Soil Thermal Properties: How Ground Stores and Conducts Heat

Every day the Sun heats the ground surface, and every night the surface cools. Every year the same thing happens on a longer cycle. How far those waves of warmth and cold reach into the ground, how quickly they travel and how much heat the ground absorbs along the way depend on a small set of physical quantities known as soil thermal properties. They decide whether a seed germinates in spring, how deep a water main must be buried to avoid freezing, how hot an underground power cable runs, how many meters of borehole a ground-source heat pump needs, and how the land surface exchanges energy with the atmosphere in a climate model.

Wikipedia's article places soil thermal properties within soil physics and notes their importance in engineering, climatology and agriculture. This article goes further: it explains each property, why soil water dominates them, how they are measured in the laboratory and the field, how they are predicted from soil composition, what numbers to expect, and how they are used. The calculations shown are worked examples with labeled assumptions, based on property values from published tables.

The three main properties

Soil thermal behavior is usually described with three quantities, together with a fourth that combines them.

Property	Symbol (common)	SI unit	What it describes
Thermal conductivity	λ or k	W·m⁻¹·K⁻¹	How readily heat flows through the soil under a temperature gradient
Volumetric heat capacity	C	J·m⁻³·K⁻¹	How much heat a cubic meter of soil must absorb to warm by one kelvin
Thermal diffusivity	α or D	m²·s⁻¹	How quickly a temperature change spreads; equals conductivity divided by volumetric heat capacity
Thermal effusivity	e	J·m⁻²·K⁻¹·s⁻¹ᐟ²	How strongly the surface resists a change in temperature; equals the square root of conductivity times volumetric heat capacity

Thermal conductivity

Thermal conductivity is the constant in Fourier's law of heat conduction, which says that heat flux is proportional to the temperature gradient. For vertical heat flow in soil, Wikipedia gives it as

Q = −λ · dT/dz

where Q is the heat flux in watts per square meter, λ is the thermal conductivity and dT/dz is the temperature gradient with depth. The minus sign means heat flows from warm to cold.

Volumetric heat capacity

Volumetric heat capacity is the product of a material's density and its specific heat capacity. Because soil is a mixture of mineral grains, organic matter, water and air, its volumetric heat capacity is approximately the sum of the contributions of each constituent weighted by its volume fraction. Air contributes almost nothing. Water contributes a great deal: its specific heat is about 4,186 to 4,188 joules per kilogram per kelvin, which with a density of 1,000 kilograms per cubic meter gives about 4.19 million joules per cubic meter per kelvin. Soil mineral solids are commonly assigned about 1.93 million joules per cubic meter per kelvin in land-surface modeling code.

Thermal diffusivity

Diffusivity combines the other two. A soil with high conductivity but also high heat capacity may have only moderate diffusivity, because the heat it conducts is soaked up in warming each layer. Diffusivity is the coefficient that appears in the heat diffusion equation,

∂T/∂t = D · ∂²T/∂z²

which Yi Dong, John McCartney and Ning Lu derive in their 2015 critical review of soil thermal conductivity models by combining Fourier's law with an energy balance. They define D = k/(ρ·c) and describe it as reflecting how fast heat will be transferred through a material.

Thermal effusivity

Effusivity governs how the soil surface temperature responds to a given heating at the surface. A soil with high effusivity (wet, dense) heats up less in the sunshine for the same absorbed energy, because heat is both conducted away and stored efficiently. A soil with low effusivity (dry, loose) heats up quickly at the surface but the heat does not go deep. The Wikipedia article groups several applications, including underground thermal energy storage and heat pump design, under the heading of thermal effusivity.

Why soil thermal properties vary so much

Three phases with very different properties

Soil is a three-phase material: solid particles, liquid water and gas, mostly air. The thermal conductivities of these phases span about two orders of magnitude. Dong, McCartney and Lu give typical values of more than 3 W·m⁻¹·K⁻¹ for mineral particles, 0.56 for water at 0 °C, and 0.026 for air. Quartz in particular is an excellent conductor among common soil minerals; values of 7.7 and 8.4 W·m⁻¹·K⁻¹ appear in different references.

