Abstract
With respect to the observed astrogeodetic deflections of the vertical at 192 stations in Japan except Hokkaido, the topographic-isostatic eflections are estimated and then the error in the Japanese system of geodetic coordinates are examined in details. The computation of topographic-isostatic deflections is carried out at 152 grid points spaced at intervals of 30' in longitude and latitude, and the reduction to any assigned observation station is made by linear interpolation.
Our main results are as follows:
1. The topographic-isostatic deflection for each of both components in the meridian and prime-vertical is about 4."5 on the average of absolute values and about 18" at the maximum.
2. The most probable dopth of isostatic compensation layer derived from the astrogeodetic defiections is estimated to be about 90km all over Japan. Regionally, however, the compellsation depth in Northeast Japan is deeper than that in Southwest Japan, and further the depth in Northeast Japan is larger in the mountain areas in general.
3. As for the system of geodetic coordinates fitted to Japan, we obtain ζ0=-12.".39±0."17, η0=9."62±0."11, δa=59m±40m and f=1/(302.2±1.9) by taking the isostatic compensation into account. These figures show that the Bessel's ellipsoid is fitted fairly well to Japan.
4. By removing the topographic-isostatic deflection and the error in the system of geodetic coordinates, the above mentioned inclination seen among the observed astrogeodetic deflections of the vertical in Japan almost disappears except in special regions such as that the area between Itoigawa and Suwa in Chubu District, the neighborhood of Miyazaki City in Kyushu District, and the eastern and southern parts of Kanto District.
5. Glancing at the revised astrogeodetic deflections as a whole, we can see a general feature that changes in the deflections appear mainly in the the perpendicular directions to the principal tectonic lines in Japan.
6. The geoidal undulation from the surface of reference ellipsoid is in a range of 10m (from -5.4m to +4.6m) which is less than about 1m from the data without such corrections.
