Depth to bedrock in Japan: insights from borehole data and terrain analysis using digital elevation models

Abstract This study presents a statistical analysis of the depth to bedrock (DTB) in Japan, based on extensive borehole data and terrain attributes derived from 10 to 30 m grid Digital Elevation Models (DEMs). The borehole data, consisting of approximately 30,000 DTB entries, were organized by verifying borehole locations, selecting representative borehole sites, and excluding sites in major river outflow zones and artificially altered areas. The DTB was estimated by performing text searches and manual assessments within XML files of borehole data, resulting in a comprehensive dataset. Using DEMs for Japan, prioritized by elevation from LiDAR sources, geomorphologic classifications and terrain attributes—such as elevation, slope aspect, slope gradient, topographic position index, height above nearest drainage, and surface texture—were calculated and statistically compared with the DTB estimates. The following observations are tentative due to the exclusion of right-censored data and the bias in the borehole locations, although they reveal several key findings. In mountainous and hilly areas, median DTBs were approximately 3–4 m. A strong power-law relationship (R 2 = 0.87) was observed between medians of DTB and slope gradient across different terrain categories. In the mountains, particularly for sedimentary rock, DTB was deeper in layered rocks compared to that in massive rocks. In hilly regions, DTB was shallower on lower slopes, where shallow landslides are more common, potentially due to landslide influence. Generally, the upper DTB limit inversely correlated with slope gradient, although in mountainous areas this upper limit decreased slightly in gentler slopes, suggesting a possible landslide effect. At the same slope gradient, sites with shallower DTB were more frequent. Despite the variability in the relationship between topography and DTB at specific locations, further statistical characterization is expected to enhance future modeling.