Constituent	Thermal conductivity (W·m⁻¹·K⁻¹)	Notes
Quartz	7.7 (Johansen method), 8.4 (Bejan and Kraus handbook table)	Most conductive common soil mineral
Other soil minerals	2.0 to 3.0 in the Johansen method	Varies with mineral; above 3 in Dong et al.'s summary
Ice	2.25 to 2.29	Roughly four times liquid water
Water	0.56 to 0.57
Air	about 0.026	Poorest conductor
Dry soil (bulk)	usually below 0.5	Air gaps between grains dominate

The bulk conductivity of a soil therefore depends on how much of each phase is present and, crucially, how they are arranged. Dong and colleagues summarize the ordering as conductivity of air < dry soil < water < saturated soil < mineral.

The role of water

Wikipedia's article notes that air is a poor conductor and reduces the effectiveness of the solid and liquid phases, and that variability in moisture largely determines thermal conductivity. The physical picture is this. In a dry soil, heat must cross from grain to grain through tiny contact points or through air-filled gaps. Dong and colleagues point out that the conductivity of dry soil is almost one order of magnitude lower than that of the mineral solids, below 0.5 W·m⁻¹·K⁻¹ in most cases, which implies that air obstructs conduction and that heat moves mainly through particle contacts.

Add a little water, and it collects first as thin films and as small bridges, or menisci, at the contacts between grains. Those bridges have far higher conductivity than the air they replace, and they sit exactly where they do the most good: in the narrow necks of the heat path. Dong and colleagues note that at low water contents adding a small amount of water dramatically improves thermal conduction. As the soil gets wetter, the gains continue but more slowly, until at saturation all the pore space is water and the conductivity reaches its maximum for that soil.

Water also raises the volumetric heat capacity, steadily and nearly linearly with water content, because water's heat capacity per unit volume is roughly twice that of mineral solids. The two effects on diffusivity partly cancel. Conductivity rises steeply at first, then levels off; heat capacity rises steadily. As a result diffusivity often rises with water content at low moisture, peaks at intermediate moisture and then falls slightly toward saturation.

Other governing factors

Dong, McCartney and Lu list the particle-level and macro-scale factors that govern soil thermal properties under isothermal conditions: mineralogy, particle size, particle shape, packing geometry, stress level, water content, porosity, gradation and cementation. In brief:

Factor	Effect on conductivity
Mineralogy	Higher quartz content gives higher conductivity; quartz conducts better than mica
Particle size	Larger particles with fewer contacts per volume give higher conductivity
Gradation	Well-graded soils conduct better, since small particles fill gaps and increase contacts
Packing and porosity	Denser packing and lower void ratio give higher conductivity
Stress	Higher stress enlarges contact areas and raises conductivity
Cementation	Cement and precipitates at contacts enlarge contact area and raise conductivity
Water content	The dominant factor in partially saturated soils

They also note that convection through the pore space becomes important if the median particle size is larger than about 6 millimeters, so that fluid can flow through the porous network. For ordinary fine-grained soils, conduction dominates.

Organic matter, gravel and freezing

Organic matter has relatively low thermal conductivity and high heat capacity, while gravel has high conductivity. Dai and colleagues, in a 2019 evaluation of soil conductivity schemes for land surface modeling, note that most older schemes considered only fine mineral soils with particles smaller than 2 millimeters and that the effects of organic matter and gravel have to be added explicitly. Peat soils are an extreme case: a frost-modeling report based on U.S. Army Cold Regions Research and Engineering Laboratory (CRREL) methods assigns dry peat a conductivity of 0.05 W·m⁻¹·K⁻¹ unfrozen and saturated peat 0.55 unfrozen and 1.80 frozen.

Freezing changes everything. Ice conducts heat roughly four times as well as liquid water, so a saturated soil generally becomes more conductive when it freezes. Its heat capacity changes too, and the latent heat released or absorbed during freezing and thawing dominates the energy balance near 0 °C. That is why cold-regions engineering treats frozen and unfrozen soils with separate property sets.

Typical values

Published tables give ranges rather than single values, because each soil's composition, density and moisture are different. Some representative figures:

Soil condition	Thermal conductivity (W·m⁻¹·K⁻¹)	Source
Dry sand	0.15 to 0.25	ISSMGE conference paper table
Moist sand	0.25 to 2	ISSMGE
Saturated sand	2 to 4	ISSMGE
Dry clay	0.15 to 1.8	ISSMGE
Saturated clay	0.6 to 2.5	ISSMGE
Saturated sand (model input)	2.4	Geothermal Energy journal table
Dry sand (model input)	0.4	Geothermal Energy journal table
Saturated clay (model input)	1.7	Geothermal Energy journal table
Dry clay (model input)	0.5	Geothermal Energy journal table

A heat transport study published in Hydrology and Earth System Sciences in 2018 used, as model inputs, 2.8 W·m⁻¹·K⁻¹ and 2.8 million J·m⁻³·K⁻¹ for saturated sand, 1.4 and 2.5 million for saturated clay, and 0.5 and 1.6 million for dry clay.

