xinyao Xie, Wei Zhao, Gaofei Yin · 发表于：IEEE Transactions on Geoscience and Remote Sensing · 年份：2023 · DOI：10.1109/tgrs.2023.3336727 · 被引用次数：13 · 研究领域：Remote Sensing in Agriculture、Remote Sensing and LiDAR Applications、Plant Water Relations and Carbon Dynamics

Remotely sensed (RS) vegetation indices (VIs) are increasingly being employed as a direct proxy for gross primary productivity (GPP). When estimating mountain vegetation GPP from VI, efforts often focus on the RS-related topographic effect (i.e., distort VIs), while the micrometeorology-related topographic effect is so far ignored. Here, a topographically adjusted VI (TAVI) scheme was developed based on removing the RS-related effect by path length correction (PLC) first and integrating the micrometeorology-related effect associated with the topography-induced redistributions of radiation and water subsequently. The proposed TAVI scheme was applied to three VIs, namely, normalized difference VI (NDVI), enhanced VI (EVI), and near-infrared reflectance of vegetation (NIRv), at 14 eddy covariance (EC) sites. The determination coefficient (${R}^{2}$) and root-mean-square-error (RMSE) between VI-estimated and EC GPP were used for evaluation. Results showed that both EVI and NIRv outperformed NDVI in GPP estimation before correction, with${R}^{2}$increased by 0.14–0.15 and RMSE decreased by 0.42–0.44 gC$\cdot \text{m}^{-2}\cdot $day−1. After correcting the RS-related topographic effect, EVI and NIRv achieved an obvious improvement (${R}^{2}$= 0.71 and RMSE = 2.00 gC$\cdot \text{m}^{-2}\cdot $day−1), while NDVI showed little sensitivity to topography. Subsequently, EVI and NIRv showed a notable improvement (${R}^{2}$= ~0.77 and RMSE = ~1.82 gC$\cdot \text{m}^{-2}\cdot $day−1) after ...