 |
||
 |
Earthquake Predictions
|
|
'City at dusk' © hywit dimyadi, iStockphotos Inc. All rights reserved
WESTERN JAVA - SUNDA STRAIT REGION
is newly defined (09-03-2009) as all areas within ≈450 km radius* of a center in Sunda Strait at Lat -5.869° and Lon 105.762°. This region covers western Java, the Sunda Strait, and adjoining areas of Southern Sumatra. Since 1975 this region has experienced 50 strong earthquakes (M6.0-M6.9) and 6 major earthquakes ranging from magnitudes 7.0 to 7.9
(USGS/NEIC & Google Earth sources)
*The radius is 450 km plus an ≈10% error margin of about 45 km to include some events that may occur close to the nominated boundary of the region.
All earthquake predictions for the Java Region are for the
largest earthquake in the nominated period, plus an error margin of about one day, but no attempt is made to make forecasts for all periods because of the constraints of private research. Each table, with its original earthquake prediction, will remain subject to revisions or updates for 10 days.
Why updates may be necessary... because earthquakes interact in a very dynamic system. Just as in reporting the weather, the latest information must be taken into account to improve forecasting.
_____________________________
SCROLL DOWN FOR TABLES
_____________________________
STATISTICS FOR EARTHQUAKE PREDICTION For period 13 August 2007 - 12 July 2008 (256 predictions)
SUCCESS RATES*
Mag Observed Events Predicted Events
(How many predicted?) (How many observed?)
Pred. Obs. %Correct Obs. Pred. %Correct
<M4 60 67 = 90% 58 66 = 88%
M4.0-4.9 76 85 = 89% 57 60 = 95%
M5.0-5.9 73 77 = 95% 90 96 = 94%
M6.0+ 25 27 = 93% 29 34 = 85%
___________________________________________ *These statistics are updated periodically and refer to tabled predictions only, as no attempt
is made to predict all large earthquakes worldwide. Note that currently successful predictions are those which are ≈80% accurate on magnitude, and within a 10-day time window (plus an error margin of about one day), reflecting the fact that earthquake prediction is still an imprecise science. However, research here is aimed at improving this margin and currently 75% of all magnitude predictions have an error margin of only 15%. Any apparent numerical discrepancies between the two halves of the above table are due to the error margins: for example, an M6 category predicted quake might have come in at M5.8
OVERALL success rate from 256 predictions worldwide-
Observed events that were predicted (234/256) = 91.4%
Predicted events that were observed (234/256) = 91.4%
SCIENTIFIC: Correlation of Predicted & Observed Magnitudes
There is a highly significant correlation (r=0.90, P< .0001) between the full set of observed and predicted magnitudes, including 'false alarms' (Worldwide N=256 as of 12 July 2008).
EARTHQUAKE PREDICTION TABLES
M= Earthquake magnitude. See HOME page for further explanation of table entries.
Earthquake Location Maps - see the USGS web site for maps of earthquakes in these tables.
New 3-layer Table Format & Prediction Rules (25 Nov 07)
Please read following points to fully explain tables below:
1. The Original Earthquake Prediction, and details relating to it, are given in the top light blue layer of each table.
The 'expected date' (or dates) given here provide the best current estimate of when the largest earthquake event is most likely within the nominated 10-day window (date prediction made +10, or possibly ≈11 days). Please note that the time window now includes an error margin of ≈1 day which may extend it to include some closely predicted events, reflecting the fact that earthquake prediction is an imprecise science. The above estimates and other details may be updated, if necessary, in the next section of the table.
The Expected Range in Magnitude, for each event after 14th July 2008, is given as the predicted magnitude +/-1 magnitude (previously it was given as +/-0.5M). Statistical analyses of over 100 of the latest predictions show that 95% of them can be expected to fall within +/-1M of the predicted values, regardless of the region of the world involved. From 14th July 2008, any predictions which fall outside these limits by as little as 0.1M will be regarded as unsuccessful.
2. Revisions, if required, are given in the mid blue layer of each table, and these revisions are used in the final assessments of the results. Please see the statement above explaining why updates are desirable, and note that revisions are limited to a 10-day window. Revisions usually relate to
earthquake magnitude and expected timing, but may also include location.
3. The Observed Earthquake and details (for the event we were attempting to predict) are given in the bottom dark blue layer of each table. The event recorded here is the largest earthquake for the nominated 10-11 day period. 'PENDING...' simply means that the anticipated event has not yet happened, or that we are waiting for the event to be posted on the web site.
JAVA REGION
See also Long-term Forecasts
These Long-term Forecasts now give the approximate timing of the largest regional earthquakes for the lunar period ahead.
The Short-term tables below will be used to make revisable and, generally, more precise earthquake predictions for some of the important events that are indicated as being imminent in the Long-term Forecasts.
No. 12
PREDICTED M5.0 Expected Date/Location Original prediction 15th April 15-24 April (UTC) 2009 Predicted at 1010 hrs UTC (time window open to 24th April +1 day margin) Probability of event >90% within ≈500 km of Semarang
|
|
|
OBSERVED M4.3 Observed Date/Location Magnitud accuracy 86% 4 days late - 29th June Lat -9.75 Lon 108.39 (EMSC source) South of Java |
No. 11
PREDICTED M5.0 Expected Date/Location Original prediction 31st Jan 6th February +/-3 days (UTC) 2009 Predicted at 1122 hrs UTC (time window open to 9th February) Probability of event >90% within 500 km of Semarang
|
|
| OBSERVED M4.4 Observed Date/Location Magnitude accuracy 88.0% 31st January (1216 hours) Predicted one hour ahead Lat -8.92 Lon 109.64 232 km SSW of Semarang, Java (USGS/NEIC & Google Earth sources) South of Java |
No. 10
PREDICTED M5.0 Expected Date/Location Original prediction 11 January 14 January +/-3 days (UTC) 2009 Predicted at 0630 hrs UTC (time window open to 21 January) Probability of event >90% within 500 km of Semarang
|
REVISION nil |
| OBSERVED M5.0 Observed Date/Location Magnitude accuracy 100% 20th January at Lat -8.37 Lon 112.07 Predicted 9 days ahead 35 km S of Blitar (EMSC source) Java, Indonesia |
Previous tables have been deleted to save space