For comparison, rocks and rock-forming materials listed in a handbook table reproduced by Dong and colleagues include granite at about 1.7 to 4.0 W·m⁻¹·K⁻¹, sandstone at about 1.6 to 2.1, marble at 2.8, limestone at about 1.26 to 1.33 and a clay at 1.28.

Temperature waves in the ground

The daily and annual cycles

Wikipedia notes that soil properties change with daily and seasonal cycles, that temperature swings are largest at the surface and weaken with depth, and that peak and minimum temperatures arrive later at greater depths, an effect called thermal lag. These observations follow directly from the diffusion equation.

If the surface temperature varies sinusoidally, as it roughly does over a day or a year, the solution of the diffusion equation in a uniform soil is a damped wave traveling downward. The amplitude falls off exponentially with depth z as exp(−z/d), and the timing lags by z/d radians, where

d = √(2D/ω)

is called the damping depth. Here D is the diffusivity and ω is the angular frequency of the cycle: 2π divided by the period. Longer periods penetrate deeper, in proportion to the square root of the period. Since a year is about 365 days, the annual wave penetrates about √365, or roughly 19 times, deeper than the daily wave.

Worked example: damping depths

Using the 2018 HESS study's property values, the diffusivity of saturated sand is 2.8/2.8 million, or 1.0 × 10⁻⁶ m²/s; saturated clay is 1.4/2.5 million, or 5.6 × 10⁻⁷ m²/s; dry clay is 0.5/1.6 million, or about 3.1 × 10⁻⁷ m²/s. Applying the damping-depth formula:

Soil (values from HESS 2018)	Diffusivity (m²/s)	Daily damping depth (m)	Depth where daily swing falls to 1% (m)	Annual damping depth (m)	Depth where annual swing falls to 1% (m)
Saturated sand	1.0 × 10⁻⁶	0.17	0.76	3.2	14.6
Saturated clay	5.6 × 10⁻⁷	0.12	0.57	2.4	10.9
Dry clay	3.1 × 10⁻⁷	0.09	0.43	1.8	8.2

The depth for a 1 percent residual swing is the damping depth multiplied by the natural logarithm of 100, about 4.6.

These calculated depths agree well with a penetration-depth table in Wikipedia's article on geothermal heat pumps, which gives the depth at which the variation falls below 1 percent of the surface value as 0.80 meters daily and 14.5 meters annually for wet sand, 0.95 and 18.0 for wet clay, 0.40 and 6.5 for dry clay, 0.30 and 4.5 for dry sand, and 1.10 and 20.5 for rock.

The lag is just as striking. In the saturated sand case, the phase lag of the daily wave at 0.5 m depth is about 0.5/0.17 radians, which corresponds to roughly 11 to 12 hours: the soil half a meter down is warmest around the time the surface is coldest. For the annual wave at 2 meters, the lag is about five weeks. The geothermal heat pump article notes that in the shallow layer, peak temperatures can lag the surface by about six months at sufficient depth.

What this means in practice

A few tens of centimeters of soil are enough to smooth out the day-night cycle almost completely, which is why root zones and burrows are thermally sheltered. A few meters are enough to strongly damp the seasonal cycle, which is why cellars stay cool in summer and frost-free in winter in many climates, and why horizontal ground loops for heat pumps are buried at 0.91 to 2.44 meters (3 to 8 feet), below the frost line, according to the geothermal heat pump article. Below roughly 8 to 20 meters, depending on soil, the ground temperature settles close to the local mean annual air temperature, then rises slowly with depth along the geothermal gradient, which the same article gives as about 0.025 °C per meter. In the northern United States, it gives undisturbed ground temperatures of roughly 7 to 12 °C at 6 meters.

Measuring soil thermal properties

Wikipedia's article identifies in situ measurement with non-steady-state heat probes as the most common approach, and describes single-probe and dual-probe methods along with remote sensing.

