A comparison of frequencies of large earthquakes between the observed and expected on the major active faults in Japan may suggest some underestimation of the occurrence rate of large events by the long-term earthquake forecasts. The main cause was thought to be the forecasts often anticipate a single large earthquake on a long fault zone, without assuming smaller earthquakes on segmented zones. However, the principal factor of the likely underestimation is that those events without enough evidence on fault zones cannot be evaluated by a trench excavation survey. Historical records show poor or no seismic surface rupture associated with a large earthquake with magnitude around 7 on some of the major fault zones.
Recent improvements on interpreting aerial photographs have brought a detailed pattern of active faults and fault segments. Together with information on underground structures obtained from geological and gravity anomaly maps, the detailed feature of active faults may enable us to decipher behavioral fault segments without solely relying on paleoseismic records.
Except for volcanic areas a short active fault on the surface should have a longer subsurface fault length within the earth's crust. The magnitude of earthquakes on such a fault is 6.9 or larger on JMA scale. This inference is based on reports on surface ruptures associated with large shallow crustal earthquakes in Japan. Only the 1945 Mikawa earthquake with magnitude 6.8 appears to be an exception to that no prominent surface rupture appears associated with earthquakes with magnitude 6.8 or smaller.