Heat Flow and Thermal Modelling
CSIR NET Geophysics Part C: Fourier's law, heat equation, heat-flow measurement, radiogenic geotherms, half-space and plate cooling, skin depth, Peclet number.
- 20 min
On this page · 15 sections
- Prerequisites
- The Hook: Kelvin’s Clock
- Heat Transfer: Conduction, Advection, Radiation
- Measuring Heat Flow
- Continental Geotherms
- Oceanic Lithosphere: the Cooling Half-Space
- Time-Dependent Problems You Can Do in a Line
- Thermal Structure of Subduction Zones
- Memory Tricks
- Don’t Confuse
- Common Mistakes
- Practice
- Quick Revision Table
- Key Takeaways
- What’s Next
Prerequisites
- Gravity, Geomagnetism, Heat Flow and Isostasy: Fourier’s law, global heat-flow values, \( q_0 = q_r + DA_0 \), the uniform-heat-production geotherm and isostasy. This page derives and extends them.
- Marine Magnetic Anomalies, Sea-Floor Spreading and Ridges: the age–depth law and hydrothermal circulation at ridges.
- The diffusion equation and the error function. The heat equation is the same mathematics as in Physical Sciences; only the numbers are geological.
The Hook: Kelvin’s Clock
Heat Transfer: Conduction, Advection, Radiation
| Mode | How | Where it matters |
|---|---|---|
| Conduction | molecular / lattice vibration, down the temperature gradient | lithosphere; thermal boundary layers |
| Advection / convection | heat carried by moving material | mantle and outer core; magma ascent; hydrothermal and groundwater flow |
| Radiation | photons | adds to effective conductivity at high temperature in the deep mantle; negligible near the surface |
Fourier’s law and the heat equation
$$\boxed{\mathbf{q} = -k\nabla T}$$\( k \) is thermal conductivity: about 2–3 W m\(^{-1}\) K\(^{-1}\) for crustal rocks and about 3–4 for mantle rocks. Conservation of energy for a volume with heat production \( A \) (W m\(^{-3}\)) moving with velocity \( \mathbf{v} \):
$$\boxed{\rho c_p\left(\frac{\partial T}{\partial t} + \mathbf{v}\cdot\nabla T\right) = k\nabla^2 T + A}$$Dividing by \( \rho c_p \) gives the thermal diffusivity:
$$\boxed{\kappa = \frac{k}{\rho c_p} \approx 10^{-6}\ \text{m}^2\,\text{s}^{-1}}$$(for \( k = 3.3 \), \( \rho = 3300 \), \( c_p = 1000 \) in SI units, \( \kappa = 10^{-6} \) exactly).
The two numbers that decide everything
Diffusion time. Heat diffuses a distance \( L \) in a time of order
$$\boxed{\tau \sim \frac{L^2}{\kappa}} \qquad \text{or, conversely,} \qquad L \sim \sqrt{\kappa t}$$| \( L \) | \( \tau = L^2/\kappa \) (with \( \kappa = 10^{-6} \)) |
|---|---|
| 1 m | \( 10^6 \) s, about 12 days |
| 1 km | \( 10^{12} \) s, about 32 kyr |
| 100 km | \( 10^{16} \) s, about 320 Myr |
| 3,000 km (mantle) | about \( 3\times 10^{11} \) years: far longer than Earth’s age |
The last line is the essence of Kelvin’s error: conduction alone cannot have cooled the deep Earth. Something faster must carry heat.
Peclet number. It compares advection with conduction over a length \( L \) at velocity \( v \):
$$\boxed{Pe = \frac{vL}{\kappa}}$$\( Pe \gg 1 \): advection dominates and isotherms are carried with the flow. \( Pe \ll 1 \): conduction dominates. Mantle upwelling beneath a ridge at 3 cm yr\(^{-1}\) over 100 km gives \( Pe \approx 95 \): advection wins.
Measuring Heat Flow
$$\boxed{q = k\,\frac{dT}{dz}}$$(with \( z \) downward, reported as positive upward flux). Both factors must be measured.
On land. Temperatures are logged in boreholes (after the drilling disturbance has decayed) and conductivity is measured on core samples. Corrections are needed for:
- topography (isotherms follow the surface in a subdued way);
- paleoclimate (the last glaciation’s cold surface still perturbs the upper hundreds of metres to kilometres; see skin depth below);
- erosion and sedimentation;
- groundwater flow, which can advect heat and distort gradients.
