Revised Origin of Cold Talus Habitats in the Columbia River Gorge and Implications for Dependent Species

Author: Steve Stampfli, White Salmon, WA, USA,

Low-elevation talus slopes of the western Columbia River Gorge appear to be geologies that can sustain cold and possibly permafrost conditions usually found at either higher elevations or more northerly latitudes.  The formation of likely sporadic permafrost is due the unusual thermal behavior of blocky-rock deposits that cause them to lose heat in winter at a higher rate than heat can be regained during summer.  As a notable result, these cold geologies / habitats support populations of cool-loving and stenothermal animals including American pika and members of the Grylloblatta insect genus.  It is furthermore possible that the process has resulted in significant deposits of underground ice, which may constitute unknown water supplies important to human economies and aquatic life in the Gorge.

As highlighted in Ice Mountain —  A Theory of Why Pikas Exist in the Columbia River Gorge, the situation in the Gorge is not unique, as there are other geologies across the globe that display the formation of cold talus and potentially permafrost in areas that have annual average air and soil temperatures that are well above 0°C (32°F).  Based on extensive work to describe the phenomena at places like Creux-du-Van in the Jura Mountains of Switzerland, scientists have theorized that the creation of localized sporadic permafrost is a result of the “chimney effect” thermodynamic process.  But before examining that term, and whether it adequately explains the processes occurring in our talus slopes, a quick look at the term “permafrost” is warranted.


Permafrost is simply defined as a patch of the earth’s regolith that is frozen, and has been frozen for at least two consecutive years.  And to be clear, regolith is defined as the layer of unconsolidated rocky material including soil between the earth’s bedrock and its atmosphere or waters.

There are four main permafrost zones on earth:  a) zone of sub-sea permafrost; b) zone of continuous land permafrost; c) zone of discontinuous land permafrost; and d) zone of alpine land permafrost.   These delineations are shown on the below map taken from Pewe’s “Alpine  Permafrost in the Contiguous United States:  A Review”.   Cross hatched areas represent the zone of continuous permafrost, hatched areas represent discontinuous permafrost, and solid black color represents alpine permafrost.  The continuous and discontinuous permafrost zones occur well north of the states of Oregon and Washington, generally above 50° north latitude.  While frozen soil in these zones is generally dictated by proximity to the North Pole, alpine permafrost can form much further south, if at increasingly higher elevations.  These areas are mapped in Figure 1, and several patches of alpine permafrost can be seen in our area (i.e., near 45° north latitude) corresponding to Mt. Rainier, Mt. Adams, Mt. St. Helens and Mt. Hood.  It is generally assumed that permafrost in all three zones can only form in places where either high latitudes or elevations result in annual average temperatures of −2°C (28.4°F) or colder.

Figure 1.  Pewe’s map of permafrost distribution across North America.  See above paragraph for legend shading definitions. 

But not all patches of perennially frozen ground on earth are found in the zones pictured above, which are all proximate to the poles or at high elevations.  These outliers are lumped together into a fifth category, called sporadic permafrost, and occur on the equatorial sides of the earth’s bands of discontinuous permafrost, or locally below the elevations of alpine permafrost.

Formation of Sporadic Permafrost in the Gorge

The existence of currently active (as opposed to relict) sporadic frozen patches in areas with annual average temperatures above −2°C must be explained in all cases by mechanisms that cause a relatively higher loss of heat (i.e., cooling) from the blocky-rock feature to ambient environment in winter, than is gained back from the surrounding ambient environment (i.e., warming) in summer.  Perpetuation of even very small seasonal net losses over time can, therefore, result in landform temperatures that are well below annual average ambient temperatures, and even below freezing.  For these mechanisms to operate, however, there must be avenues for free air passage into the soil/rock mantle.  Indeed, I would predict it impossible for sporadic permafrost to develop in fine or mixed grained soils, or solid rock landforms.

To understand how this seasonal thermodynamic imbalance can arise in either cave or blocky-rock landforms, it is useful to remember a very simple maxim which applies to wintertime overcooling:  “hot air rises vertically as a gas, while cold air flows downward as a liquid”.  Both of these represent very efficient heat transfer processes occurring in winter.

