日置幸介, カルマンフィルターによるＶＬＢＩデータ解析
測地学会誌、 33, 337-345, 1987.
測地学会誌に投稿した最初の論文。測地学会誌には当時まだ査読がありませんでした。 当初英文で書きはじめたのですが手違いで書きはじめた文書ファイルが消えてしまい、あら ためて日本語で書いたもの。鹿島支所内で衛星管制課を中心に行われていたカルマン・フィ ルターの教科書の輪講に参加して得た知識をVLBIのデータ解析に応用したもの。大気と時計 オフセットをランダムウォーク過程で観測エポック毎に推定しています。
The final stage of the geodetic VLBI data analysis is the adjustment of various
"stationary" parameters such as station positions and additional "non-stationary"
parameters such as the inter-station offsets of atomic clocks and stmospheric zenith
path delays. With the conventioal least-squares estimation method, complicated
techniques have been needed to estimate the parameters with temporal variation.
For example, the variation of clock offsets is usually assumed to be represented by
a set of parabolic curves connected by several discontinuities.
With a new parameter estimation method based on Kalman filtering technique, both
the deterministic estimation of stationary parameters and the tracking of the
values of non-stationary parameters varying according to some stochastic process are
possible. This enables the VLBI parameter estimation without operator-dependent
procedures to modelize such temporal variations. It is also shown that the higher
repeatability of the baseline lengths is obtained by Kalman filter method than by
the conventional least-squares method.