At sea. A probe several metres long is pushed into soft sediment, with thermistors along it (Bullard-type and “violin-bow” probes); some probes also measure conductivity in situ by a heat pulse. Bottom-water temperature variations can perturb shallow measurements.
Layered conductivity: the Bullard plot. In steady state without heat sources \( q \) is the same in every layer. Define the thermal resistance from the surface:
$$\boxed{R(z) = \sum_i \frac{\Delta z_i}{k_i}} \qquad T(z) = T_0 + q\,R(z)$$A plot of \( T \) against \( R \) is a straight line whose slope is \( q \). Temperature gradients are steep in low-conductivity layers (shale, sediment) and gentle in high-conductivity ones (quartzite, salt), but \( q \) is continuous.
What the global data show
| Quantity | Approximate value |
|---|---|
| Mean continental heat flow | about 65 mW m\(^{-2}\) |
| Mean oceanic heat flow | about 100 mW m\(^{-2}\) |
| Global mean | about 87 mW m\(^{-2}\) |
| Total heat loss | about 44–47 TW |
| Share through the oceans | about two-thirds |
Continents: heat flow decreases with the age of the last tectonic or thermal event: high in young orogens, rifts and volcanic areas, low over Archaean cratons. Part of this is radiogenic (crustal heat production differs), part is the slow decay of the thermal disturbance, and part reflects the thick, cold lithospheric roots of cratons. In India, geothermal areas such as Puga in Ladakh stand out against the low heat flow typical of the shield.
Oceans: heat flow falls with sea-floor age, as the half-space model predicts, except near ridges (below).
Continental Geotherms
Uniform heat production
For steady state, no flow, conductivity \( k \), heat production \( A \) and surface heat flow \( q_0 \):
$$k\frac{d^2T}{dz^2} = -A \quad\Rightarrow\quad \boxed{T(z) = T_0 + \frac{q_0}{k}z - \frac{A}{2k}z^2}$$The curvature means the gradient decreases with depth; the heat flow decreases as \( q(z) = q_0 - Az \).
Exponentially decreasing heat production
The linear relation \( q_0 = q_r + DA_0 \) observed in heat-flow provinces is reproduced by \( A(z) = A_0 e^{-z/D} \), which keeps the relation valid after differential erosion. Solving:
$$\boxed{q(z) = q_r + A_0 D\,e^{-z/D}} \qquad \boxed{T(z) = T_0 + \frac{q_r}{k}z + \frac{A_0 D^2}{k}\left(1 - e^{-z/D}\right)}$$At the surface \( q(0) = q_r + A_0D \), the observed relation. Well below \( D \), the crustal heat production adds a nearly constant \( A_0D^2/k \) to the temperature, and the gradient tends to \( q_r/k \).
Lithospheric thickness from geotherms. The base of the thermal lithosphere is often taken where the conductive geotherm meets the mantle adiabat (about 1,300 °C). Low heat flow (cold geotherm) means thick lithosphere: Archaean cratons have roots of 200 km or more, consistent with their low heat flow, high seismic velocities and xenolith thermobarometry.
Oceanic Lithosphere: the Cooling Half-Space
Set-up and solution
Mantle at temperature \( T_m \) rises at the ridge and its top is suddenly held at the sea-floor temperature \( T_0 \). Moving away from the ridge at speed \( u \), a column of age \( t = x/u \) cools mainly vertically (horizontal conduction is negligible because \( Pe = uL/\kappa \gg 1 \)). This is the one-dimensional problem of a half-space cooled from the top:
$$\boxed{\frac{T - T_0}{T_m - T_0} = \operatorname{erf}\left(\frac{z}{2\sqrt{\kappa t}}\right)}$$Three predictions
1. Heat flow (differentiate at \( z = 0 \)):
$$\boxed{q(t) = \frac{k(T_m - T_0)}{\sqrt{\pi\kappa t}}}$$With \( k = 3.3 \) W m\(^{-1}\) K\(^{-1}\), \( T_m - T_0 = 1300 \) K and \( \kappa = 10^{-6} \) m\(^2\) s\(^{-1}\), this gives about \( 430/\sqrt{t} \) mW m\(^{-2}\) with \( t \) in Myr: about 430 at 1 Myr, 86 at 25 Myr, 54 at 64 Myr. Published fits to data have the same form with constants of a similar size; the exact value depends on the parameters chosen.
