November 2017

John Langbein

Retrospectively, recreated a "cleaned" up version of xhr1 creep
from raw telemetry data for the interval between Sept 1981 and Aug
1986.  Hopefully, this is close to the data shown in the Schultz
1989 Open File Report.  To do this, I used the station notes that
told me about times when the wire broke (twice before being abandoned)
and when the instrument was 'reset' or 're-zero-d'. Key to
reconstructing the time series is a record of manual measurements,
file xhr12.m, which some how spans the wire breaks and also combines
the manual measurements from xhr2.

First, from the telemetry data, I tried to remove the multiple
telemetry glitches and fixed the offsets due to 'resets'.  Next,
after decimating the 10-minute samples to daily samples, I matched
the telemetry data with the corresponding manual measurements of
creep (about 20 simultaneous observations).  Then, I fit the telemetry
data, in digital counts, to the manual measurements of creep, in
mm. In the regression, I also accounted for the sizes of the two
unknown offsets due to the wire breaking. The rms of the residuals
to the fit is 0.1 mm.  From that regression, I applied the coefficents
to the cleaned telemetry data to obtain a time series of creep with
10 minute sampling.