|
Activity Recorded by Data Logger - by Ralph Taylor |
|
|
|
Introduction |
Castle Geyser has been monitored electronically since 1997. Data from 1997 to late 2002 is intermittent, but since that time the record is nearly complete. Our original location for the data logger was subject to damage from ice and water, and two gaps in 2004 were caused by failed loggers. We have since relocated the logger to a safer location and are having better success with continuous recording.
A gap in data from 2019 on 24 May 2008 to 1035 on 27 May resulted from the logger's memory filling.
|
|
|
Activity in 2011 |
Since Castle has two distinct sorts of eruptions the analysis is more complex than for other geysers. The statistical summary for 2011 shows the activity analysis. A pdf of this summary is at Castle Geyser Recent Activity Summary. The "all eruptions" box shows combined interval statistics for both major and minor eruptions. The boxes on the right break the intervals down into three categories: Major to major intervals, minor to major intervals, and the first major-to-major interval following a minor. The latter category is significant since we note that the first full major interval following a minor eruption is fully an hour longer than the successive major-to-major intervals. This factor greatly improves the prediction accuracy.
|
|
|
|
The overall interval graph shows all of the intervals for 2010 up to the most recent download. The graphs for the current year are updated about every six weeks from October to June and weekly from June to the end of September.
|
 Click for a larger image
|
|
|
The next graph shows the intervals for the past few months at an expanded time scale. In recent years the minor intervals have tended to be clustered; that is, for a few days there are frequent minor intervals, but there are also periods of a week or more with no minor intervals.
|
 Click for a larger image
|
|
|
The next graph shows just the major-to-major intervals, excluding the minor-to-major intervals. The black points indicate major-to-major intervals and the red points show the first major-to-major interval after a minor. Nearly all of the intervals fall in a narrow band between 13 and 14 hours, with a few much longer intervals exceeding 14h. The black points that indicate very long intervals, especially those in the winter months, may represent intervals with an undetected minor eruption.
|
 Click for a larger image
|
|
|
The overall interval distribution shows a skewed distribution for the major-to-major intervals with the minor-to-major intervals represented by the bar at the left and the long intervals represented by the short bars to the far right. Note that in this and the other histograms displayed here the labels shown on the X-axis represent the upper boundary of the class, not the midpoint. Geyser times are traditionally truncated. The graph at the right has class widths of 30 minutes. The bar appearing above the label "13:30," for example, contains intervals from 13h01m through 13h30m.
|
 Click for a larger image
|
|
|
The predictions for Castle are usually quite good providing the overnight activity (minor or major eruption) is known. The final graph is a prediction results graph, showing where the eruptions fell within the prediction window, which is plus or minus one hour for Castle. Most of the time, Castle's predictions are well above the 90% target.
|
 Click for a larger image
|
|
Activity since 1997 |
A plot of Castle's intervals shows that major eruptions are more common than minor eruptions, typically comprising more than 75% of all of the eruptions. The plot of all of Castle's intervals shows the major eruptions as the heavy band between 11 and 13 hours and the minor eruptions as the shorter intervals, dropping to under 3 hours in some cases. When Castle has a minor eruption the interval until the next major eruption is not predictable.
|
 Click for a larger image
|
|
|
However, when Castle has successive major eruptions, the intervals are quite a lot more consistent, as shown in the second graph.
|
 Click for a larger image
|
|
|
The multi-year trend is to increasing intervals, from 11 hours between major intervals in 1997 to nearly 14 hours in 2011. There was a sharp increase in Castle's intervals coincident with the large earthquake at Denali in Alaska in November 2002 resulting in average intervals 35 minutes longer after the quake than before the quake.
|
 Click for a larger image
|
|
Activity in 2010
|
Activity in 2009
|
Activity in 2008
|
Activity in 2007
|
Activity in 2006
|
Activity in 2005
|