2. Thickness. Taking the base of the thermal lithosphere where \( T - T_0 = 0.9(T_m - T_0) \), i.e. \( \operatorname{erf}(\eta) = 0.9 \), \( \eta \approx 1.16 \):
$$\boxed{z_L \approx 2.32\sqrt{\kappa t}}$$About 65 km at 25 Myr and 105 km at 64 Myr. Thickness grows as \( \sqrt{t} \): quadrupling the age only doubles it.
3. Subsidence. Cooling makes the column denser (thermal expansivity \( \alpha \)). Isostatic balance, with water of density \( \rho_w \) filling the space above, gives the depth below the ridge crest:
$$\boxed{w(t) = \frac{2\rho_m\alpha(T_m - T_0)}{\rho_m - \rho_w}\sqrt{\frac{\kappa t}{\pi}}}$$With \( \rho_m = 3300 \), \( \rho_w = 1000 \) kg m\(^{-3}\), \( \alpha = 3\times 10^{-5} \) K\(^{-1}\): \( w \approx 355\sqrt{t} \) m (\( t \) in Myr). This is the 350 in the empirical \( d \approx 2500 + 350\sqrt{t} \) m.
Where the half-space fails
graph TD
A[Half-space model] --> B[Young floor: measured heat flow below prediction]
A --> C[Old floor beyond about 70 Myr: depth and heat flow flatten]
B --> D[Hydrothermal circulation removes heat by advection]
C --> E[Plate model: fixed temperature at a depth of order 100 km]
C --> F[Physical causes: small-scale convection under old plates, heat from plumes]- Young sea floor: measured conductive heat flow is lower and more scattered than predicted, out to crustal ages of tens of Myr. Sea water circulating through permeable crust carries the missing heat out by advection; once enough sediment seals the crust, measurements approach the prediction.
- Old sea floor: beyond roughly 70 Myr, depth and heat flow change more slowly than \( \sqrt{t} \). The plate model imposes a fixed temperature at a depth \( a \) of order 100 km, so the lithosphere approaches a steady thickness; physically, extra heat must reach the base of old plates, for example through small-scale convection that destabilises the thickening boundary layer.
| Half-space model | Plate model | |
|---|---|---|
| Lower boundary | none (infinite depth) | fixed \( T_m \) at depth \( a \) |
| Heat flow | \( \propto t^{-1/2} \) forever | \( \propto t^{-1/2} \) when young; tends to a constant |
| Depth | \( \propto \sqrt{t} \) forever | \( \propto \sqrt{t} \) when young; tends to a constant |
| Fits | young sea floor | young and old sea floor |
Ridge push from the cooling model
The elevated, cooling lithosphere near ridges exerts a horizontal ridge-push force per unit length of ridge. To within a few per cent,
$$F_{rp} \approx g\rho_m\alpha(T_m - T_0)\kappa t$$For 100 Myr old lithosphere (\( t = 3.16\times 10^{15} \) s): \( 10\times 3300\times 3\times 10^{-5}\times 1300\times 10^{-6}\times 3.16\times 10^{15} \approx 4\times 10^{12} \) N m\(^{-1}\). It grows linearly with age: a body force distributed over the whole cooling plate, not a push at the axis. (Driving forces are compared on Mantle Convection and Plumes.)
Time-Dependent Problems You Can Do in a Line
Periodic surface temperature: skin depth
A surface temperature varying as \( \Delta T_0\cos\omega t \) penetrates as a damped wave:
$$T(z,t) = T_0 + \Delta T_0\,e^{-z/\delta}\cos\left(\omega t - \frac{z}{\delta}\right) \qquad \boxed{\delta = \sqrt{\frac{2\kappa}{\omega}} = \sqrt{\frac{\kappa P}{\pi}}}$$| Period \( P \) | Skin depth (\( \kappa = 10^{-6} \)) |
|---|---|
| 1 day | about 0.17 m |
| 1 year | about 3.2 m |
| 100 kyr (glacial cycle) | about 1 km |
At depth \( \delta \) the amplitude is \( e^{-1} \approx 37\% \) of the surface value and the phase lags by 1 radian; at \( \pi\delta \) the variation is exactly out of phase with the surface (about 10 m for the annual cycle: the ground at that depth is warmest in winter). The 100 kyr row explains why paleoclimate corrections matter in boreholes a kilometre deep.