Case 1.  The Ice Cave and other Topographically-Contained Landform Features

The simplest illustration of this maxim is what has been called the “Balch effect”, a process that operates in the case of a confined, downward-inclined earth opening.  This situation was perhaps first described by Edwin Swift Balch in his 1900 book titled Glacières, and is best exemplified in our area by the Trout Lake and associated Big Lava Bed lava cave systems.  During the coldest periods of winter (dominated by stable high pressure systems, clear nights and little regional winds), very cold layers of air exist at the land’s surface.  Once formed, this dense, stable, heavy air behaves as a liquid, and flows down-slope in a very non-turbulent manner perpendicular to landscape contours.  When this cold layer encounters any depressions in the earth surface (whether a broad open basin, confined cave, blocky-rock deposit, etc.), it flows in and displaces any air that is even minimally warmer, less dense, and therefore more buoyant.  This obviously causes the upward displacement of any warmer air occupying the depression, resulting in a temperature inversion (i.e., a layer of cold air trapped next to or within the ground).

Topographically-depressed reservoirs of cold inverted air may or may not be stable depending on the ambient surroundings and the nature of containment.  Inverted air masses, in general, are extremely stable as long as they remain “capped” and isolated from outside heating influences, whether caused by the rising of warm air (thermal cell convection), advection due to regional winds, direct conduction from the earth or atmosphere, solar radiation, air pumping due to barometric pressure changes, etc.  Taking the example of cold air trapped in a desert basin, the stable inversion can be rapidly “uncapped” during mornings by solar heating of the basin’s surface, which in-turn results in upward rising convection cells, mixing, and warming of the once stable and overly-cooled lower atmosphere.

The absence of such “uncapping” mechanisms in the case of the protected ice cave perpetuates the existence of inverted conditions and the reservoir of cold interior air.  Once emplaced during the coldest periods of winter, the stable layer can persist throughout the summer season, aided by the large thermal mass of bedrock, lack of solar inputs, minimal downward conduction of heat from the surface, and absence of upward mixing convection.   The same thing can be occur in other depressed landscapes with openings, including topographically-contained blocky-rock deposits.

Case 2.  The Talus Slope and other Topographically-Uncontained (Sloping) Landform Features

It is more difficult to explain how a similar imbalance might occur in sloping blocky-rock geologies, including talus slopes.  Such geologies are the second landform type seeming capable of forming cold temperatures and potentially sporadic permafrost at low elevations in the Gorge.  The cause of the winter vs summer heating imbalance in such landforms is more complex than in the ice cave, given the vast numbers of multi-level air openings.  The deposits are also generally sloping and uncontained, therefore, seemingly prone to rapid gravity discharge of cold and dense air (and excessive warming) during summer.

Chimney effect diagram

Figure 2.  Conventional “chimney effect” diagram from the Sébastien Morard, Reynald Delaloye, and Christophe Lambiel 2010 article in Geographica Helvetica titled “Pluriannual thermal behaviour of low elevation cold talus slopes in western Switzerland”.

A currently accepted explanation for the formation of “cold talus” and likely permafrost at the base of some blocky slopes is known as the “chimney effect”, explained in the earlier article Ice Mountain:  A Theory of Why Pikas Exist in the Columbia River Gorge.   In short, the chimney effect results in differential heating and cooling of the blocky-rock slope, driven by seasonal differences in temperatures inside vs outside the deposit (see Figure 2 above).  As a result of these temperature and therefore density differences, the theory predicts that relatively warm subsurface air “chimneys” up-contour to the top of the slope in winter, thus forcing the simultaneous suction of cold stable air into the base of the slope.  This action can result in the formation of unusual open tunnels where heated air is output from the top of the snow covered slope.  The rate of cold air inhalation is directly related to the temperature and density differential, thus the velocity and volume of air intake is highest when outside temperatures are coldest.  In summer, air flow direction is reversed, when cold air sinks and flow back out of the bottom of the slope due to its relatively higher density than the outside air.  As this occurs, an equal volume of warm air is presumably inhaled into the upper slope.  This is thought to result in winter overcooling, and perpetually cold conditions inside the lower base of the blocky-rock slope.

There is no question whether chimney action causes seasonal transfers of air and heat wherever there are underground spaces with multi-level openings (e.g., talus slopes, some cave networks, etc.).  On its own, however, the usual chimney effect description doesn’t seem to explain why the total slope heat loss during winter is higher than total slope heat gain in summer.  Superficially, it would appear that the heat content of cold air being drawn into the uncontained slope’s base in winter should equal the amount lost by the discharge of cold air from the base in summer.  However, as in the case of ice caves, blocky-rock slopes display heat gain/loss imbalances that can result in perpetual cold talus and permafrost.  The following two sections will attempt to describe how this might occur.

