Rate Changes detected by PBO strainmeters
by J. Langbein, USGS
A quick look at the strain changes
This page contains links to plots of strain data from the PBO borehole strainmeter network. The plots presented below explore rate changes in these data. Specifically, a period of data is selected for the previous 100 days, previous 40 days, and previous 10 days; a rate is fit to the all of the data in those time intervals and, for the last 10% of the time series (10 days, 4 days, and 1 day) a rate change is also fit. The estimated rate change is compared with its standard error. The plots are updated once per day through automated scripts.
The “stack plots” in the table below show the changes in strain in two columns; the first being the tensor strain changes in geographic coordinates; and the second column are the strain changes that have been high-passed filtered using a simple running average of 30 days; this should highlight the short term changes in strain.
|last 100 days||last 40 days||last 10 days|
|Olympic Peninsula||100 days||40 days||10 days|
|Southern Oregon||100 days||40 days||10 days|
|San Juan Bautista, Calif||100 days||40 days||10 days|
|Parkfield, Calif||100 days||40 days||10 days|
|Southern Calif||100 days||40 days||10 days|
|Mt St. Helens||100 days||40 days||10 days|
|Yellowstone||100 days||40 days||10 days|
This new page represents consolidation of recent work with processing USGS borehole strainmeter data and a newer (better?) method to calibrate the PBO strainmeters. Nearly all of the scripts (programs) used to process and update these PBO data are also used to process the USGS strainmeter data. Hopefully, similarity of the scripts applied to the different data sets will result in better “up-keep” of the results presented here. Finally, both the raw and “cleaned” data should be obtained from UNAVCO.
New Item -- Plots of past transient strain changes of interest
Retrospective plots of historical transients.
|Event||Stack Plots of older strain data|
|Vancouver Island/Olymipic Peninsula; Mid to Late-May 2008 ETS||May 2008 ETS|
|Vancouver Island/Olymipic Peninsula; Mid to Late-Jan 2007 ETS||Jan 2007 ETS|
PBO Strainmeter Locations
The solid triangles on the map below mark the locations of the strainmeter being monitored here. Open and smaller triangles show many (but not all) of the PBO strainmeters that I am not analyzing for a variety of reason.
Brief summary of processing
Each borehole strainmeter consists of 4 gauges that measure extension in four different directions. The regional strain measured by each gauge is primarily affected by the presence of the borehole, and the material used to cement the strainmeter into the borehole (about 100 to 200 meters depth). There are a number of secondary contributions that may affect the measurement, too.
Since the Earth Tide is a well known repeating source of strain, I have used both the M2 (12.42 hour period) and the O1 (25.82 hour period) to calibrate these strainmeters in terms of the dilatation (Eee + Enn), and two shear components, 2*Een and Eee - Enn. Calibration is based upon the methods outlined in the a paper by Hart et al.. There, they discuss in detail the methods when isotropic coupling can be applied to fit predicted, theoretical tensor strain tides to the tides observed on each of the strain-gages. In their appendix, Hart et al. also discuss the possibility of anisotropic coupling. For nearly all of the PBO strainmeters, the standard approach of isotropic coupling with realistic values of areal and shear coupling suggested by Hart et al. did not yield a calibration matrix such that the theoretical tides matched the observed tides. Relaxing the constraint on areal coupling provided a better fit of the calibration matrix for some sites. For other sites, however, I needed to abandoned isotropic coupling in favor of completely aniostropic coupling or, vertical anisotrophy. Finally, in some cases, B009, B012, and B018, I could not obtain a calibration matrix that matched both the observed and the theoretical tides. For both B009 and B012, they are installed within a few hundred meters of the coast where there are large, > 2 meter, variations in the local, ocean tide. Although B012 is located approximately 5KM from any inlet from Puget Sound, the tidal models that incorporate the local ocean load are probably not known to the degree needed to match the observed strain-tides.
Processing individual components of strain
Below, are plots showing the daily results of processing from each site, both the individual gauge data and the tensor components of strain. These are produced by running the processing algorithm with the option to use prior estimates of the tides and pressure coefficents. Seen in the plots are the raw strain data (grey), the atmospheric pressure data rescaled to by the pressure coefficient (green), the strain data adjusted for changes in the pressure (blue), and the residual strain after the Earth Tides have been removed (red). The strain data have had a linear trend removed. The black, dashed line is the linear trend and a change in rate for the last 10% of the time span plotted. Finally, the top graph is a plot of the first differences of the residual strain data normalized to the expected drift in the data; the “dot plot” is useful for detecting offsets and outliers in the strain data.
Comparison of current and prior analysis of strain data
The following presents output of cleanstrain+ for the most recent 100 days and results from analysis of a previous 100-day chunk of strain data. For the comparison of pressure and tides, a pdf table is provided which lists all of the estimates for the pressure and 4 Tidal constituents. Please note that I only present the analysis for the gage data and not for the components of strain to minimize space.