Permalink : http://hdl.handle.net/2115/38158

JaLCDOI : 10.14943/gbhu.72.257

A feasibility test of a kriging method is conducted for estimate of a strain field. In this test the kriging method is applied to man-made data and actual ground displacement data observed in Taiwan. It is investigated how well the method can reproduce the given strain field and extract the complex strain field of Taiwan. To clarify the merits/demerits of this method, we compare the results of the kriging to those inferred from other analysis methods; one is a conventional method in which a strain is directly calculated using a triangulation network and the other is a method incorporating a spatial smoothing procedure. The results demonstrate that the kriging method enables us to provide a largescale strain field spatially smoothly, and further to simultaneously extract a localized deformation. Additionally it is noteworthy that this method is robust for anomalous data, i.e., weights of anomalies are suppressed automatically. On the other hand, the standard triangulation method cannot stably extract a large-scale strain field and at a certain case artificial strain errors are produced. This method is very sensitive to anomalous data and/or even to small fluctuations (errors) of data. The method incorporating a spatial smoothing procedure that has been developed and applied in many recent studies can robustly evaluate strain components that are spatially dominant in the analyzed domain, but miss out a small-scale deformation due to the smoothing.