Background
The origins of non-tectonic fractures in exposed rocks have been debated for over a century (Griggs, 1926; Blackwelder, 1927, 1933; Ahnert, 1996; Bloom, 1998; Easterbrook; 1991; Cooke et al., 1993; Birkeland, 2000; Watson, 1992; Ritter et al.; 2002; Yaalon, 1970; Smith and Warke, 1997; Goudie et al., 2002). Recent work, however, convincingly targets diurnal thermal stresses as being responsible for crack initiation in exposed rocks (Hall and Andre, 2001, Halsey, et al., 1998, McFadden et al., 2005). These studies, combined with preliminary modeling and data demonstrate that: 1) the majority of cracks on geomorphic surfaces in both deserts (McFadden et al., 2005) and more humid climates (Eppes, unpublished) have preferential orientations independent of rock fabric and shape (fig 1).. Such preferential orientations of cracks that are not parallel to rock fabrics or faces is difficult if not impossible to explain by any other mechanism other than directional solar heating. Cracks caused by freeze-thaw, salt shattering ect. should exhibit random orientations; 2) temperature fluctuations on rock surfaces are rapid and spatially not uniform (Hall and Andre, 2001; Sumner, 2004), and 3) realistic variations in rock surface temperature (e.g. Hall and Andre, 2001; McFadden et al. 2005) can produce sufficient stress within and on the rock surface (approximately 7-15 MPa) to exceed the tensile strength of most rocks (preliminary modeling, this study, see below).
In addition to this work connecting solar-induced thermal stresses to the observed preferred crack orientations, interest in the role of solar exposure on rock fracture has recently been further stimulated by Morres et al., (in press); they propose that solar exposure indirectly controls cracking by influencing the water content of cracks, favoring the growth of pre-existing microcracks oriented in relatively shaded directions that favor water retention.
All of these studies provide ample motivation to examine the role of diurnal insolation in the fracturing exposed rock. Numerous questions remain unanswered quantitatively: What are the insolationinduced rock temperatures and how large are the resulting stresses in exposed rock? What role does length of rock exposure at the surface, rock location, rock type, or rock shape or size play in the potential efficacy of thermal-induced cracking? To date, as no previous study has attempted to answer these questions, the role of the sun in the propagation of rock fractures remains poorly understood.
Figure 1: Rose diagrams of orientation data for cracks not parallel (+/-10 degrees) to rock shape or fabric on A. boulders in a variety of desert geomorphic surfaces (McFadden et al., 2005, n=~450), B. a west-facing boulder field in Virginia (Collected by Eppes and her students, n=35), and C. a single large boulder on a south-facing slope in South Carolina (Collected by Eppes and her students, n=52). The preferred orientation of the cracks provides convincing evidence of the importance of directional insolation on crack formation. The variability in the preferred orientations suggests that aspect, shading, and the time of day of cracking (see below) play a role in the orientation of fractures in the different sites.
What will we do in the field?
IOur plan of field work is designed to determine how the numbers, size, location and orientation of rock fractures change with rock sun exposure, exposure age, size and shape. These field data are integral to testing and refining future numerical models and to testing the hypothesis that variability in crack characteristics can be correlated to rock properties that would influence insolation and thus thermal-related cracking. Based on our preliminary field results and the McFadden et al. (2005) study, a suite of crack and rock field data will be collected for ~1500 boulders on Quaternary age surfaces in arid sites. We focus on boulders rather than outcrops to eliminate crack orientation bias towards tectonic joints, and to avoid having to examine complicated differences in thermal stresses arising for differing outcrop geometries.
McFadden et al (2005) indicated that insolation-related cracking should generally not affect clast much smaller than the maximum depth of penetration of diurnal temperature variations. Their study, however, focused on cracks in boulders and only 25 of the ~400 clasts studied had diameters less than 10 cm. More data for small clasts will be necessary in order to document the minimum clast size below which diurnal insolation cracking is ineffective. At least some previous research in the Sahara Desert (Soleil et al., 1995) strongly supports the notion of a minimum-size threshold.
In addition to crack orientation data, McFadden et al. (2005) made several initial observations concerning correlations between cracking and rock shape. For example, even in rocks with no apparent fabric, there was a strong correlation with rock elongation and crack orientation (e.g. 54% of clasts with aspect ratios >2 exhibited one or more cracks parallel to the long axis, whereas 6% of clasts with ratios <1.5 exhibited cracks parallel to the long axis). In addition, preliminary data collected by Eppes and her students from a single large boulder in both an arid and a humid setting revealed that the total numbers of cracks varies significantly as a function of surface aspect. A possible interpretation of these data is that rock surface-sun exposure plays a role in crack initiation and orientation. If this is true, then rock shape and the orientation of non-spherical rocks with respect to diurnal or seasonal insolation variation should favor cracking parallel to some faces and not others. More detailed field data that includes rock shape and orientation data as well as our numerical modeling and instrumentation should enable us to examine the influence of rock shape on cracking by insolation-related stresses.
McFadden et al. (2005) also observed that crack width and length appeared to generally increase with the age of the surface; however, there were insufficient data to draw more specific conclusions concerning cracking and surface age. There was no documentation of youngest and/or oldest age of surface that contains a significant population of north-south cracks; or of the change in the total numbers of cracks and their size and shape as a function of age. The age control available for surfaces in our proposed arid field site should enable us to document in more detail the effects of surface exposure on long-term cracking processes.
Finally, the 5º mean resultant vector for McFadden et al. (2005) crack-orientation data does not comprise a single mode (Fig. 1a). Rather, it arises from two modes (NE & NW) of crack orientations clustered symmetrically about the north-south direction. If the formation of these cracks is linked to directional solar heating, then different crack orientations may correspond to different seasons or time-of day. For example, at latitudes equivalent to those in the Southern U.S. between sunrise and mid-morning during a significant fraction of the year (summer), the most strongly heated part of round clast surfaces is not that directly facing east or west. Instead, the southeast side of the clast will be most strongly heated in the morning. Cracks that would result from these ‘morning’ stresses would be oriented NE-SW, which is one of the observed dominant orientation modes of the desert and South Carolina crack populations (Fig. 1a and c). The mirror image population of cracks whose orientation is NW-SE could thus be the result of temperature differences that occur in the evening when the sun is in the southwest. This mode is more dominant in the Virginia crack population, which were collected on a hillslope with a west-facing aspect. In the case of this Virginia data-set, the ‘morning’ population of cracks may be absent because these boulders are shaded from morning sun. Our chosen field sites will offer us an opportunity to assess how cracking varies as a function of varying thermal forcing (different ambient temperature means and extremes, varying duration of periods of sunlight etc.).