The single needle probe

A single probe is a thin needle containing a heating wire and a temperature sensor such as a thermocouple. It is inserted into the soil and heated at a constant rate. Heat conducts away radially, and the temperature at the needle rises. After an initial transient, the temperature rise becomes proportional to the logarithm of time, and the slope of that line is inversely proportional to the soil's thermal conductivity: a highly conductive soil carries the heat away quickly and the needle warms slowly; a poorly conductive soil lets the needle heat up faster.

This is the soil version of the transient hot-wire method. The method's history goes back to Joseph Priestley's crude heated-wire experiment on gases in 1781; Sven Pyk and Bertil Stålhane proposed the first transient hot-wire method for solids and powders in 1931. The single probe measures thermal conductivity only. Wikipedia lists two standards that cover it: IEEE Standard 442-1981, from the Institute of Electrical and Electronics Engineers, and ASTM D5334-08, an ASTM International thermal needle probe test method for soil and soft rock. The Wikipedia article mentions a complete field system designed for measurements at about 1.5 meters, the typical burial depth of high-voltage cables.

The dual-probe heat-pulse method

A dual probe has two parallel needles a fixed distance apart, one containing a heater and the other a temperature sensor. A short pulse of heat is released from the heater, and the sensor records how the temperature rises and falls as the pulse diffuses across the gap. The time to the temperature peak depends on the diffusivity, and the size of the peak depends on the volumetric heat capacity. From the two, conductivity follows. The dual probe therefore determines all three main properties, as Wikipedia notes.

Because the volumetric heat capacity depends almost linearly on water content, a dual probe can also be used to estimate soil moisture. Bristow, Kluitenberg, Goding and Fitzgerald described a small multi-needle probe in 2001 in Computers and Electronics in Agriculture that measures soil thermal properties, water content and electrical conductivity together. The drawbacks Wikipedia lists are the small measurement volume and sensitivity to probe-soil contact and to the exact spacing between heater and sensor. A small error in that spacing produces a large error in heat capacity, because the standard analysis uses the square of the spacing.

Steady-state methods

In the laboratory, conductivity can also be measured by establishing a steady temperature difference across a soil sample, as in a guarded hot plate, and measuring the heat flux. Steady-state methods need no complex signal analysis but take a long time to settle, and in moist soil the sustained temperature gradient can drive water vapor and liquid away from the warm side, changing the very property being measured. That is one reason transient probes became standard for soils.

Thermal response tests

For ground-source heat pump boreholes, the thermal properties that matter are those of the whole formation along tens to hundreds of meters of borehole. A thermal response test circulates heated fluid through a test borehole for many hours and analyzes the fluid temperature over time to infer an effective formation conductivity. The geothermal heat pump article says such tests are often used to size the loopfield properly, especially for commercial sites with more than about 10 wells. Wikipedia's soil article adds that in the United Kingdom, the Microgeneration Certification Scheme (MCS) design criteria for shallow-loop ground-source heat pumps require an accurate in situ thermal conductivity reading, which heat probes can provide across a site.

Remote sensing

Satellites and aircraft offer a way to see thermal variation over large areas. Wikipedia notes that reflected visible light reveals only the topmost few molecular layers of the soil, while thermal infrared reveals energy variations at shallow depths, and that microwave sensing has the advantage of not being blocked by cloud cover, unlike thermal infrared. Remote sensing cannot measure conductivity directly, but by observing how quickly surface temperatures rise and fall over a day, it can infer properties such as effusivity, which in turn relate to soil moisture.

Predicting soil thermal conductivity

Because measuring every soil is impractical, a large literature has developed on predicting thermal conductivity from soil composition. Dong, McCartney and Lu group the models into three families.

Mixing models

These treat soil as a combination of solid, water and air blocks arranged in series, in parallel or in some combination. The parallel arrangement (an arithmetic average weighted by volume fraction) gives an upper bound, and the series arrangement (a harmonic average) gives a lower bound. These are called the Wiener bounds, after Wiener's 1912 analysis. They are very wide when the phases differ greatly, as soil phases do. More refined mixing models include Mickley's 1951 unit-cell model, Woodside and Messmer's 1961 three-path model, McGaw's 1969 model and Gori and Corasaniti's 2002 cubic-cell model, each with assumptions that make it suitable for some conditions and not others.

Empirical models

The most widely used family began with Kersten's extensive tests on clays, silts and sands, published in 1949, which produced separate empirical equations for silt-clay soils and sandy soils based on water content and dry density. Dong and colleagues report that Kersten's equations can be off by about 25 percent or more for soils high in silt and clay and are valid only within the range of water contents tested.

