日本地球惑星科学連合2019年大会

講演情報

[E] ポスター発表

セッション記号 H (地球人間圏科学) » H-TT 計測技術・研究手法

[H-TT20] Environmental Remote Sensing

2019年5月26日(日) 17:15 〜 18:30 ポスター会場 (幕張メッセ国際展示場 8ホール)

コンビーナ:Wei Yang(Chiba University)、近藤 昭彦(千葉大学環境リモートセンシング研究センター)

[HTT20-P02] Normalized Difference Greenness Index (NDGI): A new vegetation index for improved estimation of vegetation phenology

*楊 偉1小林 秀樹2近藤 昭彦1 (1.千葉大学、2.海洋研究開発機構)

キーワード:植生フェノロジー、リモートセンシング、雪、MODIS

Satellite monitoring of vegetation phenology is crucial to understanding the biological impacts of climate change. Nevertheless, satellite-based phenology estimation can be biased by effects of snow using conventional vegetation indices (VIs), such as the normalized difference vegetation index (NDVI). To eliminate effects of snow, snow-free VIs that take advantage of the shortwave infrared (SWIR) band have been proposed (i.e., phenology index [PI] and normalized difference phenology index [NDPI]). However, PI cannot properly capture the presence of sparse vegetation, and NDPI does not account for the influence of dry vegetation. Here, we propose a novel snow-free VI, designated the normalized difference greenness index (NDGI), using reflectance of green, red, and near-infrared (NIR) bands. NDGI is a semi-analytical index deduced based on the linear spectral mixture model and the spectral characteristics of vegetation, snow, soil, and dry grass. Its performance for estimating the start of growing season (SOS) and end of growing season (EOS) was evaluated using simulation datasets, time-lapse camera data of tundra sites, and flux tower gross primary production (GPP) data of grassland sites. Simulation results demonstrated that NDGI can exclude the influence of snow on SOS and EOS estimations. In the camera sites, NDGI markedly outperformed NDVI, PI, and NDPI for SOS estimation with RMSE=6.46 days and Bias=-1.32 days (vs. RMSE range 12.39–14.8 days and Bias>3.5 days for the other VIs), and for EOS estimation with RMSE=8.33 days and Bias=0.11 days (vs. RMSE range 8.00–26.74 days and Bias>2.67 days for the other VIs). In the GPP sites, NDGI also outperformed the other VIs for SOS estimation with RMSE=10.32 days and Bias=-4.91 days (vs. RMSE range 12.9–24.74 days and Bias>6.13 days for the other VIs). Although its performance was lower for monitoring EOS than SOS in GPP sites, it still outperformed the other VIs in EOS estimation. These results indicate the potential of NDGI for operational monitoring of plant phenology in the tundra and grassland ecosystems based on satellite observations.