Like the much more visible and famous rivers of flowing glacial ice that have been in overall retreat since the end of the last Ice Age around 13,000 years ago, permafrost around the world is also in retreat.  In this current interglacial period the retreat of permafrost is generally slower and much less obvious but in each case, the place of the retreat is usually an interesting, and sometimes hazardous place to be.  A tidewater glacier, for example, is famous for producing spectacular vistas of calving ice and turquoise water strewn with floating hunks of ice.  Such places represent boundaries between the retreating frozen glacial ice and the advancing mostly liquid sea and the contrast between the two is usually stark and impressive and these places are a popular vacation excursion the world over. 


As a crude analogy, in the same way that a tidewater glacier is a boundary separating the retreating glacial ice from the advancing, mostly liquid seawater; the place of permafrost retreat also represents a boundary between permafrost terrain and terrain that is free of permafrost.  One big difference is that in permafrost areas, such boundaries aren't always nearly so obvious as they are at the face of a retreating glacier.  But in both cases, the presence of the boundary can produce some spectacular results.  


In each case, the boundary between frozen and non frozen conditions isn't a fine line.  The liquid tidewater lapping away at the retreating glacial face is only MOSTLY liquid and scattered masses of breakaway ice can still be found floating in the liquid water for some distance away from the glacier's retreating face.  When these masses of ice are found floating in the sea, sometimes hundreds of miles from the glacier or ice sheet that bore them we call them icebergs and depending on their size, they can take months to thaw even after being exposed to direct sunlight and to non-frozen seawater. 

In the case of the permafrost boundary, interestingly enough and for different reasons, scattered masses of ice can also sometimes be found some distance from the retreating permafrost face.  And they too can take time to thaw.  Sometimes, especially when they are well-insulated by Fairbanks loess and are shaded by dense forest, they can take a long time to thaw.  

When scattered masses of ice are encountered in mostly non-frozen ground, calling them icebergs is just crazy talk.  So many permafrost enthusiasts simply call them massive ground iceArguably, not frightfully creative, but it gets the point across.  One thing that permafrost boundaries have in common is that the scattered masses of ice can take large structures and cause them to sink.  Each event in its respective domain can be pretty spectacular.

In each setting, the reasons for the existence of the ice are the similar: thaw at each retreating face doesn't occur uniformly everywhere.  Pan back far enough from each boundary and you do eventually get an imaginary line like that shown in the image below.  But up close the line becomes curvy and broken and lobes or isolated masses of permafrost are found near the boundary spatially, in much the same way that icebergs are found floating in the water some distance away from the retreating face of the glacier.  And like icebergs that are calved from ice sheets or glaciers, ground ice deposits can be holdovers from a cooler climate.  In the case of the glacier, movement of the iceberg away from the glacier or ice sheet by the ocean currents is responsible for the rapid change in climate (and thaw) seen by the iceberg.  But in the case of retreating permafrost, the naturally induced climatic changes are much slower.  Under the right conditions, ground ice deposits can be many tens of thousands of years old.  But clearing the ground surface overlying permafrost can also change the "climate" that the permafrost sees at an increased rate.  Finally construction of a warm foundation (in the interior of Alaska) instantly and dramatically changes the climate that the permafrost sees as ground temperatures below the warm building change from sub-arctic (average of 27F) to tropical (average of 60F) the moment the building is heated.    

Excerpt from USGS mapping showing an approximate permafrost boundary. The boundary is estimated and is intended to seperate permafrost terrain from terrain that is generally permafrost free. It may not account for lobes or isolated masses of permafrost (or non-frozen conditions) on either side of the boundary.
In each case, the impact that these masses of ice can have on man-made structures can be devastating so an awareness of their existence and their location is critical.  On the sea, masses of ice drifting away from polar ice sheets or glaciers are usually (but not always) fairly easy to spot but have still been known send very large ships to the bottom of the ocean.  On land, the masses of ice (or ice-rich frozen soil) near permafrost boundaries usually not easy to spot from the surface and they are sometimes completely hidden, buried in the earth.  But they don't sink ships.  They lie in wait for the introduction of heat from a warm foundation.  Then they sink the building.

Fairbanks lies at around 63 degrees north latitude and the climate of the area is sub-arctic.  The average annual air temperature hovers just a few Fahrenheit degrees below freezing.  Fairbanks lies within a region of discontinuous permafrost.  Permafrost in the area has been retreating since the last ice age and so permafrost boundaries are common.  


The images below illustrate an actual residential construction site located near a permafrost boundary in Fairbanks.  The site was located on the non-permafrost side of a boundary similar to the mapped example pictured above.  The photos were taken from a site located in a different area than that which is depicted on the map above.  


Owing to past experience in the area, the orientation of the site (located on a northwest-facing slope) and the close proximity of the site to a known permafrost area, nine boreholes were used to gather preliminary information about subsurface conditions on the property. 


Three of the nine borings were located downhill of the preferred construction site to investigate suspicious uneven surface terrain observed there.  Permafrost consisting of frozen silt containing variable clear ground ice was encountered in each of these borings.  The remaining boreholes were located upslope, in the area preferred for construction.  Four of these six borings were located near the four corners of a rectangular building footprint with the remaining two borings being drilled at the mid-point of each wall on the long dimension.  Permafrost was not identified in any of these boreholes.


As can be seen below in the first photo (on the left), the predominant soil type at the site was loess (windblown silt) and the predominant vegetation was mature birch forest.  While permafrost was not identified in any of the six borings drilled within the building footprint, isolated masses of permafrost were a concern between boreholes and careful observations were recommended during foundation construction.  Excavation for a daylight basement was begun and was nearly complete at the time of the first site visit.


During the first site visit, careful observations were made of the exposed soil conditions in the excavation.  As expected, the excavation revealed mainly Fairbanks loess (windblown silt) as indicated by the borings.  This material is typically massive (meaning that individual depositional layering is not easy to see) and homogenous (meaning that it's material consistency is uniform in every direction).  The silt is tan to buff in color and is usually naturally weakly cemented such that it holds its shape, and steep excavation cuts are possible.


During the visit, some staining was observed in the wall of the excavation near one corner of the building footprint which suggested a historic presence of water (see second photo from the left).  The staining was observed in an area of the excavation that was adjacent to an apparent drainage gulley which ran across the property from higher elevations of the hill.  During the observations, the author walked the excavation boundary and probed the soil at the base of the excavation with a 48 inch long surveyor's stake.  At the footing location at the base of the excavation immediately adjacent to the staining, the stake could easily be pushed by hand to its full length into the soil.  For the remainder of the foundation location around the building footprint, the stake could only be pushed by hand one or two inches.


A test pit was recommended at the location of the probe and revealed massive ground ice.  The deposit was located between the two borings drilled at the building corners.  The boreholes were spaced around 30 feet apart.  The property is surrounded by residences supported by conventional concrete foundations. 


A careful study of the ground ice and its surroundings indicated that the ice was associated with the nearby drainage and was likely the result of ground water that had seeped into a cast in the silt made by a previous ground ice deposit which had thawed and drained away during a previous warming period, leaving the open void (cave) behind.  The climate cooled again and shallow surface water trickled into the cave and re-froze.  Evidence supporting this hypothesis was seen in cores of ground ice (sixth photo from left)  taken from a borehole located downslope of the preferred building site as well as the massive ice exposed in the excavation shown (third, fourth, and fifth photos from left).