Ethiopia Rift Measurements July 1997 - Field Report September 1997

Eight points previously measured by GPS methods in 1992 across the northern Ethiopian rift near 8°N were re-measured in July 1997. The remeasurements show that widening of the rift is 3.7±0.9 mm/year, and that most of this widening is concentrated in the deepest, most youthful part of the rift zone (2.9±1 mm/year). Masonry monuments were constructed over bedrock points separated by 119 km on the edges of the rift and a telemetry link installed to bring these data to the Geophysical Observatory for analysis. Data from these two sites are recorded 24 hrs/day to monitor rift widening and shear strain across the rift. They will also be used to search for possible amplification of periodic shear or dilatational strain signals applied to the rift zone at tidal and seismic frequencies. The proposed analysis of tidal amplitudes will permit us to estimate rheological conditions in the upper mantle, and will permit us to assess the contribution of tidal shear strain to rift volcanism. Additional GPS points were installed near Gode in Ethiopia and west of Kampala in Uganda.

Fig.1. The Afar triple junction showing locations of the GPS arrays across Afar and the Ethiopian rift, and estimated velocities relative to the Nubian plate.

In January and again in November 1992 scientists from EIGS , EMA, the Geophysical Observatory of Addis Ababa University, and from CIRES, the University of Colorado, measured the relative positions of approximately 20 points across the northern Ethiopian Rift and across the Afar depression to an accuracy of 3 mm. The objective of the measurements was to measure directly the rate of widening of these two rift zones.

By measuring at points that had been established 20 years earlier by the Geophysical Observatory in Afar (Mohr, 1970) it was possible immediately to learn something of the recent activity on these rift zones. Thus in 1992 we established that the Dobi graben had widened approximately 30 cm since 1970. Although this is probably associated with the 1989 swarm of M5 earthquakes that activated normal faulting throughout the graben, the measurement corresponds to a mean opening rate of 15 mm/year, 75% of the mean widening rate for the Red Sea rift obtained from global plate studies.

Surprisingly, no widening (1±1 mm/yr) was detected in 1992 in the main Ethiopian rift zone near Addis Ababa, where control points had been first measured in the interval 1969-74 (Asfaw et al. 1992). Various reasons for the null result were proposed: the opening rate was perhaps less than the ±2 mm/year detection threshhold of the measurements, rift widening perhaps occurs only during seismic activity, rift activity may have occurred outside the 48 km central part of the rift measured, or perhaps the rift zone is currently acting as a strike-slip system of faults.

The measurements in 1997 show that the 1992 estimate was contaminated by what we now infer to be a spurious measurement at the point Ganti toward the SE, resulting in an apparent contraction of the line Ganti-Mietchi. The 1992 estimate of widening was based on three non-contiguous lines (Figure 2), and a fourth that was not compared because it was to an auxillary point at Boku. The 1992 and 1997 observations of this point, when adjusted correctly for the eccentricity of the 1970 geodimeter measurement shows consistent elongation, indistinguishable from a linear strain rate. The strain rate we obtain for the Boku-Mietchi line is of the same order as strain of contiguous lines in approximately the same azimuth, and since the Boku-Mietchi line and the Ganti-Mietchi line differ in azimuth by less than 20° we infer instability of the Ganti point is responsible for its apparent contraction 1970-1992. The point was destroyed by villagers soon after our measurements in 1992.

Figure 2 First epoch measurements of the central rift were completed in 1970 using geodimeters (Mohr et al, 1976). In 1992 and 1997 some of these lines were remeasured using GPS methods. The line Ganti-Mietchi alone shortened presumably due to spurious motion of the point Ganti which is now lost. Mariam remains but could not be measured because an army had camped above it in 1997.

Because the 1970's geodimeter points we occupied in 1992 formed lines with azimuths normal to the trend of the rift, their comparisons with 1992 and 1997 line-length data are almost completely insensitive to rift parallel displacements. Some authors have proposed that shear strain may be significant in the Ethiopian Rift (Bonini et al, 1997). The 1992-1997 GPS data in principle permit the resolution of rift-normal and rift-parallel strain, however, the crucial observations across the rift in 1992 (KOLO, ADD1 and BOLO with several day occupations) were undertaken during a time of anti-spoofing (AS code implementation), and consequently we were unable to obtain improved orbits in our estimates of 1992-1997 east and north motions. The data we have processed suggest that east and south components of displacement are somewhat similar, i.e. that the rift apparently opens normal to its margins. Current uncertainties permit 2±2 mm/year of sinestral slip in addition to the 3.9±0.6 mm/year opening rate in a SE direction.

The 1997 remeasurements of 1992 points: Kolobo, Addis Obs. (ADD1 & ADD0), Dukam, Ridge (near Nazret), Mietchi (Dera), Selassie (NW of Sire) and Bolo Michael (East of Sire) were completed with 6 receivers in 8 days with occupation durations of 3-10 days. The point Ridge was measured for the first time with GPS in lieu of a nearby point currently inaccessible to us (Mariam, east of Mojo, which is now beneath a temporary army telemetry post). The Dukam point we occupied is one of the 4 French points of that name and the point we occupied was 11.26 m from our 1992 point requiring an eccentricity estimate before comparison with the 1992 data.

