This is part of a research project to develop and test the efficacy of stochastic models for earthquake probability forecasts, in particular, the ETAS model. While such models show considerable promise, they still have many deficiencies. The ETAS model is specified by its intensity function (instantaneous rate), which can be displayed as a movie. Forecasts are made by utilising all recorded events up to the start of the forecast time interval, and then simulating the model over the forecast time interval. This is repeated many times (see title, pg 5 counts.pdf), enabling the calculation of an empirical representation of the forecast probability distribution (below).
File/Item | Description |
---|---|
Most Recent Forecast | 2025-02-10 18:00 (UTC) |
Current Forecast Length | 28 days |
Current Frequency | Every 2nd Monday at 18:00 (UTC)
Forecast online at approx 20:25 (UTC) |
Forecast Region | Rectangular Region: (165.5°E, 180°E, 47.5°S, 36°S), as above |
eda.pdf | Event Summary Over Previous 100 Days |
spatial.pdf | Forecast Spatial Distribution (low resol png above) |
counts.pdf | Distribution of Forecast Event Counts
M≥3.8, 4.8, 5.8 & 6.8, blue distribution on pp 1–4, respectively (low resol png M≥3.8 above) |
summary38.pdf | Summary and Evaluation of Previous Forecasts: CUSP magnitude scaled to be consistent with SeisComp3 (since 2019-07-14) |
summary40.pdf | Summary and Evaluation of Previous Forecasts: No magnitude correction for CUSP inconsistency (until 2021-02-15) |
Previous Forecasts | Before 2019-10-01: GNS sftp website
After 2019-10-01: Google Drive |
Data | An updated version of package ssNZ is made to make and again to assess each forecast, see details.txt |
Data Archive | Before 2019-10-01: GNS sftp website
After 2019-10-01: Google Drive |
Further details are given on the technical details page.