Four elements are required to estimate the potential losses from any one peril: the hazard itself, its probability of occurrence, the vulnerability of the risk, and the spatial distribution and aggregation of risk.

		Aerial view of Valdez, Alaska, showing the extent of inundation along the coastline following 1964 tsunami.

Once understood, a model may be created which estimates the loss from any one event. By repeating the exercise thousands of times and recording the loss and the probability of it arising, a graph can be constructed showing how often a loss of X is expected to exceed a value of Y which provides the basic information from which any insurance or reinsurance can be technically priced.

In this section we give an overview of how it is possible to quantify future losses and the complexities of modeling it. Currently, it is not possible to model tsunamis on a global basis because the extremely large amount of data necessary to do so doesn't exist. Furthermore, as it would take an extremely long time to develop and compute an answer even using a super-computer, we have chosen to demonstrate the principles by looking at a small region of Japan. Japan is one of the most densely populated countries on earth and is sighted on a major plate-tectonic junction. It is subjected to numerous earthquakes every year, some of which have caused extreme loss to life, property and infrastructure. In 1923 Tokyo was flattened by one of the greatest earthquakes in recent history to effect a heavily populated developed country. The Kobe earthquake of 1995 caused $3 billion worth of damage which would now cost insurers dearly as the Japanese insurance market for earthquake risk has recently opened up.