This is not a rigorous treatment of this subject, but rather an administrative overview.
Volumetric behavior of wet high plastic clay during and after freezing is important to understand when building small surface structures which are partly supported on such conditions. When adjacent structures are attached and founded below the effect of freezing soils and moisture change, there will be differential movement that occur annually. In addition, if some of the clay is uncontrolled back fill and or fill, more variation is introduced.
There are two separate issues of volume change with freezing, phase change and secondly the development of ice lenses or frost heave. The second requires freezing front, as "frost susceptible" soil, and a source of water. High plastic clay is not normally considered as frost susceptible soil since the hydraulic conductivity is quite low; however, if organics are present, all bets are off.
For this mental exercise we need to consider the soil as a five phase system, rather than the normal three. Solid, liquid and void, as normal, but the water is also ice, liquid and water vapor. Water increases in volume about 9% in the phase change. Water vapor moves through the soil toward the freezing front; it is drawn by substantial thermodynamic forces; the vapor freezes at the freezing front, and shows up as ice crystals. The same happens in a frost susceptible soil, except it is liquid water that moves, and forms ice lenses. This thermodynamic force draws water into the soil, increasing the saturation. This suction has very little impact on the void ratio, at this stage. Also, in most cases, water is found with depth.
Where saturation occurs before freezing, the water turns to ice, and increases in volume by 9 %. This does not happen abruptly but rather as a rapid change, as there is a temperature gradient, and super cooled liquid water due to water-soil adhesion. Very soon, as cooling continues, the water is fully frozen. The amount of expansion can be best understood by considering the void ratio (volume of voids/volume of soil solids) and the degree of saturation (approximately volume of water/volume of voids). All the physical state properties; density, moisture content, are directly linked to void ratio and degree of saturation. In the end, wet clay expands on freezing. If a soil was 50% voids, and fully saturated, the expansion would be 4.5 %. Where the ground is laterally restrained, the only it can go is up, therefore in this case the amount of uplift would be approximately 4.5% of the depth of frost penetration. The amount of expansion is dependent on the soil, moisture, and depth of freezing, which in turn, is dependent on ground cover, amount and duration of cold, wind, building heat loss, and other factors.
Buildings are designed to stay more or less stable, while everything on the ground outside will move seasonally with the frost. This includes grade supported slabs, steps, walkways. Building supported steps typically do not move as much. The design and construction of these units does not account for these movements, when combined with settlements.
Fill and loose materials settles. It is dependent on moisture, grain to grain stress and strength. When it is wet, saturated, the settlement is time dependent, and is termed consolidation, and is characterized with the expulsion of water, greater density, lower void ratio, lower moisture content, greater strength and the like. When it is less than saturated, it compacts with the expulsion of air. It tends to stay in lumps with voids between the lumps until it gets water and then settles quickly to some degree. It is never finished until it becomes saturated and consolidates. Now for the ugly part, consolidation has pressure dependent end point; it consolidates until the strength is equal to the imposed load. It will change end points when the pressure changes. It may also dry, which increases the strength and stop consolidation. The whole process is water supply dependent, which is not predictable. Consolidation it's self is time dependent, but not the supply or removal of water.
Now that we know what consolidation is about, during freezing, water is drawn in, and some expansion occurs. Each spring, after melting, drainage may occur, and the frost moved materials can consolidate back to a new state of equilibrium. One area is not necessarily the same as the next. Ramps may be high in the center, like the center swelled, another trick of high plastic clay, when it gets water. It may be that the edges were backfilled after the retaining walls were constructed, while the center was native or well compacted clay.