Local Observational Evidence

Two side-by-side thermodynamic processes appear to drive the very efficient loss of heat from uncontained, blocky-rock landforms during winter, both understood using the simple maxim “warm air rises up as a gas, while cold air sinks down as a fluid”.

The long-wave infrared (LWIR) images below illustrate recently observed surface temperature effects present on talus during both the summer heating and winter cooling seasons.  Both frames show the northwest facing flank of Shellrock Mountain, Oregon, lying near sea-level just south of the Columbia River in the western Gorge.

The following observations related to summer heating period are made after examination of Figure 3.  First, the drainage of dense cold air (blue color) from the slope is immediately apparent along the lower margins of the talus skirt.  (It is hidden below the slope transect on this image due to tree canopy interference).  Second, above the cold air vent, there is almost no overall bottom-to-top trend in summer surface temperature along the slope transect.  Third, there is little variability in slope temperatures, resulting in a fairly smooth temperature graph that lacks spiky bumps and dips.  Surface temperatures are limited within a narrow 1.7°C range.


Different observations are apparent during the winter cooling period, as illustrated in Figure 4.  First, and as expected, the drainage of cold air from the bottom of the slope is absent due to internal temperatures now being above outside temperatures.  Second, as also seen in summer, there is little overall trend in surface temperatures from bottom-to-top of the slope transect.  Third, winter surface temperature variability across the transect is high, as evidenced by the graph below the image that ranges between 3.9 and 7.0°C  (3.1°C range).  This observation is best realized when viewing the very spiky nature of the temperature graph.  Finally, the image does not show indication of the presence of large patchy warm air vents on upper portions of the unbroken talus plane.   This supports the contention that warm air is not being conveyed upslope and within the blocky-rock mass, as the chimney effect theory predicts.  While large warm air vents are present on the left side of the Figure 4 image, these vertically-elongated winter features are clustered at the bottom and middle portions of the talus mass, and not at the top of the slope.  The large vents shown occur at topographically-uneven faces of the slope, characterized by convex ridges and concave rills associated with past slope movements.  They are not, therefore, considered relevant to consideration of the chimney effect, or the theory currently offered.

The above observational evidence indicates that significant winter overcooling is being driven by the formation of a multitude of small vertically-rising convection cells originating within the talus, which occur across the entire breadth of the unbroken talus field.  The vertically-rising convection cells are indicated by the individual spikes seen along the winter fall-line transect graph.  Each spike likely represents a column of rising warm air that is very efficiently conveying heated air vertically-upward and out of the slope. By my calculation (and pending possibly unknown limitations of the LWIR camera’s resolution), the average on-ground distance between the presumed convection cells equals 16 feet.  This equates to a density of 170 convection cells per acre.

The winter image also hints that the vertically-rising convective heat transfers are coupled to vertically-downward fluid-like intakes of cold and dense winter air into the landform between convection cells.  These induction areas are likely indicated by the troughs seen on the graph.  The high cooling efficiency of this arrangement somewhat assumes that the closely proximate upward and downward trending air parcels do not significantly intermix either underground or above the surface.  During cold and calm winter weather  patterns, it may be possible that the convection cells persist in a confined upward direction well after they emerge from the talus surface, thus allowing the proximate downwelling of dense cold air with little intermixing.  If this is the case, the effect may be identical to the up-and-down convective movements of air parcels during thunderstorms, where there is little intermixing.

Assuming the above two linked mechanisms drive the primary winter cooling process, it is not likely that significant amounts of cold air are being inducted into the base of the slope, as the chimney effect model predicts.  Instead, it appears that the entire blocky-rock deposit is being cooled from surface to base, as one uniform mass, during winter.  If bottom-of-slope intake does happen, it would probably only be forced when a continuous blanket of snow is present, creating a closed conduit for the warm and low-density air being buoyed upward.

It is also necessary to touch upon why summer heating mechanisms are comparatively weaker than the cooling mechanisms of winter.  This is fairly easy to understand in the relatively simple case of the ice cave.  Once warm air is displaced by the down-gradient flow of cold air into the contained and “capped” underground cavity, the cold inverted air is very stable in temperature and volume.  In a relatively small cave, the only means of introducing summer heat and uncapping of the inversion would be via the very inefficient downward conduction of ambient heat from above.