Cooling of a dyke or sill
The time for a sheet intrusion of half-thickness \( b \) to lose most of its heat is of order \( b^2/\kappa \). A 10 m dyke (\( b = 5 \) m) cools in about \( 25/10^{-6} = 2.5\times 10^{7} \) s, under a year. A 1 km sill takes of order \( 2.5\times 10^{11} \) s, about 8 kyr. Doubling the thickness quadruples the time.
Kelvin’s calculation
For a half-space cooling from initial temperature excess \( \Delta T \), the surface gradient is \( G = \Delta T/\sqrt{\pi\kappa t} \), so
$$t = \frac{\Delta T^2}{\pi\kappa G^2}$$With \( \Delta T = 2000 \) K, \( G = 25 \) K km\(^{-1}\) and \( \kappa = 10^{-6} \): \( t \approx 2.0\times 10^{15} \) s \( \approx 65 \) Myr. Kelvin’s own inputs differed, but the order of magnitude is the same, and so is the lesson: conduction from a hot start explains a gradient only for a young Earth. Radioactivity (supplying heat) and convection (keeping the interior hot while only a thin lid conducts) remove the paradox.
Thermal Structure of Subduction Zones
A slab is cold and thermally thick, and it descends faster than it can warm by conduction (high \( Pe \)), so isotherms are dragged down. Its interior warms on a time scale of order (thickness)\(^2/\kappa \): for a 100 km thick slab that is hundreds of Myr, longer than the time it takes to reach the transition zone. The thermal parameter \( \Phi = t\,v\sin\delta \) (see Plate Boundaries, Kinematics and Seismicity) measures how far down the cold core survives. Above the slab, corner flow in the mantle wedge brings hot mantle towards the arc, and heat flow is low in the forearc and high in the arc and back-arc.
Memory Tricks
Don’t Confuse
| This | vs | That |
|---|---|---|
| Thermal conductivity \( k \) (W m\(^{-1}\) K\(^{-1}\)) | Thermal diffusivity \( \kappa = k/\rho c_p \) (m\(^2\) s\(^{-1}\)) | |
| Surface heat flow \( q_0 \) | Reduced heat flow \( q_r \) (from below the radiogenic layer) | |
| Half-space model (no bottom; root-t forever) | Plate model (fixed temperature at depth \( a \); flattens) | |
| Geothermal gradient (K km\(^{-1}\); changes across layers) | Heat flow (mW m\(^{-2}\); continuous in steady state) | |
| Thermal lithosphere (conductive lid, base near 1,300 °C) | Elastic thickness \( T_e \) (from flexure; thinner) | |
| Hydrothermal heat deficit (young sea floor, too low) | Flattening (old sea floor, depth and heat flow level off) |
Common Mistakes
Practice
All questions below are original practice questions written for this page. They are not from any CSIR NET paper.
Level 1: Foundation
- Write the thermal diffusivity in terms of conductivity, density and specific heat.
- How does the time for heat to diffuse a distance \( L \) scale with \( L \)?
- What is meant by reduced heat flow?
Answers
- \( \kappa = k/(\rho c_p) \), about \( 10^{-6} \) m\(^2\) s\(^{-1}\) for rocks.
- As \( L^2 \): \( \tau \sim L^2/\kappa \).
- The intercept \( q_r \) of the line \( q_0 = q_r + DA_0 \): heat flow from below the layer of enriched crustal heat production.
Level 2: Part B Style
Practice Q1. In a granite borehole the temperature gradient is 20 K km\(^{-1}\) and the conductivity is 3.0 W m\(^{-1}\) K\(^{-1}\). The heat flow is:
(a) 6.7 mW m\(^{-2}\) (b) 15 mW m\(^{-2}\) (c) 60 mW m\(^{-2}\) (d) 150 mW m\(^{-2}\)
Answer: (c) 60 mW/m^2
\( q = k\,dT/dz = 3.0\times 0.020 = 0.060 \) W m\(^{-2}\) = 60 mW m\(^{-2}\). (a) divides the gradient by the conductivity; (d) divides the conductivity by the gradient. Always multiply, and convert K km\(^{-1}\) to K m\(^{-1}\) first.
