Simulating Spatial Correlation for Catastrophic Events

13 Sep 2016

Catastrophe models calculate the stochastic distributions of loss originating from events like hurricanes and earthquakes. These models are typically based on a stochastic event catalog. For each event spatial correlation needs to be simulated. The standard approach is based on the evaluation of a copula. However the complexity of the corresponding algorithm is O(n3) and it becomes difficult to execute for n in the thousands.

So as the number n of locations in the footprint of the event grows, this approach quickly becomes infeasible. We propose a slight modification of the well-known Kriging technique, in order to solve this problem. With our solution the creation of simulation data for catastrophe models becomes manageable with the use of Big Data techniques.

Please contact your underwriting representative to discuss these or other Research topics further.