Some thoughts on the usefulness of Lineament Maps

“Of what use at all is a lineament map?” constitutes a fair question posed with positively indecent irregularity, if ever at all. Here we suggest that because topographically derived lineament maps depict landscape elements that form under physically differing processes at different rates and times, they have historically conflated time-transgressive morphological evolution of the region at hand with a simplistic message, stated or implied, akin to “this map, then, represents the structural framework underlying such-and-such geological province.” This slippery slide down the razor presupposes two essential conditions: That incision occurs only where the substrate is most easily eroded and that all faults, fractures, and shear zones are less resistant than undeformed rock. It is a demonstrable fact that neither condition is necessarily true. Yet lineament mapping persists, from Earth to Mars and by now, presumably, on even more distant planets. What can we do with these things, and why do we bother?

Whilst quantifiable linearity does convey information that may be useful for landscape classification, abstracting topographic surfaces into lineaments does not, a priori, expose the structural template of the underlying bits of any given planet. Analyses of azimuth, density, length, or any other quantity provide little benefit unless one can constrain exactly what it was that one measured. Digital vectorization of modern DTM data does offer hope. Furthermore, linearity can be expressed in topographically positive features such as ridgelines. Individual vectors can (must!) be given local attributes (depth… width… sinuosity… slope…) that may be coupled to a postulated (theoretical, probably optimistic, and in most cases surely a relative) morphologically dependent ‘age’ of incision. Regional (‘environmental’) attributes pertaining to bulk properties of the area traversed by that very same vector (flat… inclined… U or V shape… carbonate… granite…) can be collected to provide external context. Equally important is a careful inventory of bias such that the mapping method generates reproducible vectors with representative and homogeneous coverage from the upper left-hand corner of the dataset to both the penultimate and ultimate pixels at the (assume southeastern) End Of File. Absent these, a lineament map is not dissimilar to a basket of tropical fruits plucked from one’s local Arctic haberdashery at or around Christmastime.

Because she exhibits a wide range of topographic styles, human-generated lineament maps have struggled to extract unbiased, homogeneous signals from Baltica’s tired old bones. Having experimented for some years with algorithms designed to map lineaments automatically from (x, y, z) data we feel inclined to present a few observations, interpretations, confessions of bias, and recommendations for how we might possibly do better. We will briefly describe one successful algorithm for computerized lineament mapping, present results that purportedly describe distinctly different Norwegian landscapes, illustrate some connections to certain known structures and disconnections with others, and attempt a sort of general, undoubtably conflicted synthesis of the potential use of lineament maps in assessing the landscape evolution of certain small parts of the Norwegian rifted margin.

“There are more things in heaven and Earth than are dreamt of in our Science….”