A008-0026
Contrasting diurnal and seasonal Advanced Himawari Imager vegetation index patterns with land cover type in Australia

Monday, 7 December 2020
Poster
Ngoc Tran1,2, Alfredo R Huete2, Natalia Restrepo-Coupe3, Song Leng2, Qiaoyun Xie2, Yelu Zeng4 and Tomoaki Miura5, (1)Hanoi University of Science and Technology, School of Information and Communication Technology, Hanoi, Vietnam, (2)University of Technology Sydney, Faculty of Science, Ultimo, NSW, Australia, (3)University of Arizona, Department of Ecology & Evolutionary Biology, Tucson, AZ, United States, (4)Pacific Northwest National Laboratory, Joint Global Change Research Institute, College Park, MD, United States, (5)Univ Hawaii, Honolulu, HI, United States
Abstract:
The Advanced Himawari Imager (AHI) on board the Himawari-8 geostationary (GEO) satellite provides sub-hourly reflectance and vegetation index observations raising the possibility of improved monitoring of highly dynamic ecosystems. However, since AHI has unfamiliar diurnal and seasonal observation geometries, not experienced by MODIS or VIIRS, VI patterns and issues resulting from AHI observations need to be investigated. Diurnal reflectance and greenness profiles are of interest in BRDF studies, hotspot influences, alignment with flux tower data, and in determining optimal methods for daily composited products. In this study, we investigated diurnal variations of surface reflectances, NDVI, and EVI at different seasonal periods and across a wide range of forest, woodland, grassland, and crop land cover types. Our aim was to better understand diurnal and seasonal VI time series acquired from unique AHI backscatter and small phase angle observation geometry. Our results showed large diurnal variations in reflectance and VI values with large contrasts across land cover types. The NDVI and EVI profiles were diurnally negatively correlated, with relationships that varied with land cover type and with phenological period of the year. Grassland/ cropland diurnal EVI behaved distinctly to woodlands, with forests exhibiting solar noon EVI peaks, while grasslands showed a solar noon EVI troughs or no profile at all. On the other hand, NDVI values were lowest at solar noon in all vegetation types. This was attributed to a negative NDVI backscatter and hotspot effect, while EVI showed positive backscatter effects and positive to no diurnal hotspot effect. The diurnal patterns encountered were very useful in generating daily composite for seasonal analyses. We found the seasonal AHI VIs also exhibited strong backscatter effects and hot spot influences around spring and autumn equinox periods. The diurnal patterns also offered opportunities for tracking diurnal courses of carbon dioxide exchange, such as gross primary production from flux tower sites.