The inefficiency of summer heating in a sloping blocky-rock landform such as a talus slope is, however, more difficult to explain than in the contained ice cave example.  The difficulty stems from the fact that talus deposits are inclined and lack the bowl-like topographic containment of an ice cave; therefore, any cold air reservoirs forming in winter above the bedrock surface would tend to flow out of the slope’s base in summer.  While this unquestionably occurs (see evidence of lower-slope cold venting in Figure 3), the fact that this does not result in a high degree of summer heating is likely because the sub-surface flow of the inverted air mass is greatly slowed by physical resistance within the porous talus.  This resistance is compounded by the long distance the cold subsurface air must follow from top to bottom of the talus slope interior.  The inverted air mass contained in the base of a cold talus slope, therefore, might be comparable to a perennial lake having a restricted overflow/outlet channel, which due to its configuration lacks the ability to rapidly drain, and is thus fairly stable in volume.

Finally, note that this same internal resistance is likely responsible for the earlier hypothesized low rate of inhalation of cold air into the base of the slope in winter.  It is presumably much more efficient that the intake of cold air occurs vertically and downward, and immediately proximate to the multitude of vertically-acting warm air convection cells.  A vertically downward pathway for winter air intake is logically the shortest, least resistant, and most efficient pathway.

Applied Engineering Evidence

Civil engineers working in northerly latitudes have long recognized that roads, pipelines and other structures built upon relatively warm permafrost are subject to failure upon melting of the permafrost layer and deflation of the local landscape surface.  Melting is a consequence of replacing native soil and vegetation layers with non-porous compacted earth and pavement, and therefore, loss of the temperature buffering and cooling mechanisms important to maintaining the permafrost layer.

To counteract the loss of permafrost after construction, D.J. Goering at the University of Alaska Fairbanks has experimented with systems known as air convection embankments for protecting roadways and other structures built on warm permafrost.  Such systems involve perching structures atop a layer of porous gravel or rock having low fines content and narrow particle size gradation.  In the beginning, such construction may have been believed simply a means of insulating the frozen ground surface from the summer’s heat.  Eventually, however, Goering came to realize that a passive refrigeration effect was largely responsible for maintaining frozen ground conditions in the underlying soil layers.

To understand the apparently passive cooling mechanism, Goering built an experimental eight-foot thick, unconfined porous rock embankment near Fairbanks, and laced it with electronic temperature probes placed in a three-dimensional grid pattern.  After monitoring the internal and external temperatures for two years, and then modelling the internal heat flows, he was able to decipher how ground temperatures at the base of the embankment were being maintained below freezing.  The experimental findings closely match the LWIR imagery observations and the ideas presented in the last section of this article.

The following quote from the 2003 paper titled Thermal response of air convection embankments to ambient temperature fluctuations, effectively summarizes Goering’s findings:

“During the winter, the embankment is cooled at its upper surface due to low ambient air temperatures.  If the cooling is strong enough and the embankment material is of sufficient permeability, natural convection of the pore air will occur during winter months due to the unstable pore-air density gradient that develops.  The convection can transfer heat upward out of the embankment at a rate that may be more than an order of magnitude larger than conductive heat transfer, resulting in greatly enhanced winter cooling.  During summer the pore-air density gradient is stable and convection does not occur. Thus the embankment acts as a one-way heat transfer device or thermal diode that effectively removes heat from the embankment and underlying foundation material during winter without re-injecting heat during subsequent summers”.

Diagrams showing the isotherm lines and convection cell boundaries discovered by Goering illustrate the internal processes likely responsible for over-cooling the interior of Columbia River Gorge talus slopes.  These diagrams are presented below as Figures 5 and 6.  The non-spiky summer surface temperatures earlier shown in Figure 3 are explained by the stable and laminar sub-surface thermal patterns depicted in the experimental embankment shown in Figure 5 below.  Likewise, the more spiky surface temperature range seen during winter in Figure 4 is explained by the formation of the vertically-oriented internal convection cells seen in Figure 6.



