In this post I will elaborate on how to understand the Icelandic borehole strain and seismicity graphs. For the experts I might just be stating the obvious, but for the more general public (like myself) this might be a guide on how to understand all these enigmatic waves and ripples.
This map shows the locations of three kinds of instrument that monitor earthquake and volcanic activity around Hekla volcano. SIL stations (of the South Iceland Lowland automatic earthquake data acquisition and evaluation system; black triangles), GPS stations (yellow) and volumetric borehole strainmeters (green squares).
Location of the SIL and GPS stations and borehole strainmeters.
Image courtesy of IMO
http://hraun.vedur.is/ja/hekla/Stadsetning_stodva_31052011.jpg
Strainmeters can be of various design. In Iceland we are dealing with Sacks-Everton volumetric strainmeters. Wikipedia reveals: “a design that uses specially shaped volumes to measure the strain tensor.” In other words, changes to the volume of a fluid filled chamber anchored in the borehole.
Image courtesy of IMO
http://hraun.vedur.is/ja/hekla/thensla_burfell.html
The sample rate of the volumetric strainmeter data is one second (1 sps = samples per second, i.e. 1 Hz). The unit “strain counts” on the vertical axis is arbitrary, because a gain is manually set to determine what amount of relative change in strain or stress is one count. Strainmeters indicate ground velocity (displacement per time). Positive strain values mean volume increase in the bedrock (extension due to tension force, i.e. strain), negative values decrease of volume (contraction due to compressive force, i.e. stress). If you think of driving a vehicle, this plot shows your velocity relative to the starting velocity, since the start of the trace is always set to zero. A massive drop or rise might for example indicate you came to full stop at a tree or reached escape velocity for space travel.
Whether a strainmeter shows extension or contraction during an eruption depends on its relative position to the conduit/rift, see the opposite reactions during the Hekla 2000 eruption.
Búrfell darkblue,
Saurbær blue, Skálholt red, Geldingaá yellow, Stórólfhvoll violet, Hella light blue. Image courtesy of IMO
http://hraun.vedur.is/ja/englishweb/heklafigure1.html
Besides the Hekla strainmeter Búrfell is the second closest to Hekla, roughly 15 km at a perpendicular angle to the rift direction. The huge strain drop (i.e. massive stress increase) at Búrfell was interpreted as magma forcing it’s way up, opening a conduit. On the other hand, the simultaneous strain increase (decreased stress) at the other stations was due to emptying of the magma chamber. Here is further (paywalled) read on the strain during the 1991 Hekla eruption. The unit nanostrain indicates a change by a billionth part of the volume, i.e. 10-9. Earthtides are known to have an amplitude of about 50 nanostrains. The 2000 eruption caused a sudden drop about an order of magnitude larger.
A seismometer literally measures shaking, i.e. motion of the ground, which can be recorded as a seismogram.
The seismometers of the SIL array can both measure ground displacement (unit is meters per second, m s-1) or be used as accelerometers (unit meters per square second, m s-2).
Most Icelandic seismometers are 5 sec (0.2 Hz resonant frequency, limiting the frequency range) Lennarz seismometers. The sampling frequency is 100 Hz. The Haukadalur seismometer (63°58´08.4´´ N / 19°57´54.0´´ W, appr. 10 km West of Hekla) is a LE-3D/5s, measures oscillations in three dimensions (“transverse”, North-South; “radial”, East-West; “vertical”, Up-Down).
Tremor amplitude time series with different frequency bands. Vertical axis: One-minute averages of the vertical component of the tremor amplitude, x micro meters s-1. Image courtesy of IMO
http://hraun.vedur.is/ja/hekla/oroi_hau.html
First of all, this graph does not show the raw seismogram, but is a spectral analysis. You remember the colorful spectrograms from the El Hierro stations? A spectral analysis is performed on the waves of the seismogram to extract oscillations of different frequencies. Several algorithms can be used to create a spectrogram, for example STFT, short-time Fourier transformation, or CWT, continuous wavelet transform. For El Hierro the amplitudes are given over the whole frequency range while in Iceland they show averages of three frequency bands.
This example is a tremor amplitude time series showing averages of the frequency bands 0.5–1.0 Hz (red line), 1.0–2.0 Hz (green line) and 2.0–4.0 Hz (blue line), of the vertical component (Z) for the station HAU. Unfortunately the vertical axis is not labelled, but is presumably representing the amount of bedrock displacement in micro meters per second multiplied by a variable scaling factor (x). The values are presumably one-minute averages. An example for this analysis is described e.g. in this thesis, see p. 564 ff.
