JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. E4, 8032, doi:10.1029/2002JE001864, 2003
 This section briefly reviews the possible formation of the features seen in Figure 2. The word “debris” refers to material, whether from weathering or the products of rockfall events, for the rock detritus which is involved in the formation of these features, hence the more general term debris-related or debris-derived forms.
3.1. Rock Glaciers
 The three main models of rock glacier formation have been proposed and are discussed in detail by Whalley and Martin . These are a permafrost origin, a glacier-derived origin, and a mass-wasting (landslide) origin. The first two involve the creep of ice held in the body of the feature whereas the third model may involve, but does not require, the presence of ice. In summary, these models have the following properties.
3.1.1. Permafrost Model
 The permafrost model for rock glacier formation follows the ideas of Wahrhaftig and Cox  and has been promulgated in particular by Barsch  and Haeberli . The “congelation” ice is formed from freezing water, either by ice segregation or water injection under pressure. A pre-requisite is a mean annual air temperature of, at most, -1.5°C. This thermal condition implies a “zonal” occurrence of rock glaciers and this attribute has led to the use of rock glaciers as being indicators of permafrost, both present and relict [Barsch, 1996]. The presence of any glacier ice which plays a part in the formation of rock glaciers is generally disputed by adherents to this model. The literature often implies that rock glaciers necessarily have a permafrost origin.
3.1.2. Glacial Model
 The glacial model (for a comprehensive review, see Whalley and Martin ), relies on the preservation of a thin (generally <50 m) body of ice by an insulating weathered rock debris layer. The ice is considered to be derived from glacial, i.e., “sedimentary” sources. The thin ice creeps, giving a typically low velocity and the debris preserves this in an otherwise ablation-dominant environment. The controls on maintaining this buried ice are thus related to thickness of debris cover as much as local climate (measured by, e.g., degree-day estimates). As such, they are “azonal” features and cannot be used to delimit temperature regimes such as the presence of permafrost.
3.1.3. Landslide Model
 The landslide, or “catastrophic,” model [Johnson, 1974, 1984] has used similarity of topographic form to suggest that rock glaciers may be derived from rapid landslides/rock avalanches (Bergsturtz or Sturtzstroms) [Whalley, 1976; Whalley and Martin, 1992]. These will generally be forms which do not flow after emplacement. However, it has been recognized that some Bergsturz have fallen on retreating/down-wasting glaciers and so have produced “instant” rock glaciers. This is a variation of the glacier ice cored model rather than the landslide model [Whalley, 1976]. In the case of fossil rock glaciers, it may not be easy or possible to distinguish between these origins.
3.2. Protalus Lobes and Other Ice/Debris-Derived Features
 Two other components of ice plus debris now need to be considered. Some have argued that they are part of a continuum of features, which includes rock glaciers. All of the three possible modes of formation mentioned above could be included in this continuum [Shakesby et al., 1987].
3.2.1. Protalus Lobes
 Protalus lobes (Figures 3, 4a, 4b, 4c, and 4d), usually being away located away from glacier ice, are generally accepted to be of nonglacier origin, although the ice could originate as snowbanks and then become buried by debris from cliffs above. They do not then necessarily require permafrost conditions. The preservation of these features may again depend upon azonal conditions such as debris thickness cover, aspect and altitude as well as thermal conditions. However, the large numbers of these features in high latitudes suggests that permafrost may be a sufficient, although not necessary, condition for their formation. Finite element modeling shows the extremely low creep rates of these features and the dependence of surface velocity upon the size of ice bodies contained, their disposition and the depth of burial in the ice-debris mass [Azizi and Whalley, 1995, 1996]. This modeling supports the, relatively few, observations on velocities of protalus lobes [e.g., Sollid and Sørbel, 1992].
3.2.2. Protalus Ramparts
 Protalus ramparts (Figures 2, 4a, 4b, 4c, 4d) are generally attributed to debris accumulating at the front of snow patches or even small “glacierettes.” This association with glacial conditions would place them in the glacial, rather than permafrost, realm (i.e., the ground temperature may not necessarily be <c 1.5°C). However, there have been suggestions [Barsch, 1996] that they are incipient rock glaciers of permafrost origin. Again, it may be that both could be realistic models, according to local antecedents and contingent factors, and they have been considered as part of a continuum of landforms [Shakesby et al., 1987]. The choice of rock glacier model has an impact on the interpretation of Martian forms. In particular, the possibility of massive ice bodies, derived from glaciers, is fundamentally different from the necessity of permafrost for the formation of rock glaciers [Barsch, 1996]. Because of the divergence of opinion about ice presence in terrestrial rock glaciers and protalus lobes, the next section illustrates the diversity of ice locations in ice-debris features.
Citation: , Rock glaciers and protalus landforms: Analogous forms and ice sources on Earth and Mars, J. Geophys. Res., 108(E4), 8032, doi:10.1029/2002JE001864, 2003.
Copyright 2003 by the American Geophysical Union.