← Back to Sessions

On the Use of Statistical Tests for Evaluating Spatial Earthquake Forecasts

Session: Constructing and Testing Regional and Global Earthquake Forecasts III

Type: Oral

Date: 4/22/2021

Presentation Time: 09:45 AM Pacific

Description: 

Spatial forecasts of triggered earthquake distributions have been ranked using statistical tests, such as the Receiver Operating Characteristic (ROC) test and the Collaboratory for the Study of Earthquake Predictability (CSEP) S-Test. The ROC test is a binary comparison between regions of positive and negative forecast against positive and negative presence of earthquakes. Forecasts predicting only positive changes score higher with the ROC test than Coulomb methods, which predict positive and negative changes. We hypothesize that removing the possibility of failures in negative forecast realms yields better ROC scores. We create a ‘perfect’ Coulomb forecast where all earthquakes only fall into positive stress change areas and compare with an informationless all-positive forecast (stress as a function of distance as r^-3). The ‘perfect’ Coulomb forecast barely beats the informationless forecast, and adding as few as 4 earthquakes occurring in the negative stress regions causes the Coulomb forecast to be no better than an informationless forecast under a ROC test. ROC tests also suffer from data imbalance when applied to earthquake forecasts because there are many more negative cases than positive. This imbalance also affects the CSEP S-Test, which normalizes forecasts, comparing rates at places where earthquakes occurred. A forecast that identifies places where earthquakes won’t happen (like Coulomb calculations) is not rewarded for being correct under the S-Test.

There are acknowledged issues with rapid, and/or simplified Coulomb stress changes being able to forecast the position of every triggered earthquake, and static stress changes are only part of the overall effect that a mainshock has on the stress field. However, Coulomb stress change calculations that account for the complexity in the mainshock rupture and the surrounding fault structures can provide resolution on where earthquake rates will be suppressed as well as where they will be enhanced.

Presenting Author: Tom Parsons

Student Presenter: No

Authors