The blue trace (high frequency band, fast shaking) mainly represents earthquakes and the green and red traces (low frequency bands, slow shaking, harmonic tremor) tremor from magma movement, which for Hekla is usually in a well-defined spectral band at 0.5–1.5 Hz (see the thesis).
Based on previous observations, the following scenario might occur when the next eruption is about to happen: First there will be more earthquakes opening a fissure, showing as an increase of the blue earthquake trace amplitude by an order of magnitude. When the fissure is opened earthquake activity seizes and the blue trace will go back to normal. Meanwhile the magma starts spilling out and a sudden increase in the red and green tremor trace amplitude by at least an order or magnitude will be seen, which gradually decays with decreasing pressure. What we should actually look for in this graph is not the width of the traces, which only indicates how much the shaking amplitudes vary, but a really really strong rise of the curves as seen in 2000:
Tremor amplitude time series with different frequency bands. Vertical axis: One-minute averages of the vertical component of the tremor amplitude, x micro meters s-1. Image courtesy of IMO
http://hraun.vedur.is/ja/hekla/hau20000226.gif
Lastly, the following graph is a composite of data derived from the volumetric borehole strainmeters and from the Haukadalur seismometer, plus information on local earthquakes determined by the SIL system.
Upper panel: Volumetric strain rate.
Lower panel: earthquake magnitude (left), horizontal components of tremor amplitudes (right)
Image courtesy of http://hraun.vedur.is/ja/hekla/borholu_thensla.html
The upper part shows the “two-minute median from one-second data” of borehole strain rate (strain counts per second) measured by the four stations Búrfell, Hekla, Hella and Stórólfhvoll. See the green squares on the map. A change of the strain rate means the bedrock is compressed or extended faster or slower than before. The cause of this is a change in the pushing or pulling forces. Think of it as your vehicle being accelerated or decelerated when pushing your gas or brake pedal. This graph shows what your feet do. When Hekla erupted in 2000 the strain rate looked like this:
The rate of strain changes in Búrfell (blue, 15 km from Hekla) and Skálholt (red, 45 km from Hekla) (nanostrain per hour)
Image courtesy of IMO http://hraun.vedur.is/ja/englishweb/heklafigure3.html
The minimum in the strain rate indicates the time of the surface breakout of the magma, along with the visual observation of the eruption at 18:17.
Because the ground is moved by several variable sources, mainly earth tides (very slow change in strain counts rate) and microseismicity (very fast change in strain counts rate) the above mentioned two-minute time range is chosen by which these events are filtered out. Then the median, the mean value separating the higher half of a data sample from the lower half, is plotted.
The left axis in the lower part shows the magnitude (in Ml) of local earthquakes. Since most of the time there are no earthquakes (counted in the lower right corner) no trace appears.
The right vertical axis in the lower part indicates the bedrock displacement, i.e. velocity in micro meters per second. The data is derived from the horizontal components (North and East) of the Haukadalur tremor amplitude time series data, which are 60-sec averages. Short-lasting shaking, for example caused by single earthquakes or a sledge hammer, are averaged out by plotting the the three-minute median. When an eruption is imminent, the blue (high frequency) trace will rise first indicating fissure opening and the green and red traces will follow when the eruption starts.
Standard VolcanoCafé disclaimer: I am not an expert on this topic, just read a few papers while researching for the post. Please excuse me if I jumped to false conclusions and feel free to post corrections!
chryphia
Many thanks to the dragons who read the draft and special thanks to Geolurking for helpful comments! 
Other links:
-The SIL seismological data acquisition system – As operated in Iceland and in Sweden. Abstract only (2003)
-How a seismometer works from Sep 25, 2012 by Geolurking
-Summary about long and short period and broadband seismometers in this blog post
-ON THE USE OF VOLUMETRIC STRAIN METERS TO INFER ADDITIONAL CHARACTERISTICS OF SHORT-PERIOD SEISMIC RADIATION
-Seismometers of the SIL used as accelerometers
- Earthquake engineering research center, University of Iceland operating the Icelandic strong-motion network since 1984.
-Sturkell et al., 2005, Volcano geodesy and magma dynamics in Iceland
-Description by IMO of the Hekla 2000 eruption.
-Visualizing Stress is a good site, even if you are not into the math aspects of it, it has some really good narative data in the tutorials.
Name those Volcanoes Riddle