Practice Q2. Heat flow on 4 Myr old sea floor is 200 mW m\(^{-2}\). By the half-space model, heat flow on 64 Myr old sea floor is:
(a) 12.5 mW m\(^{-2}\) (b) 50 mW m\(^{-2}\) (c) 100 mW m\(^{-2}\) (d) 800 mW m\(^{-2}\)
Answer: (b) 50 mW/m^2
\( q\propto t^{-1/2} \): \( 200\times\sqrt{4/64} = 200/4 = 50 \) mW m\(^{-2}\). (a) uses \( 1/t \).
Practice Q3. With \( \kappa = 10^{-6} \) m\(^2\) s\(^{-1}\), the depth at which the annual surface temperature wave is reduced to \( e^{-1} \) of its surface amplitude is about:
(a) 0.17 m (b) 1 m (c) 3.2 m (d) 32 m
Answer: (c) 3.2 m
\( \delta = \sqrt{\kappa P/\pi} = \sqrt{10^{-6}\times 3.156\times 10^{7}/\pi} = \sqrt{10.0} \approx 3.2 \) m. (a) is the daily skin depth.
Practice Q4. Using \( d = 2500 + 350\sqrt{t} \) m, the depth of 49 Myr old sea floor is about:
(a) 3,725 m (b) 4,950 m (c) 5,950 m (d) 19,650 m
Answer: (b) 4,950 m
\( 2500 + 350\times 7 = 4{,}950 \) m. (d) uses \( t \) instead of \( \sqrt{t} \).
Level 3: Part C Style
Practice Q5. A continental province has \( q_r = 30 \) mW m\(^{-2}\), surface heat production \( A_0 = 2.5\ \mu \)W m\(^{-3}\) decreasing as \( e^{-z/D} \) with \( D = 10 \) km, \( k = 2.5 \) W m\(^{-1}\) K\(^{-1}\) and \( T_0 = 10 \) °C. Which statements are correct?
- The surface heat flow is 55 mW m\(^{-2}\).
- The temperature at 40 km depth is about 590 °C.
- The heat flow at 40 km depth is about 30.5 mW m\(^{-2}\).
- Extrapolating the surface gradient linearly would give about 480 °C at 40 km.
(a) 1 and 2 only (b) 1, 2 and 3 only (c) 2, 3 and 4 only (d) 1, 2, 3 and 4
Answer: (b) 1, 2 and 3 only
- 1: \( q_0 = q_r + A_0D = 30 + 25 = 55 \) mW m\(^{-2}\). True.
- 2: \( T = 10 + 480 + 100(1 - e^{-4}) = 10 + 480 + 98 = 588 \) °C. True.
- 3: \( q(40) = 30 + 25e^{-4} = 30 + 0.46 = 30.5 \) mW m\(^{-2}\). True.
- 4: False. The surface gradient is \( 55/2.5 = 22 \) K km\(^{-1}\); linear extrapolation gives \( 10 + 22\times 40 = 890 \) °C. (480 °C is only the \( q_r z/k \) term.)
Practice Q6. Take \( k = 3.3 \) W m\(^{-1}\) K\(^{-1}\), \( T_m - T_0 = 1300 \) K and \( \kappa = 10^{-6} \) m\(^2\) s\(^{-1}\) in the half-space model. Which statements are correct?
- Heat flow at 25 Myr is about 86 mW m\(^{-2}\).
- The thermal lithosphere (90% isotherm) at 25 Myr is about 65 km thick.
- Doubling the age doubles the thickness of the thermal lithosphere.
- Heat flow measured with probes on 5 Myr old crust is often lower than this model predicts.
(a) 1, 2 and 4 only (b) 1 and 3 only (c) 2, 3 and 4 only (d) 1, 2, 3 and 4
Answer: (a) 1, 2 and 4 only
\( t = 25\times 3.156\times 10^{13} = 7.89\times 10^{14} \) s; \( \kappa t = 7.89\times 10^{8} \) m\(^2\); \( \sqrt{\kappa t} = 28.1 \) km.
