Large discrepancies among remote sensing indices for characterizing vegetation growth dynamics in Nepal
| Authors: | Decheng Zhou, Liangxia Zhang, Lu Hao, Ge Sun, Jingfeng Xiao, Xing Li |
| Year: | 2023 |
| Type: | Scientific Journal |
| Station: | Southern Research Station |
| DOI: | https://doi.org/10.1016/j.agrformet.2023.109546 |
| Source: | Agricultural and Forest Meteorology |
Abstract
Mountain ecosystems provide multiple ecosystem services and are “natural laboratories” to understandecosystem responses to global change. Because of the inaccessibility and the high cost of field surveys, remotesensing indices are the major and sometimes the only measures to monitor the vegetation growth dynamics inmountains. However, there are large discrepancies in those indices that should be quantified in mountainousregions. This case study in Nepal, a highly mountainous region, explores the consistency and inconsistency of sixwidely used remote sensing indices in monitoring vegetation growth from 2000 to 2020. The study considersthree greenness indices of normalized difference vegetation indices (NDVI), enhanced vegetation index (EVI),and near-infrared reflectance of vegetation (NIRv), one cover index of leaf area index (LAI), and two productivityindices of gross primary productivity (GPP) and solar-induced chlorophyll fluorescence (SIF). We find highspatial consistency in the multiyear means (r = 0.79~1, N = 4300, p < 0.01), especially in the highlands andbetween EVI and NIRv, and a logarithmic relationship between greenness indices or GOSIF and LAI or GPP. Incontrast, the long-term trends differ substantially by index and space. Only 7% of the lands show synchronizedsignificant increase though all the indices show a widespread increasing tendency (77~87% of the lands). Theprevalent non-significant changes of all the indices primarily contribute to the trend uncertainties, especially inthe highlands. The inconsistencies between greenness and productivity indices and in them further exaggeratethe uncertainties. Our results emphasize the large discrepancies of remote sensing indices in quantifyingmountain vegetation growth dynamics. Larger inconsistency is expected if we consider disparities among thequality-control schemes, study seasons, remote sensing models, satellite platforms, and sensors. Reinforcedremote sensing data, model improvements and/or new indices are needed for an accurate quantification of thevegetation growth dynamics in mountain regions.