JpGU-AGU Joint Meeting 2020

Presentation information

[E] Oral

A (Atmospheric and Hydrospheric Sciences ) » A-CG Complex & General

[A-CG53] Terrestrial monitoring using new-generation geostationary satellites

convener:Yuhei Yamamoto(Center for Environmental Remote Sensing, Chiba University), Yunyue Yu(NOAA National Environmental Satellite, Data, and Information Service), Tomoaki Miura(Univ Hawaii), Kazuhito Ichii(Chiba University)

[ACG53-05] Analysis of differences between NDVI-based indices of Himawari 8 AHI and Aqua MODIS using Rayleigh-corrected reflectances

*Kenta Obata1, Hiroki Yoshioka1 (1.Aichi Prefectural University)

Keywords:Himawari 8, MODIS, NDVI, FVC

A new generation of geostationary (GEO) satellite sensors, including the Himawari 8 Advanced Himawari Imager (AHI), has been launched in past five years. These sensors have been widely used for monitoring spatial and temporal changes of terrestrial vegetation with high temporal resolution. Data from polar-orbiting, low earth orbit (LEO) satellite sensors such as the Moderate Imaging Spectroradiometer (MODIS) can be used to complement observations made using the new-generation GEO sensors; for example, LEO data obtained prior to the launch of GEO sensors can be used to create long-term data sets. Evaluation of the consistency of GEO and LEO products is thus crucial for the integrated application of GEO and LEO sensors.
A major issue in comparing GEO and LEO products is that the ray-matched condition, in which the illumination and viewing angle condition is nearly identical, does not appear in middle to high latitude regions. The mismatch of the angular condition and effects of surface anisotropy, i.e., bidirectional reflectance distribution function (BRDF), cause biases in reflectances of sensors that propagate into downstream products such as the Normalized Difference Vegetation Index (NDVI). In addition, NDVI data may differ between sensors due to differences in their spectral response functions.
In this study, we scrutinize an index related to greenness, specifically, the fraction of vegetation cover (FVC), and examine its consistency between AHI and Aqua MODIS (near-nadir viewing) for land surfaces in mid-latitude regions. The FVC is referred to as an “NDVI-based index” as it is a function of NDVI and can be less than 0 or greater than 1. Previously, we developed an algorithm that, using top-of-atmosphere (TOA) reflectances as input data, makes the output (NDVI-based index) consistent between sensors in order to mitigate effects of viewing zenith angle (nadir and off-nadir) and spectral response functions. However, the spatial heterogeneity of the atmospheric layer arising from Rayleigh scattering, aerosol loading, ozone and water vapor contents added uncertainties when comparing NDVI-based indices.
Here, as the next step in our studies, we use Rayleigh-corrected (RAC) reflectances to evaluate differences between NDVI-based indices from AHI and Aqua MODIS. Results indicate that NDVI-based indices using RAC reflectances show more consistent results relative to NDVI using RAC reflectances. Additionally, absolute differences between RAC NDVI-based indices of two sensors are similar to or slightly smaller than that between TOA NDVI-based indices. Finally, the benefit of NDVI-based index for transforming NDVIs between sensors is introduced for comparing/integrating observation data obtained using AHI and MODIS for monitoring terrestrial vegetation.