Debris classification based on detailed spectral observations using micro-satellite

Remote sensing observations from satellites have the great advantage of surveying large areas in a short time. On the other hand, the pixel size of satellite-borne camera on the ground is larger than that of drones or ground-based measurements, making it difficult to classify types of litter based on their detailed shape. Detailed spectral measurements using hyperspectral cameras are expected to be effective in classifying plastics and wood floating on the ocean, or litter accumulated on the beach, from vegetation, sand and stones, but the typical ground resolution of existing satellite-borne hyperspectral cameras is about 30 m. It is not easy to discriminate between types of litter and other objects. We have established imaging technology with a bandwidth (FWHM) of 10-20 nm, 1 nm steps at the centre wavelength and ground resolution of 5-120 m in the 0.4-1.0 micron wavelength range by developing and operating a 50 kg class micro-satellite equipped with a liquid crystal tunable filter (LCTF). In order to capture plastic features, it is necessary to observe even longer wavelength ranges. Currently, by developing a new spectral camera and satellite attitude control technologies, we plan to achieve a bandwidth of less than 10 nm and a ground resolution of about 10 m at 0.4-1.6 um. It is also important to build up a spectral library of spectra for different types of litter and plastics based on ground-based measurements. In this presentation, we report on the development of our micro-satellites and on-board cameras, as well as the methodology and status of the construction of the spectral library.