The amount of positional difference is, usually, measured, using the positions between which the factor of correlation taken for small nearby corresponding areas in two pictures is the greatest. The altitude measurement precision is, therefore, directly dependent upon how accurately such picture correlation can be taken. In the case of ASTER, the difference in observation time between two pictures is as small as 55 seconds and this means that we see the earth surface twice almost at the same time free from any significant change in pictures due to optical conditions and atmospheric fluctuations, allowing us to expect a high-precision picture correlation. This picture correlation can be applied over entire pictures to prepare a digital elevation model (DEM) where the earth surface altitude is measured all over at a certain interval for each grid. It is also possible to extend DEM by preparing DEM for each scene and mosaically connecting neighboring scenes.

A DEM preparation is currently planned at an interval of 30m over the entire east Asian region. The east Asian region normally encompasses China, the Korean Peninsula, and Japan but, if simply calculated from the ground system's processing capacity (possible 30 scenes-equivalent DEM preparation a day), an Asian-wide DEM can be prepared in several years. A DEM now available by the public in the Asian region contains the data taken mainly by the U.S. FROS Data Center at an interval of 30 seconds (about 1km), suggesting how epochal ASTER's east Asian-wide 30m-interval DEM data is.

This DEM currently prepared for each scene is called a "relative DEM," subject to possible difference such as in position and scale, causing some deviation from geodetic coordinates on the ground, since its data are all automatically taken by computer, not using any reference ground points. It is, however, considered that this difference may be given as quite a small error of some 70m per 3 sigmas on the picture in terms of ground plane position due to the platform's high-precision postion/attitude information. In fact, a mosaic DEM, as shown in the figure, was prepared from simulation pictures at and around Mt. Yatsugatake obtained after、provided with possible position/attitude variables and compared with DEM of the Geographical Survey Institute, demonstrating the former's extremely high precision with its standard deviation of about 8m. Of course, a more extensive mosaic DEM could have cumulative errors, probably requiring reference ground points.

Conversely,however, the DEM preparation using a reference ground point for each scene is unrealistic due to unexpected sufficient such use in Asian region, we think that the most appropriate approach is to take DEM data for the Asian region, using reference ground points at main places. Thus, the above-mentioned DEM for the Asian region is promising as one of the most basic geophysical data for use in extensive areas.