The Sensitivity Analysis of Anisotropic and Atmospheric Effects on NDVI Estimation using AHI Data

Currently, various types of reflectance are used to calculate NDVI in various researches and institutions.

However, there is little research on the sensitivity analysis according to the kind of reflectance used in the calculation.

Therefore, in this study, the vegetation index is calculated by using the Himawari-8 / Advanced Himawari Imager (AHI) data

and We will analyze the sensitivity to atmospheric correction and anisotropic effects.

We used the Second Simulation of a Satellite Signal in the Solar Spectrum (6S) Radiative Transfer Model (RTM) to at atmospheric correction and kernel-based semi-empirical BRDF modeling in order to remove the anisotropic effect of the surface.

Based on these results, TOA NDVI, TOC NDVI, and Fixed Viewing BRDF-adjusted reflectance (FVBAR) NDVI were produced. We conducted a sensitivity analysis on three products.

All three types of NDVIs exhibited similar spatial distribution.

However, there were different trends depending on land type and the vegetation vitality in the area.

TOA NDVI was estimated to be about 0.6 in underground vegetation area. This result tends to be underestimated compared to other NDVIs.

On the other hand, in the area with low vegetation distribution, TOC NDVI was very high compared with other NDVI.

In addition, TOA NDVI and TOC NDVI calculated the missing value in specific sections. However, for FVBAR NDVI, the least observable area by cloud was the least, the most consistent value in time series was calculated compared to other NDVI. This result tends to be underestimated compared to other NDVIs.