Recent results of the Shellrock Mountain LWIR imagery work, combined with experimental evidence reported by Goering in 2003, support the conclusion that vertically-oriented, cell-confined “pore air convection”, and proximate subduction of cold and dense winter air is responsible for the existence of cold and sometime frozen ground in open, blocky-rock deposits of the Columbia River Gorge.  Very significant temperature declines can occur within blocky-rock slopes and embankments within a period of a few days or even hours during cold periods. These rapid events represent periods of intensely non-linear cooling, and have a disproportionally large impact on ground temperatures when averaged over the year.  The experimental evidence also indicates that gravity-driven outflow of cold air from a slope or embankment base in summer is a relatively inefficient means of landform warming, which does not necessary result in the melting of permafrost.  Even lacking the topographic containment of an ice cave, the inverted cold air mass remains relatively stable within the sloping blocky-rock deposit during summer.  This is probably due to high internal resistance to summer gravity-driven outflow of the stable inverted air system, as it flows downward along the long talus slope / bedrock interface.

Given the likelihood that pore air convection is the main mechanism responsible for winter over-cooling and sporadic permafrost formation within sloping blocky-rock deposits, I suspect that the winter-time chimney effect can be viewed as an inefficient manifestation of pore air convection occurring under snow layers.  In such circumstance, the very efficient co-existence of upward trending warm air convection cells with side-by-side cold air downwelling is interrupted by the snow layer.  As warm air rises within the blocky-rock deposit and nears the rock / snow interface, it is forced to follow an upslope path through the porous rock just below the snow surface.  Perhaps more important is the fact that the volume of warm air displaced upward is not being directly replaced by cold downwelling air from the open atmosphere.   Instead, rising warm air trapped below the snow layer causes a slow suction of cold air into the slope via transfers through the snow layer, and into the base of the slope.  The resulting mixture of inside / outside air then moves upward below the snow / rock interface at a relatively low velocity.  The rate of winter slope cooling is, therefore, significantly retarded by a) the low volumes of cold and warm air that can transfer through the snow cover, b) the high flow resistance encountered by rising air being sucked upward within the porous rock deposit, and c) the fact that the external cold air and internal warm air are being constantly mixed within the porous slope just below the snow layer.

The above conclusions indicate that the creation of sporadic permafrost and periglacial landforms is most likely to occur in zones characterized by low snowfall accumulation and periods of cold winter temperature.   Such conditions were at a maximum in the late Pleistocene, some 16,000 to 20,000 years ago throughout the Columbia Basin of the Pacific Northwest.  During that time, climates were dominated by cold and dry continental air masses (high pressure systems), that blocked entry of wet maritime weather systems onto the continent.  There is also evidence that lesser such conditions existed into the Holocene period, and may be still present today, evidenced by sporadic permafrost scattered across the Pacific Northwest.  In the case of the western Columbia River Gorge, the occurrence of scattered permafrost can be partially attributed to the fact that the region exhibits large masses of porous talus occurring in low-elevation snow-free zones, which is favorable to efficient pore air convection.  Equally important, however, is the fact that dry and cold continental weather patterns are still very dominant aspects of our winter weather.  Each winter, we experience periods when calm and cold air settles into the Gorge from the east, resulting in prolonged periods of sub-freezing temperature.  These periods, which Goering has shown to have the ability to cause intensely non-linear cooling of blocky-rock deposits, must largely account for the occurrence of our cold and likely frozen blocky-rock slopes.

Finally, the apparent overwhelming dominance of non-linear winter cooling within blocky-rock landforms indicates that long-term plant and animal habitat temperature trends are not simply dictated by annual ambient temperatures, or even by annual high temperatures.  It is therefore impossible to conclude that simple rises in regional or global average temperatures can be directly correlated to an increased threat to populations of stenothermal talus or cave dwellers.  Instead, experimental findings reported in this article indicate that the long-term thermal stability of blocky-rock habitats is dictated by the number and length of periods when temperatures drop below specific levels.  This, of course, greatly complicates any attempt of forecasting long-term population trends or threats of species declines for organisms dependent upon such geological habitats.  This is not to say that such threat analysis is impossible for rock and cave dwelling organisms such as pikas and grylloblattids, but only that conclusive answers will likely depend upon multi-disciplinary efforts involving geologists, meteorologists, climate scientists, statisticians, biologists and physicists committed to the work.

Steve Stampfli

21 May 2019 at White Salmon, Washington

– END –

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s