How To Process Borehole Strainmeter Data

The basic steps involved in processing NOTA borehole strainmeter data are outlined in the flowchart below. Full details are given in How Strain Data Are Processed and summarized in Strainmeter Products Summary. The PBO Data Critical Design Review describes the processing techniques used for all NOTA datasets.

For NOTA strainmeters extensional strain is positive.

1. Download Raw Data

The raw data from each of the four gauges within a PBO strainmeter are available in 20 Hz, 1 Hz and 5-minute interval data rates. All BSM data are available in Bottle and SEED format (see PBO SEED channel codes) plus the low frequency data set is available in ASCII format.

2. Linearize Raw Data

The raw measurements from each gauge are outputs from capacitance bridges. To convert the gauge output, in counts, to strain along the axis of the gauge the following equation is used.

et=((Rt/1e+8)/(1-(Rt/1e+8))) – (Ro/1e+8)/(1-(Ro/1e+8)))*G/D

where et is the linearized strain at time t, Rt is the raw data point, Ro is a reference data point, G is the reference gap within the strainmeter (100 or 200 microns) and D is the instrument diameter (87 mm). All Level 2 time series are contain the linearized strain data.

3. Edit Data

Before tides and long term trends can be extracted from the gauge data non-tectonic and tectonic steps must be removed. Examples of such signals are coseismic offsets, instrument restarts or lightning strikes. UNAVCO’s BSMAC isolates and finds such signals and flags them in the Level 2 Data Product.

4. Estimate the Earth and Ocean Load Tidal Signal

Once large steps have been removed from the time series the clean data can be input to a tidal analysis program such as BAYTAP-G and the amplitude and phase for the main tidal constituents determined. Each gauge is analyzed separately. When the tidal signal is being estimated for the first time it is often necessary to repeat steps three and four as small data steps will become apparent once the tides have been removed. These amplitudes and phases are calculated by UNAVCO’s BSMAC and provided as a Level 2 Data Product both in an XML file and as a simplified ASCII file for each BSM.

5. Estimate The Barometric Pressure Response Coefficient

The strain changes are assumed to be linearly related to change in barometric pressure response. One linear response coefficient is estimated for each gauge . Typical responses range from a few to 10 nanostrain per millibar. These responses are also provided as a Level 2 Data Product.

6. Estimate the Long Term Trends

Estimating the long-term trends is one of the most difficult steps in processing BSM data as care must be taken to not remove the signal of interest. The expected long-term trend is one of compression as the borehole tends towards closure. For simple cases the trends can be forward modeled using an exponential plus linear trend. These forward modeled trends are provided as Level 2 Data Products. Alternative methods for detrending are high, low or band pass filtering to isolate the signal for interest.

7. Generate Areal and Shear Strains

The final step in BSM data processing is the calculation of areal and shear strains. In this step the gauge measurements are combined to form areal (Eee+Enn), differential extension (Eee-Enn) and tensor shear strain 2Ene . Currently the areal and shear strain measurements are generated using the manufacturer’s calibrations and the tidal calibrations as published by Roeloffs (2010). These matrices are listed in the Level 2 XML files and a simplified ASCII version.

Requesting High Rate Processed Strain Data

High rate (1 sps) processed strain data is generated by UNAVCO after a M7 event any where in the world, any geophysical event of interest within the Plate Boundary Observatory and upon user request. The processed data set contains all the corrections that are normally seen in the 5 minute Level 2 data set : tidal, barometric pressure and borehole trend corrections plus offsets and data quality flags. In addition the files will contain tiltmeter, pore pressure and high rate barometric pressure data if they are collected at the site.

The data will be placed in an anonymous FTP directory for download and an email sent to the data requestor. Data sets generated following events may be updated with additional data in the hours following the event if required. The format is that of a space delimited ASCII file.

High Rate Data Format

For each event there is one tar file per strainmeter. Within each tar are five gzipped files, one for each gauge (CH0,CH1,CH2 and CH3) and one for tensor strain.

File Name Convention:

  • BNUM.CH0.event.txt.gz
  • BNUM.CH1.event.txt.gz
  • BNUM.CH2.event.txt.gz
  • BNUM.CH3.event.txt.gz
  • BNUM.tensor.LAB.event.txt.gz
  • BNUM.tensor.PBO_01.event.txt.gz

BNUM refers to the 4-character ID, e.g., B004 and “event” refers to the event name. E.g., B004.CH0.20110311Tohoku.txt.gz.

File Format

The format for the CH0, CH1, CH2 and CH3 files is:

1Date TimeUTC
2Gauge measurementDigital counts
3Gauge strainMicrostrain
4Interpolated missing valuesMicrostrain
6Barometric correctionMicrostrain
7Barometric pressureMillibars
8Pore pressureMillibars
9VersionGeneration date

The format for the tensor strain file is:

2Eee+Enn StrainMicrostrain
3Eee+Enn TideMicrostrain
4Eee+Enn Interpolated missing valuesMicrostrain
5Eee+Enn barometric correctionMicrostrain
6Eee-Enn StrainMicrostrain
7Eee-Enn Interpolated missing valuesMicrostrain
8Eee-Enn TideMicrostrain
9Eee-Enn Barometric correctionMicrostrain
102Ene StrainMicrostrain
112Ene Interpolated missing valuesMicrostrain
122Ene TideMicrostrain
132Ene Barometric correctionMicrostrain
15Barometric Pressuremillibar
16Pore Pressuremillibar
17VersionGeneration date
18Tilt XMicroradian
19Tilt YMicroradian

The null value is 999999. If 1-sps barometric data are available it will be given otherwise there will only be 30 minute interval data. Pore Pressure and tiltmeter data are only available for a subset of sites. The “Interpolation” columns contained interpolated values to replace the null values. This column should be substracted from the linear strain to remove the null values. Barometric, pore and tiltmeter time-stamps have been rounded to the nearest second.