Step 12: Identify fault trains/sections with >50% aseismic creep

Many (most?) continental faults display stick-slip (seismogenic) behavior, while other faults display steady-creeping (aseismic) behavior.
In a previous step (Step 9) I suggested that you divide any fault train into sections wherever the dominant behavior changes.
Now it is time to identify the dominant mode of slip for each digitized fault trace/train/section in your dataset.

Programs NeoKinema and Long_Term_Seismicity both need to know which faults are creeping.
NeoKinema represents seismogenic (stick-slip) faults by a number of rectangular and triangular dislocation patches during the step where it computes the current model estimate of the “mean coseismic velocity” of all GPS benchmarks.  (This mean coseismic velocity is then added to the observed mean interseismic GPS velocity to get the long-term-average velocity of each benchmark.)  However, faults that are aseismic and creeping do not contribute to the total mean coseismic velocity field.
Long_Term_Seismicity forecasts long-term seismicity rates along active faults with seismogenic (stick-slip) behavior.  However, it ignores aseismic faults which are creeping.

For these reasons, the fault-offset-rate table that we started to build in the previous Step has a column labelled “C?”.
This is an abbreviation of the question, “does this fault Creep at >50% of its long-term slip-rate?”.
The possible answers are “True” or “False”, which will be abbreviated as “T” and “F”, respectively, in the table.

There are several types of data that can contribute to answering this question:

· If a historical seismic catalog or an instrumental seismic catalog shows that the fault produced a large earthquake, we should assume that the answer is “False”.

· However, if historical and instrumental seismic catalogs show NO earthquakes along the fault, it is not necessarily creeping!  A fault with long-term-average slip-rate of ~0.1 mm/a may only produce ~one earthquake, with ~1 m of surface slip, every ~10,000 years!  All negative results from seismic catalogs are therefore inconclusive, because the catalogs are too short.

· Offsets of man-made lines such as fences, sidewalks, roads, canals, and pipelines often give the first evidence of fault creep, and can be used to estimate the creep-rate.  Unfortunately, we cannot answer the critical question (>50%, or not?) without knowing something about the long-term-average slip-rate of the same fault.  Ideally, this could be obtained from larger offsets of dated geologic features.  Otherwise, it may be necessary to remember such a fault as problematic, and to return to this question after getting a preliminary model long-term-average slip rate for this fault by running NeoKinema!

· Geodetic velocity measurements (GPS-drived horizontal velocities of benchmarks, and/or InSAR images of the horizontal velocities of radar reflectors) are very useful, especially when we have observations at many different distances from the fault.  Observations close to the trace can give the fault creep-rate, and observations further from the trace (outside the belt of elastic strain accumulation) can give the long-term-average slip rate.

The recent discovery that many normal faults in the Apennine Mountains of Italy have fault-scarps that grow steadily [Kastelic et al., 2017, J. Geophys. Res. Earth Surf., 122, 114-129, doi: 10.1002/2016JF003953], even in the absence of earthquakes, raises a new problem for this classification.  Perhaps the hanging-walls of these faults (which are often composed of carbonate talus) are not purely elastic, and move (faster than any plausible tectonic rate) due to surficial strain mechanisms such as solution-transfer compaction and/or landsliding?  However, this does not necessarily mean that these faults are aseismic; instead, the creep might terminate (or diverge from the main fault plane) at shallow depths.  Hopefully, geodetic data will be able to reveal whether these faults are locked or creeping at depths of about 1~12 km, where earthquakes might be expected to nucleate.  It is the (inferred) deeper behavior of such faults that should determine their treatment in NeoKinema and Long_Term_Seismicity models.