PRISM geometric Cal/Val and DSM performance
Junichi Takaku(1) and Takeo Tadono(2)
(1) Remote Sensing Technology Center of Japan, (RESTEC), Technical Development Department, Japan
(2) Japan Aerospace Exploration Agency, (EORC/JAXA), EO Research Center/Japan Aerosp Exploration Agency, Japan
The creation and updating of maps on a scale of 1:25,000 as a global mapping scheme are the one of the goals of the PRISM mission and it needs high resolution and accurate Digital Surface Model (DSM) and its derived ortho- images on a global scale. According to this aim, the accuracies and stabilities of PRISM geometry have been widely assessed and calibrated as the Calibration and Validation Science Team (CVST) activities organized by EORC/JAXA since the ALOS launch.
First, the geometric model accuracies and stabilities of PRISM are presented with the latest results of calibrations and validations in EORC/JAXA. ALOS equips GPS receiver for satellite orbit determinations and star tracker instruments for satellite attitude determinations to automatically fix exterior orientation parameters of PRISM geometry without any Ground Control Points (GCP). Those automatic geo-location accuracies are important for almost areas which have no GCP in the world. The calibration details of those exterior parameters and current issues are discussed with the validation results of GCP site scenes for various areas and various data periods. The triangulation accuracies of PRISM triplet stereo images are estimated following the geometric model accuracies from the traditional photogrammetric point of view.
Next, the latest performance analysis of PRISM-DSM generated from the latest calibrated PRISM geometric models is presented. PRISM consists of independent three stereo-viewing panchromatic radiometers for forward, nadir, and backward and acquires each image in the same orbit at almost the same time. This unique triplet stereo observation enables to generate more precise DSM data frequently than usual pair stereo observations. For this specific stereo image configuration, the correlation based exclusive triplet images matching algorithm is being applied for the PRISM-DSM generation in EORC/JAXA. This algorithm enables the higher robustness against the matching blunders due to various image conditions (occlusion, clouds, etc.) than the conventional pair stereo matching. The reference DSM data sites were provided by CVST activities and used for the validations of PRISM-DSM. Those reference DSM data sets derived from airborne LiDAR instruments and aerial photo image matching have high ground resolutions (up to 1m) and height accuracies (up to 0.5m) at the various terrain futures. The absolute and relative accuracies of PRISM-DSM generated with and without GCP are presented with the comparison from those reference DSM data.