Abstract -
It is well known in a qualitative sense that some faults are more segmented than others. This work uses results from fault zone mapping in outcrop, mine and seismic datasets to establish a quantitative understanding of the segmentation of normal faults. At its simplest, we ask questions such as: "how many intact relays would I expect to find on a 1 km trace of this 30 m throw fault?" A slightly more complex question of practical relevance is: "what is the probability that this 15 m thick aquifer is self-juxtaposed at some point along this 1 km trace of this 30 m throw fault?" In both these cases, the question is posed as a function of localised fault throw, but an alternative means of characterising the faults is to do so as a function of local zone width, by asking: "what is the frequency of relay zones at least 10 m wide on faults in this system?" We describe results from both forms of characterisation using data from 153 fault zones from 15 different normal fault systems with maximum throws in the range of tens of centimetres to hundreds of meters.
The most direct means of measuring fault zone segmentation from the practical perspective of risking possible across-fault juxtaposition, is by recording the number of distinct across-fault flow paths produced by self-juxtaposition of a flow unit of a given thickness on a fault of a given throw. We normalise the flow unit thickness by the fault throw to give a dimensionless parameter we term Threshold Fractional Throw (TFT), and find that the frequency (number/km) of across-fault flow paths at particular TFT values for faults in particular systems is inversely proportional to the fault throw, with different fault systems having different segment frequency characteristics. If TFT = 0.0 then the only segment boundaries considered are intact relay zones, which, for a 10 m fault, have frequencies ranging from 1 every 10 km of fault trace, to 3 per kilometre (Fig 1a). If TFT = 0.5 (Fig 1b), some faults examined have regions where self-juxtaposition of a flow unit of only half the thickness of the fault throw, occurs as often as once every 100 m on a 10 m throw fault.
The second way of examining the frequencies of relay zones is as a function of their width. The advantage of this approach is that faults with different throws in the same system can be grouped together, since the width of a relay zone does not change as a fault accumulates throw. The resultant relay zone width populations (Fig 2) are characterised by approximately self-similar distributions (i.e. the frequency of zones in inversely proportional to their width). Normal faults typically have dimensionless frequency values for width populations of fault lenses and relays of between 15 and 120. These values imply that one of these features greater than 10 m wide typically occurs between 2 and 12 times per kilometre of fault trace.
We find that segmentation frequency on faults from a given area can vary significantly, but this variability is smaller than the differences between different areas. Preliminary analyses indicate geomechanical controls on segmentation frequency. Additionally, we have compared our measurements with data for strike-slip faults published by de Joussineau and Aydin, and find that both fault modes have similar plan-view segmentation characteristics, with discrepancies accounted for by sampling differences.
Abstract of talk given to:
Geometry and Growth of Normal Faults, Geological Society of London, June 2014.