Preliminary hyperspectral volcano observations using Airborne Radiative Spectral Scanner (ARTS)

Airborne-imaging spectral systems can often efficiently identify volcanic phenomena that are difficult to detect by satellite imagery. Since 1990, the National Research Institute for Earth Science and Disaster Prevention (NIED) has been developing our original airborne-imaging spectral systems for volcano observations. In 2006, we developed a new airborne hyperspectral sensor, the Airborne Radiative Transfer Spectral Scanner (ARTS), for hyperspectral volcano observations. ARTS is a push-broom imaging spectrometer covering wavelengths from 380 to 1100nm (VNIR; 288 bands), 950 to 2450nm (SWIR; 101 bands), and 8000 to 11500nm (LWIR; 32 bands) and has precise position and attitude measurement systems (GPS/IMU) to achieve direct geo-correction of the acquired image. The ARTS specifications were planned to provide hyperspectral images to support developing algorithms for remotely sensing the geothermal distribution, ash- fall areas, and content of volcanic gas columns. ARTS will also be useful for operational volcanic observations to assess volcanic activity and to mitigate volcanic disasters.Before beginning the operational use of ARTS, it is important to validate its in-flight performance. Therefore, we have been conducting validation on the B200 platform. In this study, we present the results of two experiment observations, the overflight of ARTS instrument at the NIED building site on April 5, 2007, and the volcano observations flight over active volcano (Sakurajima volcano) just after its eruption on April 8, 2008. At the NIED building site, we validated the radiometric fidelity of all bands and the accuracy of geo-corrections. At the Sakurajima volcano, we tried to demonstrate the functions of ARTS, especially those for volcano observation. At the NIED building site, the validation results indicate that the geo-correction accuracy is typically less than a two-pixel difference (RMS), and that there was good agreement between the predicted radiance at the sensor and the measured radiance at the sensor at a flight altitude of 1000m AGL. The percent difference of the radiance is typically 0.1 to 5 percent for VNIR and SWIR bands, and 0.1 to 2 percent for LWIR bands, except for the bands strongly affected by the atmosphere. At the Sakurajima volcano, the geo-corrected image was calculated directly using the data from the GPS/IMU system. This image can be superimposed onto the topographical map with sufficient accuracy for practical use. The trace area of the pyroclastic flow caused by the eruption can be measured using VNIR images. We could detect the geothermal activities of Sakurajima crater (Minamidake A-crater and Showa crater). The estimated maximum brightness temperature of Minamidake A-crater is 854 degrees C as measured from the radiance at 1001nm and 354 degrees C as measured from the radiance at 10260nm. The estimated maximum brightness temperature of Showa crater is 435 degrees C as measured from the radiance at 1625nm and 176 degrees C as measured from the radiance at 10260nm. These results indicate the existence of surface-temperature fields of subpixel resolution. Under these conditions, the shorter wavelengths of ARTS yield better maximum temperature estimation than the longer wavelengths. These results demonstrate ARTS"f ability to estimate temperature. In addition, inside the Minamidake A-crater area, the sulfur dioxide gas abundance could be estimated from the LWIR data. From these results, we conclude that ARTS is a well calibrated instrument for assessing volcanic activity and can be used for operational volcano observations.