- 1: \( q = \frac{3.3\times 1300}{\sqrt{\pi\times 7.89\times 10^{8}}} = \frac{4290}{4.98\times 10^{4}} = 0.086 \) W m\(^{-2}\). True.
- 2: \( 2.32\times 28.1 \approx 65 \) km. True.
- 3: False. Thickness scales as \( \sqrt{t} \): doubling the age multiplies it by \( \sqrt{2} \approx 1.41 \).
- 4: True. Hydrothermal circulation in young crust removes heat by advection.
Practice Q7. Mantle beneath a ridge rises at 3 cm yr\(^{-1}\). Over a length scale of 100 km, with \( \kappa = 10^{-6} \) m\(^2\) s\(^{-1}\), the Peclet number and its meaning are:
(a) about 0.95; conduction dominates (b) about 9.5; conduction and advection comparable (c) about 95; advection dominates (d) about 950; radiation dominates
Answer: (c) about 95; advection dominates
\( v = 0.03/3.156\times 10^{7} = 9.5\times 10^{-10} \) m s\(^{-1}\). \( Pe = \frac{9.5\times 10^{-10}\times 10^{5}}{10^{-6}} \approx 95 \).
With \( Pe \gg 1 \), heat is carried upward with the rising mantle faster than it can conduct away, which is why the mantle reaches shallow depth nearly at its adiabatic temperature and melts by decompression. Peclet number compares advection with conduction; it says nothing about radiation, so (d) is wrong on two counts.
Quick Revision Table
| Idea | Fact or formula |
|---|---|
| Fourier | \( q = -k\nabla T \); \( k \) about 2–3 (crust) and 3–4 (mantle) W m\(^{-1}\) K\(^{-1}\) |
| Heat equation | \( \rho c_p(\partial_t T + \mathbf{v}\cdot\nabla T) = k\nabla^2T + A \) |
| Diffusivity | \( \kappa = k/\rho c_p \approx 10^{-6} \) m\(^2\) s\(^{-1}\) |
| Diffusion | \( \tau \sim L^2/\kappa \); \( L\sim\sqrt{\kappa t} \) |
| Peclet | \( Pe = vL/\kappa \) |
| Bullard plot | \( T = T_0 + qR \), \( R = \sum\Delta z_i/k_i \) |
| Global | continents about 65, oceans about 100, mean about 87 mW m\(^{-2}\); 44–47 TW |
| Uniform \( A \) | \( T = T_0 + q_0z/k - Az^2/2k \) |
| Exponential \( A \) | \( q_0 = q_r + A_0D \); \( T = T_0 + q_rz/k + (A_0D^2/k)(1 - e^{-z/D}) \) |
| Half-space | \( T \propto \operatorname{erf}(z/2\sqrt{\kappa t}) \); \( q = k\Delta T/\sqrt{\pi\kappa t} \approx 430/\sqrt{t} \) mW m\(^{-2}\) |
| Thickness | \( z_L \approx 2.32\sqrt{\kappa t} \) |
| Subsidence | \( w = \frac{2\rho_m\alpha\Delta T}{\rho_m - \rho_w}\sqrt{\kappa t/\pi} \approx 355\sqrt{t} \) m |
| Misfits | young: hydrothermal deficit; old (beyond about 70 Myr): flattening, plate model |
| Ridge push | \( \approx g\rho_m\alpha\Delta T\kappa t \approx 4\times 10^{12} \) N m\(^{-1}\) at 100 Myr |
| Skin depth | \( \delta = \sqrt{\kappa P/\pi} \): day 0.17 m, year 3.2 m, 100 kyr about 1 km |
| Kelvin | \( t = \Delta T^2/\pi\kappa G^2 \); missed radioactivity and convection |
Key Takeaways
What’s Next
- Mantle Convection and Plumes: the heat transport that conduction cannot provide.
- Marine Magnetic Anomalies, Sea-Floor Spreading and Ridges: revisit the age–depth data this model explains.
- Extensional, Compressional and Strike-Slip Terrains; Mantle Plumes: thermal subsidence of rift basins uses the same cooling physics.
- Plate Tectonics and Geodynamics: Quick Revision.