betaFeedback

WorksPhysics and Astronomy

Convection in the active layer speeds up permafrost thaw in coarse grained soils

Abstract

Permafrost thaw is a major concern raised by the ongoing climate change. An understudied phenomenon possibly affecting the pace of permafrost thaw is the onset of convective motions within the active layer caused by the density anomaly of water. Here, we explore the effects of groundwater convection on permafrost thawing using a model that accounts for ice - water phase transitions, coupled with the dynamics of the temperature field transported by the Darcy's flow across a porous matrix. Numerical simulations of this model show that ice thawing in the presence of convection is much faster than in the diffusive case and deepens at a constant velocity proportional to the soil permeability. A scaling argument is able to predict correctly the asymptotic velocity. Since in the convective regime the heat transport is mediated by the coherent motion of thermal plumes across the thawed layer, we find that the depth of the thawing interface becomes highly heterogeneous.

Introduction. The thawing of permafrost - ground that has been frozen for at least two consecutive years - poses important societal and climatic challenges owing to its control on soil stability, health hazards and mobilization of compounds trapped in cold soils [1, 2, 3, 4]. Permafrost evolution is deeply tied to the thermal and hydrological dynamics at ground surface and within the so-called “active layer”, that is the surface layer subject to annual thawing and freezing usually present above permafrost. The dynamics of both the permafrost and the active layer are mostly estimated on the basis of models [5, 6], while direct measurements are rather sparse and spotty [2, 6].

Climate models mostly project permafrost evolution accounting for the vertical conductive heat transfer across the active layer [7, 8, 9, 10, 11]. Nonetheless, a crucial role in determining the speed of permafrost thaw can be played by fluid convection in the active layer soil matrix at large permeability, sustained by the density anomaly of water. During the thawing process, the temperature within the active layer decreases with depth from the surface value to the freezing temperature met at the contact with the permafrost. Across this thermal profile, fluid density varies nonlinearly (Figure 1a): below T≃4 ∘CT \simeq4\,^{\circ}\mathrm{C} water has a negative thermal expansion coefficient, i.e., its density increases with temperature, whereas above this value the temperature-density relationship is inverted. Therefore, in the lower part of the active layer where 0 ∘C<T<4 ∘C0\,^{\circ}\mathrm{C} < T < 4\,^{\circ}\mathrm{C}, soil water is unstably stratified and convection can occur.

Schematic dynamics of the active layer and vertical sections of the temperature fields

Figure 1. (a) Schematic dynamics of the active layer considered in this work. In the top layer, where T>4 ∘CT > 4\,^{\circ}\mathrm{C}, the fluid stratification is stable, resulting in a diffusive heat transfer. Below this layer the unstable stratification can produce a convective layer characterized by the presence of thermal plumes. The permafrost layer where T<0 ∘CT < 0\,^{\circ}\mathrm{C} is represented in grey. Conceptual representations of temperature and density water profiles are sketched by the red and blue line, respectively. (b) Vertical sections of the temperature fields for run C (upper panel) and D (lower panel) at time t=150t = 150 days, starting from the same initial conditions. Colors indicate the temperature field. The grey area represents the frozen region where ϕ=1\phi= 1.

In this work we study the effect of active layer fluid motion on permafrost thaw. We consider an homogeneous-isotropic porous medium saturated with water in the presence of gravity. For the application to permafrost thaw, we assume that the lower part of the domain is frozen (at temperature T0<TM=0 ∘CT_0 < T_M = 0\,^{\circ}\mathrm{C}) while an upper thin porous layer is already melted and in contact to the surface at a temperature T1>TMT_1 > T_M (see the scheme in Figure 1a). Three-dimensional direct numerical simulations of this system show that, depending on the permeability of the porous medium, convective motions can develop in the active layer. This accelerates substantially the thawing process with respect to the diffusive behavior and produces a complex water-ice interface which deepens ballistically in time.

