Retrospective Evaluation of the Five-Year and Ten-Year CSEP-Italy Earthquake Forecasts

摘要

On 1 August 2009, the global Collaboratory for the Study of EarthquakePredictability (CSEP) launched a prospective and comparative earthquakepredictability experiment in Italy. The goal of the CSEP-Italy experiment is totest earthquake occurrence hypotheses that have been formalized asprobabilistic earthquake forecasts over temporal scales that range from days toyears. In the first round of forecast submissions, members of the CSEP-ItalyWorking Group presented eighteen five-year and ten-year earthquake forecasts tothe European CSEP Testing Center at ETH Zurich. We considered the twelvetime-independent earthquake forecasts among this set and evaluated them withrespect to past seismicity data from two Italian earthquake catalogs. In thisarticle, we present the results of tests that measure the consistency of theforecasts with the past observations. Besides being an evaluation of thesubmitted time-independent forecasts, this exercise provided insight into anumber of important issues in predictability experiments with regard to thespecification of the forecasts, the performance of the tests, and the trade-offbetween the robustness of results and experiment duration. We conclude withsuggestions for the future design of earthquake predictability experiments.