In recent years, continuous observations of gravity with time have been achieved by employing superconducting gravity meters with long-term stability and high precision. As a result, even small gravity changes associated with atmospheric loading are now detectable. It is, therefore, possible to obtain information about the structure of the earth from signals caused by atmospheric loading. An attempt to estimate a load Green's function for gravity was performed using the gravity changes associated with time-varying atmospheric pressure loads. Model parameters corresponding to discrete load Green's functions were, at first, estimated by applying the conventional least squares method, but those absolute values were too large to be acceptable. A similar estimation by applying the non-negative least squares method was then performed. As a result, one parameter obtained for the nearest region was fairly well consistent with the theoretically predicted one. However, the other parameters could not be determined. The cause of such a result may be noises included in gravity changes associated with atmospheric pressure variations. It was found by a numerical experiment that reduction of noises to the level of about 10-5 pgal would be needed to estimate the parameters precisely. But it is practically impossible to reduce the noise to that level. Thus, reduction of noises such as the influence of underground water which will be a major source of observational noise and a potent method of inversion are required for the improvement of the result. The present method can be employed to estimate ocean response to atmospheric pressure variations. Furthermore, it is possible to apply the present attempt to a case of oceanic loading, if the change of sea level is accurately determined.
View full abstract