Johansen's 1975 thesis introduced the approach that became the standard. Instead of fitting conductivity directly, Johansen interpolated between a dry value and a saturated value using a normalized conductivity, the Kersten number:

Ke = (k − kdry)/(ksat − kdry)

which he related to the degree of saturation Sr with simple logarithmic functions. In the forms used by CRREL-based frost modeling:

coarse unfrozen soils: Ke = 0.7·log₁₀(Sr) + 1.0
fine unfrozen soils: Ke = log₁₀(Sr) + 1.0
frozen soils: Ke = Sr

The dry and saturated end points are computed from soil properties. The dry conductivity is a function of dry density γd (in kilograms per cubic meter):

kdry = (0.135·γd + 64.7)/(2700 − 0.947·γd)

The saturated conductivity is a geometric mean of the solids and water conductivities weighted by porosity n: ksat = ks^(1−n) · kw^n, with kw = 0.57 W·m⁻¹·K⁻¹ for unfrozen water. The solids conductivity itself is a geometric mean of quartz (7.7) and other minerals (2.0, or 3.0 for coarse soils low in quartz), weighted by quartz content.

Omar Farouki's 1981 CRREL monograph, Thermal Properties of Soils, compared 11 early schemes, including de Vries's 1963 model and Johansen's, and found that Johansen's gave the most accurate predictions across a range of soil types and saturation levels, except for dry (Sr below 0.2), unfrozen, coarse soils. Later refinements built on the Kersten-number idea: Côté and Konrad (2005) used a hyperbolic relation with a fitted parameter and a new dry-conductivity equation; Lu and colleagues (2007) used an exponential relation; Balland and Arp (2005) proposed another widely used derivative.

Mathematical and theoretical models

These borrow formulas developed for other properties of mixtures, such as electrical conductivity or magnetic permeability. They include the Hashin-Shtrikman bounds of 1962 and 1963, which are the narrowest possible bounds for an isotropic mixture given only volume fractions; effective medium theory; and de Vries's 1963 model, which treats soil as a continuous medium with dispersed ellipsoidal inclusions and weights each phase by shape factors.

Which model works best?

There is no universal winner. Dong and colleagues tested several models against data for a sand, a silt and a clay and found, for example, that de Vries's model had the lowest normalized error for the sand, Côté and Konrad's for the silt and Lu's for the clay. They concluded that none of the existing models realistically accounts for pore structure or interface properties, and proposed a conceptual model based on how soils retain water. Dai and colleagues, testing seven schemes inside the Common Land Model, found that Balland and Arp's 2005 scheme consistently performed best for land surface modeling and that the choice of scheme had significant effects on simulated soil temperature in dry regions but little effect in wet regions, where high heat capacity dominates.

Worked example: Johansen's method for a sand

Consider a hypothetical sandy soil with a porosity of 0.40 and a quartz content of 60 percent, treated as a coarse soil.

Dry density: γd = 2,700 × (1 − 0.40) = 1,620 kg/m³
Dry conductivity: kdry = (0.135 × 1,620 + 64.7)/(2,700 − 0.947 × 1,620) = 283.4/1,165.9 ≈ 0.24 W·m⁻¹·K⁻¹
Solids conductivity: ks = 7.7^0.6 × 2.0^0.4 ≈ 4.49 W·m⁻¹·K⁻¹
Saturated conductivity: ksat = 4.49^0.6 × 0.57^0.4 ≈ 1.97 W·m⁻¹·K⁻¹

The Kersten number and resulting conductivity at different saturations:

Degree of saturation	Kersten number	Predicted conductivity (W·m⁻¹·K⁻¹)
0.05	0.09	0.40
0.10	0.30	0.76
0.25	0.58	1.24
0.50	0.79	1.60
0.75	0.91	1.82
1.00	1.00	1.97

The table reproduces the behavior described earlier. Going from 5 percent to 25 percent saturation triples the conductivity; going from 50 to 100 percent adds only about a quarter. Keep Farouki's caution in mind: Johansen's method is least reliable for very dry coarse soils, where the logarithm becomes unreliable.

The volumetric heat capacity of the same soil, using 1.926 million J·m⁻³·K⁻¹ for solids and 4.188 million for water, is 1.16 million when dry, 1.57 million at a volumetric water content of 0.10, 1.99 million at 0.20 and 2.83 million when saturated (water content 0.40). Heat capacity rises steadily while conductivity rises steeply then flattens, which is why diffusivity peaks at intermediate moisture.