Mathematical model and numerical simulations. The physics of thawing in a porous medium is described by a phase field method, a convenient numerical tool for simulating the dynamics of multiphase fluids in the presence of phase transitions [12, 13, 14]. The physical state of water is represented by a continuous phase field ϕ(x,t)\phi(x,t) which takes the value ϕ=0\phi= 0 in the liquid phase and ϕ=1\phi= 1 in the solid (ice) phase. The equation for the phase field is a Allen-Cahn type equation [15] coupled with an advection-diffusion equation for the temperature field T(x,t)T(x,t) and the Darcy’s equation, which describes the velocity field u(x,t)\mathbf{u}(x,t) in the porous medium

u=kμϕ(−∇p+(1−ϕ)2ρg)(1)\mathbf{u} = \frac{k}{\mu\phi}\left(-\nabla p + (1-\phi)^2 \rho\mathbf{g}\right) \tag*{(1)}
∂T∂t+u⋅∇T=κ∇2T+ΔT St ∂ϕ∂t(2)\frac{\partial T}{\partial t} + \mathbf{u}\cdot\nabla T = \kappa\nabla^2 T + \Delta T\,St\,\frac{\partial\phi}{\partial t} \tag*{(2)}
∂ϕ∂t=6κ′5St[∇2ϕ−1δ2ϕ(1−ϕ)(1−2ϕ+T−TMΔT)](3)\frac{\partial\phi}{\partial t} = \frac{6\kappa'}{5St}\left[\nabla^2\phi- \frac{1}{\delta^2}\phi(1-\phi)\left(1-2\phi+\frac{T-T_M}{\Delta T}\right)\right] \tag*{(3)}

where g=(0,0,−g)\mathbf{g}=(0,0,-g) represents gravity, kk is the permeability, μ\mu is the fluid viscosity, ϕ\phi the porosity, pp is the total pressure and κ\kappa represents an effective thermal diffusivity (linear combination of the solid and the liquid contributions [16]). St=L/(cpΔT)St=L/(c_p\Delta T) is the dimensionless Stefan number defined in terms of the latent heat for unit mass LL, the specific heat capacity of water cpc_p and the temperature jump ΔT=T1−T0\Delta T=T_1-T_0, while δ\delta and κ′\kappa' represent the interface thickness and mobility. Finally, density and temperature of water are related by the empirical non-monotonic model, valid around the temperature Tmax⁡=3.98 ∘CT_{\max}=3.98\,^{\circ}\mathrm{C} of maximal density, ρ=ρ0(1−α∗∣T−Tmax⁡∣q)\rho=\rho_0(1-\alpha^*|T-T_{\max}|^q) where α∗\alpha^* is a generalized thermal expansion coefficient, q=1.895q=1.895 and ρ0=999.97 kg/m3\rho_0=999.97\,\mathrm{kg/m^3} [17]. We remark that the model (1-3) assumes local thermal equilibrium by which a single temperature field describes the solid and the fluid phases in the porous medium [18]. More complex, non-equilibrium, models which take into account heat exchanges within the solid matrix have been proposed [16, 19] and could be used for future studies in the present setup.

Because of the density anomaly, the thawed water has an unstable density stratification in the layer where TM<T<Tmax⁡T_M < T < T_{\max}. The dimensionless number which controls the dynamics of this fluid layer is the Rayleigh-Darcy number

Ra=gΔρkHϕμκ(4)Ra = \frac{g\Delta\rho k H}{\phi\mu\kappa} \tag*{(4)}

where HH and Δρ\Delta\rho are respectively the thickness of and the density difference in the convective layer. For values of RaRa below the critical value Rac=4π2Ra_c=4\pi^2, the heat transfer across the active layer is diffusive and the fluid is at rest [20]. In contrast, when Ra>RacRa>Ra_c the unstable density stratification causes a convective motion which displaces the fluid and the temperature field.