Fig.3. Points measured across the Northern Ethiopian Rift in July 1997, and the new fixed GPS link ADD1-REDG.

Summary of strain changes and velocity field observation

Figures 4-6 show least squares fits to all available data from geodimeter and GPS occupations as of 6 Sept. 1997. The weighted least squares fits to the data are summarized in NW /SE rift profiles of strain and velocity. These summary plots are composite plots merging GPS and geodimeter data. The 27 year data sample yields displacement and strain rates 2-3 times more precisely than do the GPS only data for the past 5 years, but they are confined to the most active 25% of the width of the rift. The combined line length increase between Ridge and Selassie is 2.9 mm/year with an uncertainty of 1.6 mm/year or 0.8 mm/year if all the errors are uncorrelated. The Addis-Selassie line extended 3.1±0.9 mm/year 1992-1997, and the Addis-Bolo line extended 3.7±0.9 mm/year 1992-1997, suggesting that little additional deformation occurred outside the zone of deformation monitored by the geodimeter lines.

GPS Telemetry - an electronic ruler across the rift

Two permanent GPS receivers were installed to improve the rift widening estimates and constrain the azimuth of maximum rift separation. The points are 119 km apart in Addis Ababa (ADD1) and Bolo Michael (REDG) respectively, and operate 24 hours/day. Continuous observations permit improved tropospheric modeling, and the estimation of seasonal effects on the data, and in principle permit us to achieve 1 mm/year positioning accuracy. Masonry pillars were built at each location and a spread-spectrum radio link used to relay data from each Trimble 4000SSI receiver to the observatory without disturbing data acquisition. The line-of-sight radio link crosses the entire rift zone without repeaters. Data from these two sites are recorded every 30s with a 15 degree window resulting in 3 Mb/day of data (90 Mb/month). Data downloading takes 15 minutes via radio. The data are archived on 100 Mbyte removable ZIP disks stored in the observatory.

Fig 7. Profile along the GPS line of Fig.2 corrected for the Earth's curvature (dashed curve). Point REDG if used as a repeater has line-of-sight access to most of the rift zone.

A point was installed remote from the Ethiopian rift near Somalia at Gode, and at four points in Uganda to provide estimates of plate deformation and the rotation rates between blocks adjoining the rift. The 513 km distance from Bolo Michael to Gode potentially yields a rotation precision of 8 nanoradians/year for the Somali block.

References

Asfaw, L., R. Bilham, Mike Jackson and P, Mohr, Recent Inactivity across the African Rift, Nature(Lond), 357,447, 1992.

Bonini, M, T. Souriot, M. Boccaletti, & JP Brun., Successive orthogonal and oblique extension episdes in a rift zone: Laboratory experiments with application to the Ethiopian Rift, Tectonophysics, 16, 347-362, 1997.

Mohr, P., Ethiopian Rift Geodimeter Survey. Smithsonian Astrophysical Observatory, Special Report 376, 111 pages. 1974

The following positions were calculated by Kristine Larson using precise orbits. Their absolute positions are accurate to 10 cm and their relative positions (one to another) are accurate to approximately 1 cm horizontal and 2 cm vertical.

Ethiopia Positions July 1997, ITRF94 (International Reference Frame 1994)

absolute accuracy 10 cm in all three coordinates

ADD1 4913665.3315 3945901.0005 995400.6389

BOKU 4884809.7068 3995656.6800 933731.3982

BOLO 4871316.2271 4018483.8800 910342.9756

DUKA 4908226.9368 3958127.4352 971030.0564

KOLO 4925779.6875 3929903.8700 1000448.8505

MIEH 4882511.7496 4001618.7075 919934.4099

RIDG 4886768.3213 3991738.3873 939189.0715

SELA 4876132.9001 4010714.9717 913954.4508

GOD1 4596951.6915 4372688.9951 654710.1504

REDG 4870482.0790 4019341.6996 911217.1941

WGS84 - HEIGHTS ARE ELLIPSOIDAL, NOT RELATIVE TO SEA LEVEL

absolute accuracy 10 cm in all three coordinates

relative positions within 1 cm horizontal, 2 cm vertical

ADD1 9.0352955446 38.7660788042 2438.1467606155

BOKU 8.4721290826 39.2823393330 1860.3851542715

BOLO 8.2576351630 39.5201676573 2477.5325655313

DUKA 8.8128860044 38.8836469669 2044.5369219407 (RM1)

KOLO 9.0811179533 38.5836959689 2700.7611098504

MIEH 8.3461146082 39.3374289105 1820.2445489438

RIDG 8.5222070482 39.2435392281 1714.0352494195

SELA 8.2915278158 39.4379672029 1788.3428764101

GODE 5.9311343987 43.5677654209 256.8432751121

REDG 8.2655811793 39.5309878067 2506.6868776418