Reliable land subsidence mapping using a spatial interpolation procedure based on geostatistics

This paper describes a spatial interpolation procedure for settlements to increase the reliability of subsidence mapping. For this study, a spatial interpolation method based on the ordinary Kriging method was investigated for applicability to a land subsidence area in the northern Kanto plain. First, the spatial variability of settlements using a semivariogram was investigated in detail. For a short distance, the semivariogram of settlements increases. However, after the maximum value at about 35 km, the semivariogram decreases because settlement observation locations are typically arranged around an area in the center of a severe land subsidence zone. Therefore, the semivariogram is only useful up to that maximum value for investigating spatial variability of settlements in a land subsidence area. A spatial interpolation procedure based on Kriging method is proposed. Applicability of that procedure is investigated through comparison of estimations with observations. Drawing both estimations and standard deviations of those estimations represents a reliable land subsidence map. Moreover, to improve the map's accuracy, new allocation for monitoring settlements is undertaken by installation in areas where the estimation standard deviation is greatest. The effect can be investigated using simulated results in the distribution change of the estimated standard deviation.