Can ALOS-3/HISUI detect seaweed beds more precisely than ALOS/AVNIR-2?

Abstract

Hyperspectral Imager Suite (HISUI) is a Japanese future spaceborne hyperspectral sensor system. It will be launched in 2015 or later as one of mission instruments onboard JAXA's Advanced Land Observation Satellite 3 (ALOS-3). HISUI will consist of a hyperspectral imager and a multispectral imager with 30 m and 5 m spatial resolution and 30 km and 90 km swath, respectively. In order to characterize capability to detect seaweed beds of HISUI multispectral data with 5 m spatial resoution, we comprared classification results between ALOS/AVNIR-2 with spatial resolution of 10 m and simulation data of HISUI produced from WorldView-2. Study site was seletced in Oita Prefecture in Kyushu Island, Japan, where seaweed beds were broadly distributed along the coast. We used AVNIR-2 data taken on 20 February 2007 and simulation data produced from WorldView-2 taken on 18 April 2010. Supervised and unsupervised classifications were applied to these data sets. Miss-classification of analysis using AVNIR-2 was identified in deep waters in offshore waters. Since radiometric resolution of AVNIR-2 is 8 bits, it is considered that low radiometric resolution causes missclassification. Spatial resolution of HISUI multiband data with 5 m permits to detect seaweed beds clearer. Precision of classification using HISUI simulation data was about 90% and 10% higher than that using AVNIR-2. Thus HISUI multiband data are suitable for mapping seaweed beds in coastal waters.