A short history of the subject

The study of soil heat draws on two older traditions, the mathematics of heat conduction and the engineering of foundations, pavements and buried services. The milestones mentioned in this article can be gathered into a single timeline.

Year	Development
1781	Joseph Priestley's heated-wire experiment on gases, an early ancestor of hot-wire and needle-probe methods
1912	Wiener's parallel and series bounds for the conductivity of mixtures
1931	Pyk and Stålhane propose the first transient hot-wire method for solids and powders
1949	Kersten publishes extensive measurements and empirical equations for clays, silts and sands
1951	Mickley's unit-cell mixing model
1961	Woodside and Messmer's three-path model
1962 to 1963	Hashin and Shtrikman's bounds for composite materials
1963	De Vries's model of soil as a medium with dispersed ellipsoidal inclusions
1969	McGaw's model with an interfacial efficiency factor
1975	Johansen's thesis introduces the Kersten-number interpolation method
1981	IEEE Standard 442 for soil thermal resistivity; Farouki's CRREL monograph compares 11 schemes
2001	Bristow and colleagues describe a small multi-needle probe for thermal properties, water content and electrical conductivity
2005 to 2008	Côté and Konrad, Balland and Arp, Lu and colleagues, and Chen publish refinements of the Kersten-number approach
2015	Dong, McCartney and Lu's critical review concludes that no existing model captures pore structure and interfaces realistically
2019	Dai and colleagues evaluate seven schemes inside a land surface model and recommend Balland and Arp's

Two themes run through this history. The first is a steady move from steady-state laboratory measurements to fast transient probes that can be used in the field and repeated as conditions change. The second is a shift from purely empirical curve fitting toward models that try to represent how water actually sits in the pores, a goal that Dong and colleagues argue has still not been fully reached.

Applications

Buried power cables

High-voltage underground cables generate heat from their electrical losses, and the surrounding soil must carry it away. If the soil's thermal resistivity (the reciprocal of conductivity) is high, the cable runs hotter and its permissible current, its rating, must be reduced. Worse, the heat from the cable can dry out the soil around it, raising the resistivity further in a feedback that can lead to thermal runaway. This is why IEEE 442 exists, why measurement systems are designed for the typical 1.5-meter burial depth, and why cable trenches are often backfilled with engineered material of known, stable thermal properties.

Ground-source heat pumps

Ground-source heat pumps exchange heat with the ground through buried horizontal loops or vertical boreholes. The geothermal heat pump article reports vertical boreholes typically 15 to 122 meters (50 to 400 feet) deep, spaced at least 5 to 6 meters apart, and gives the example of a detached house needing 10 kilowatts (3 tons) of capacity, which might require three boreholes 80 to 110 meters deep. Detailed analysis of soil or formation conductivity generally yields more accurately sized and more efficient systems, and wet ground performs better than dry ground because water conducts and stores heat better. Undersizing leads to the ground temperature drifting over the years; oversizing wastes drilling cost.

Underground thermal energy storage

Wikipedia's article describes using the ground to store summer solar heat for use in winter, with ground-source heat pumps retrieving it. The design question is how much heat the ground can store (heat capacity), how fast it can be charged and discharged (conductivity) and how much leaks away (diffusivity and the size of the store).

Frozen ground and pavement design

In cold regions, the depth to which frost penetrates determines how deep foundations, water pipes and drainage must go, and how pavements must be built to avoid frost heave. Frost penetration depends on soil thermal conductivity in both frozen and unfrozen states, heat capacity and the latent heat of the soil water. This is the context of CRREL's work, including Farouki's monograph, and of the frost-modeling methods that use Johansen's equations. Wikipedia also mentions expansion and contraction of construction materials in freezing soils as an engineering concern.

Agriculture

Seed germination and early root growth depend on soil temperature. Wikipedia's article notes that thermal properties affect the timing of planting for good seedling emergence and crop growth. Practices that change soil moisture, mulching, residue cover or tillage all change how quickly the seedbed warms in spring. A dark, dry, loose soil warms quickly at the surface but stores little heat, while a moist, compact soil warms more slowly but holds its warmth into the night.

