Long range earthquake forecasting moves a step closer as forecast model demonstrates significant forecast skill with 95% confidence over a 151 day period. At day 151 the QS Earthquake Forecast Mission is on track to be the first “statistically significant” long range earthquake forecast ever produced. With 26 out of 49 major earthquakes forecast exactly and 96 days out of 151 days correctly identified as either having major events or not, QSM I is now entering the final critical phase.
The Mission:
Quakescanner Mission I (QMI) is a six month mission that seeks to understand and model the affect of gravitational forces and space fabric distortions on our planet, not by changing gravity science but by modelling space (from which gravity emerges) itself - space-fabric! Using new mathematical models of the solar system that treat space like a special kind of liquid called "superfluid", the QMI project is forecasting the days it identifies as most at risk to determine if the model can be used to bring us a step closer to understanding what causes these deadly events, and more importantly if they can be predicted.
At day 151 the QS earthquake forecast is running within the 95% confidence requirements. The number of significant earthquakes forecast exactly was 26 out of 49 and over the 151 days, 96 days were forecast exactly.
What does statistically significant mean?
Well, love them or hate them statistical significance is the general measure by which theories, drug studies and all manner of scientific aspirations live or die by. In general, what statistics try and relate to the reader is a confidence level of how likely the observed outcome of a test (be it an opinion poll, a weather forecast or the performance of a drug) is actually the result of actual influence or if the result could have been down to sheer luck. In the case of the Quakescanner Long Range Earthquake Forecast Mission, we are testing if the forecast results are more or less likely to be a result of space phenomenon such as gravity disturbances (which the experiment seeks to prove) or if the results obtained might as easily be obtained by a simple guess.
The most accepted test of significance (i.e the forecast has some skill at predicting earthquakes) is the p-value. The p-value for 95% confidence is set at 0.05. If the statistical performance of the forecast is greater than 0.05 then the results “could” be down to luck. If the result is less than 0.05 then we consider that the forecast model has significant skill and by connection, the model that created the forecast cannot be discounted as a plausible theory of how the forecast system behaves.
So how is the mission doing?
At day 151, the mission is currently on track to deliver a statistically significant result. The table calculates several statistical metrics including Fisher Exact Test, Chi-Squared and Odds Risk / Risk Ratio. The Fisher exact test p-value in particular is often considered the 'gold standard' of statistical measurements. The 95% confidence target requires a p-value < 0.05. The Fisher exact test p-value currently attained is 0.0126, which is well within the margins for statistical significance. It is valid to say that with 95% confidence the forecast is delivering results based on skill and not chance.
What does this all mean then?
With just 31 days of mission time remaining the goal of attaining statistical significance over the entire 6 month forecast period hangs on the performance of the New Year forecast which will be issued on December 31st. If by that time the result still upholds the theory then it will be both the first time that a long range earthquake forecast has successfully forecast earthquakes with any degree of reliability and the first acceptable evidence that relates the planets of our solar system to earths climate and seismic activity.
Check out the full mission page for latest information here
