Earthquakes in the ANSS combined catalog, 1962 to present.

Historical earthquakes in the vicinity of the Rio Grande Rift (including west Texas, New Mexico, Colorado and southeastern Wyoming) hint at the potential for larger events in the future. An 1882 earthquake west of Fort Collins, CO had moment magnitude 6.6 (plus or minus 0.6), and there have been four earthquakes with magnitude greater than 5 in the past forty years.

To be useful for understanding earthquake hazard, seismicity should be observed from a dense seismic network over a long period of time (i.e., most of an earthquake cycle). In the Rio Grande Rift region, we have neither a good seismic network nor a long historical record. Written records in the region extend back only 150 to 300 years (and are longest in New Mexico), but the earthquake recurrence interval on major faults is at least 5000 years (see next section!). Also, the CO-WY-NM region has had poorer seismic instrument coverage than any other earthquake-prone region in the western United States (for example, until very recently there were two or fewer permanent seismographs in the entire state of Colorado).

In the map at left, seismicity in the Intermountain Seismic Belt (the dense cloud of earthquakes in Utah and stretching along the Idaho-Wyoming border to Yellowstone) is well-measured down to about magnitude 2 or 2.5, because of a dense (<50 km-spaced) network of seismographs maintained by the University of Utah. By comparison, the catalog in the RGR is well-measured only down to about magnitude 3.5 or 4. Consequently, spatial concentrations of background seismicity and other important information such as magnitude-frequency relations are very poorly known in this region.

Surface Fault Ruptures from the USGS Quaternary Fault and Fold Database, 130ka to present.

The study of prehistorical earthquakes (or "paleoseismology") is based on geologic mapping of fault structures and recognizable surface scarps. The best data comes from scarps that have been trenched (i.e., a trench dug crossing the scarp), which allows geologists to examine the offset soil horizons and date the offsets. Most of what we do know about earthquake hazard in the RGR comes from paleoseismology.

For an earthquake to rupture cleanly all the way to the Earth's surface, it must be relatively large (a minimum magnitude ~6.5 and more often nearer 7.0). The figure at left shows fault scarps that moved in the last 130,000 years in green, and scarps that moved in the past 15,000 years are shown in red. Some of these faults (for example, the Sangre de Cristo fault near Great Sand Dunes National Park in southern Colorado) have moved more than once in the past 15,000 years in events that likely had magnitudes 7.0 to 7.5.

The faults shown in the figure are by no means the only ones that pose a threat of future large earthquakes. The data depicted here are incomplete: Some faults have been trenched but have not yet found their way into the database, and there are other fault scarps which geologists have not yet gotten time or funds to trench. Moreover, recognizing young, small-offset scarps in the geomorphology of forested regions can be extremely difficult.

GPS Surface Velocities using NGS Continuously Operating Reference System (CORS) data.

Surface deformation is derived from geodetic data, most commonly continuous measurements of GPS position. The figure at left shows the best data currently available for the Rio Grande Rift region. Some of these GPS sites show very fast motions which likely are not due to tectonics, but ignoring the anomalous motions, the data suggest about 1 mm/year of east-west extension across the rift combined with about 1 mm/year of left-lateral strike slip. If that estimate of deformation rate is correct, then one could estimate the approximate earthquake hazard by comparison with rates of motion in another region where earthquakes are more frequent. The San Andreas fault and related plate boundary deformation in California totals about 35 mm/year over an area comparable to that of the Rio Grande Rift. That would suggest that we should expect (roughly) one earthquake in the Rio Grande Rift for every 25 earthquakes of comparable size in California.

There are several problems with estimating earthquake hazard in this way however. The biggest problem is that the GPS velocities shown at left are very, very small relative to potential sources of noise. One very serious problem is that most of the CORS sites used in the analysis were not designed to measure the slow rates of tectonic deformation in the Rio Grande Rift. Accurately measuring tectonic motion requires special monumentation that couples well to the motion of deeper bedrock. The CORS GPS sites were installed for other purposes besides tectonics and so were not carefully monumented (many of the sites were set on building roofs or monumented in soil). Consequently, these sites are very sensitive to localized movements resulting from, e.g., frost action in the soil, groundwater withdrawal, mine-related deformation, and structural settling. Moreover, separating out such small tectonic signal also requires very careful analysis of the GPS data and correction for other known deformation effects. For example, seasonal changes in position related to the weight of snow and water depressing the Earth's surface can be several mm in the horizontal and several cm in the vertical!

In part to overcome these problems, we are installing a network of 24 tectonic-quality GPS sites throughout the Rio Grande Rift region of Colorado and New Mexico. These sites were chosen to complement the existing GPS instrumentation and to densify in the region of the Rio Grande Rift where we expect maximum deformation. In about five year's time, we should have a much better idea of the earthquake hazard posed by this tectonic feature!





At Left: Network of GPS Sites we are installing for this project are shown as triangles. Green triangles are sites we are preparing to install, and red triangles have been installed and are collecting data. Stars are sites being installed for the EarthScope-Plate Boundary Observatory (blue = installed; purple = planned). Blue circles are existing CORS sites.