2020 年 40 巻 2 号 p. 14-25
Seismic hazard maps play an important role in earthquake disaster risk reduction. The availability of spatial data is crucial to generate these maps that plot the spatial distribution of hazard potentials to emphasize spatial differences. The past few decades have seen an exponential increase in the availability of geospatial data. However, we cannot ascertain whether the amount of available data is sufficient, and we have no guidelines to draw the maps based on the available data consistent with the data accumulation. In this study, we address these issues in terms of data visualization techniques. Using information theory, we propose a parameter that measures the incremental information gain as maps are updated with new data over time. Data saturation occurs as the proposed parameter approaches zero. The concept is applied to a case study area in the Furukawa district of Japan where earthquake data has been collected over 7 years from 31 seismometers in a dense seismic array. Convergence in site amplification maps generated over different observation periods conclude that the mapping in Furukawa district is approaching data saturation and from the viewpoint of information theory, the current operation may be terminated.