Climate and land-surface models

Every weather and climate model needs to represent how the land surface stores and releases heat. Dai and colleagues showed that in dry regions, differences between soil conductivity schemes can produce significant differences in simulated soil temperature, especially at depth, by changing downward heat conduction and freeze-thaw cycles, while in wet regions the high heat capacity of moist soil makes the choice of scheme matter less. They found the sensitivity largest over arid and semi-arid regions at middle and high latitudes.

Soil carbon

Wikipedia's article links soil heat to carbon dioxide release, citing a 2010 Nature Geoscience paper by Allison, Wallenstein and Bradford, titled "Soil-carbon response to warming dependent on microbial physiology." Soil respiration by microbes is sensitive to temperature, so how warming at the surface propagates into the soil, and how soil moisture affects that propagation, bears on how soils respond to a changing climate.

Other uses

Wikipedia also lists tracking water movement around radioactive waste sites and locating buried land mines. Mines and other buried objects have different thermal properties from the surrounding soil, so the soil surface above them heats and cools slightly differently over the day, which thermal imaging can detect under the right conditions.

Common misconceptions

"Wet soil is cold because water is cold"

Wet soil warms slowly in spring because water greatly increases the soil's heat capacity, and because evaporation from a wet surface uses energy that would otherwise warm the soil. The water is not colder in itself.

"Dry soil insulates best because it has the lowest conductivity"

Dry soil does have low conductivity, but it also has low heat capacity, so its diffusivity can be similar to or even higher than a moist soil's. Whether dry soil "insulates" depends on whether you care about steady heat flow (conductivity) or the damping of temperature swings (diffusivity and effusivity).

"Soil thermal properties are constants"

Wikipedia stresses that the properties change constantly with the daily and seasonal cycles, because moisture, and sometimes freezing, change. A single laboratory value describes one moisture state.

"More water always means more conductivity, in proportion"

Conductivity rises steeply with the first small additions of water and then much more slowly. The relationship is strongly nonlinear, as Johansen's logarithmic Kersten number captures.

"The ground below a few meters is always 10 °C"

The temperature at depth is close to the local mean annual air temperature, which varies by climate, then rises with depth along the geothermal gradient. In the northern United States the geothermal heat pump article gives about 7 to 12 °C at 6 meters; elsewhere it differs.

Frequently asked questions

What is the most important factor controlling soil thermal conductivity?

For a given soil, water content. Between soils, mineralogy (especially quartz content), density and particle size also matter.

What is thermal resistivity?

The reciprocal of thermal conductivity. It is the quantity traditionally used in power cable engineering, often expressed in kelvin-meters per watt.

How deep does the daily temperature cycle reach?

Typically a few tens of centimeters. Calculated 1 percent penetration depths for typical soils range from about 0.3 to 0.8 meters, and the geothermal heat pump article's table gives 1.10 meters for rock.

How deep does the annual cycle reach?

Several meters to about 20 meters, depending on soil and rock type.

Why do needle probes need good contact?

Because the measurement depends on heat crossing from the needle into the soil. An air gap around the needle adds a resistance that the analysis attributes to the soil, making it appear less conductive. Dual probes are also sensitive to the spacing between needles.

Can soil thermal properties be estimated without measurement?

Yes, with models such as Johansen's, Côté and Konrad's or Balland and Arp's, using porosity, density, quartz or sand content and water content. Accuracy varies; Johansen's own work reported that dry-soil conductivity could be calculated within about plus or minus 20 percent using porosity alone, and Kersten's older equations can be off by 25 percent or more for silty and clayey soils.

Why do engineers care about soil drying around a buried cable?

Because drying lowers conductivity, which makes the cable hotter, which dries the soil further. The rating of a cable must account for the worst plausible soil condition.

Summary

Soil thermal properties describe how ground conducts, stores and spreads heat. Thermal conductivity, volumetric heat capacity and diffusivity, along with effusivity, depend on mineral composition, density, structure and above all water content, because water replaces poorly conducting air at the grain contacts and adds heat capacity. The result is that the same soil can vary several-fold in conductivity between dry and wet states. Temperature waves from the surface penetrate a few tens of centimeters per day and several meters per year, with timing lags that grow with depth. Engineers measure the properties with needle probes, heat-pulse probes and borehole response tests, and predict them with mixing, empirical and theoretical models, of which Johansen's 1975 approach and its successors are the most widely used. Those numbers feed into decisions about cable ratings, heat pump boreholes, frost protection, planting dates and climate simulations.