Goals and Components
System Documentation
Download Code
QuakeTables Resources
People
Publications
Related Links
FORECAST/Scorecard
Animations
Pictures
QuakeSim Library
(authorized users only)
|
The "Scorecard" is a forecast map, or "seismic hotspot" map, for the period of time January 1, 2000 - December 31, 2009. The colored areas (hotspots) are found by a method that computes the increase in potential for large (magnitude M > 5.0) earthquakes over the years January 1, 1990 - December 31, 1999. Research has shown that this procedure locates areas (hotspots) that are sensitive not only to the large events that actually occurred between January 1,1990 - December 31, 1999, but also to the large events that will occur over approximately the next ten years. The green triangles in the plot represent the large events that did occur during January 1,1990 - December 31, 1999. The blue circles are the large events that have occurred since January 1, 2000. As the scorecard indicates, 22 of these 25 events occurred after the scorecard was first published in Proceedings of the National Academy of Sciences on Februrary 19, 2002. The scorecard as published in PNAS used only the data from the Southern California catalog, with selected additions from the Northern California catalog.
Scorecard using original catalog
Computed July 29, 2008
[download a pdf version]
The figure on the new scorecard below was published in [1], based on seismicity data through August 31, 2005. The new scorecard displays a new statistic on accuracy. The "mean forecast error" is the average distance from an earthquake (blue circle, M > 5.0) to the nearest red pixel at the given forecast area coverage (175 red pixels out of a total of 14,400 total pixels on the map. Thus the fraction f = 175/14400 => 1.2% is a measure of the false alarm rate. Note that roughly half of the total area is seismically active, meaning that there was at least one M > 3.0 earthquake in a 0.1 degree x 0.1 degree pixel box. The mean forecast error distance obtained, 7 km +/- 16 km, measures the error in the forecast accuracy.
Most of this error is due to earthquakes # 6 and #12, which are about 50 km off the nearest red pixel. Without those
earthquakes, the error would be less than 5.5 km, half of one pixel box size.
[1] JR Holliday, CC Chen, KF Tiampo, JB Rundle, DL Turcotte and A Donnellan, "A RELM Earthquake Forecast Based on Pattern Informatics," Seism. Res. Lett., v78, pp. 87-93 (2007)
Computed July 29, 2008
[download a pdf version]
|