We performed direct numerical simulations of equations (1-3) by a fully parallel (MPI) pseudo-spectral code up to resolution 2048×2048×5122048 \times2048 \times512. Boundary conditions on the top and bottom layers are imposed by penalization terms, while we use periodic condition in the horizontal directions. In all the simulations the fluid is initially at rest and the vertical temperature profile within the active layer of depth H0H_0 is a linear ramp, connecting the temperature T0=−1 ∘CT_0 = -1\,^{\circ}\mathrm{C} in the permafrost bulk to T1=Tmax⁡T_1 = T_{\max} at the ground surface. In order to destabilize the flow, a small random perturbation is added to the initial fluid temperature profile. Units for space and time are meters and days respectively and for the physical parameters we set κ=0.012 m2/day\kappa= 0.012\,\mathrm{m^2/day}, μ=86.0 kg/(m day)\mu= 86.0\,\mathrm{kg/(m\,day)}, L=3.3×105 J/kgL = 3.3 \times10^5\,\mathrm{J/kg} (for simplicity we use the same values for liquid and ice phase). Three different runs have been performed, all with porosity ϕ=0.5\phi= 0.5, with permeabilities k=4.4×10−9 m2k = 4.4 \times10^{-9}\,\mathrm{m^2} (A), k=5.9×10−9 m2k = 5.9 \times10^{-9}\,\mathrm{m^2} (B) and k=8.8×10−9 m2k = 8.8 \times10^{-9}\,\mathrm{m^2} (C), typical of water in gravel. As a benchmark, we also consider an additional run with much smaller permeability k=6×10−11 m2k = 6 \times10^{-11}\,\mathrm{m^2} which remains in the diffusive regime. The numerical method has been tested in the case of a purely diffusive process, i.e. with u=0u = 0 in (2), corresponding to the standard Stefan problem, for which analytical predictions are available.

Results. Assuming an initial small thickness H0H_0, such that Ra<RacRa < Ra_c, the early stage of the thawing process is characterized by diffusive heat flux from the surface. Consequently, the width of the liquid layer is expected to increase as H(t)∝t1/2H(t) \propto t^{1/2} [21]. This growth is accompanied by the increase of Rayleigh-Darcy number (4), which eventually becomes larger than RacRa_c causing the onset of convection. It is well known that, in general, convection is much more efficient to transfer heat than diffusion, and porous convection is not an exception [22, 23]. Therefore, in the convective regime H(t)H(t) is expected to grow much faster than in the diffusive case. This results in a further increase of RaRa and of the convective motion, which accelerates the thawing process.

A first insight into the difference between the conductive and convective dynamics can be obtained from the active layer temperature field. Figure 1b shows two vertical sections on the (x,z)(x,z) plane of the water temperature field (colors) taken at t=150t = 150 days for the conductive case (run D, lower panel), and one convective case (run C, upper panel). The frozen part is represented in grey.

When the permeability is small (run D) the heat transfer in the active layer is purely diffusive, resulting in a uniform temperature gradient in the liquid layer and a flat fluid-ice interface. Conversely, the convective cases (such as run C in Figure 1b) are characterized by the presence of thermal plumes that are responsible for the efficient exchange of heat across the active layer and speed up the thawing process. Thermal plumes are generated close to the layer at temperature Tmax⁡T_{\max} and they move downwards causing an inhomogeneous heating and thawing of the ice. As a consequence, the fluid-ice interface is not flat, and ice and water coexist at a given depth zz, in clear contrast with the diffusive case (compare the top and bottom simulation in Figure 1b).

Thermal plumes that have reached different depths in the thawing process create complex patterns at the thawing surface h(x,y,t)h(x,y,t), defined in this study as the depth where the local phase field ϕ(x,t)=1/2\phi(x,t)=1/2. One example of this surface is shown in Figure 2a for run C at time t=150t=150 days. The average depth reached by convection is in this case H(t)≃1.6H(t)\simeq1.6 m, but we observe local fluctuations ranging from 1 m up to 2.2 m below downwelling plumes. Remarkably, similar patterns have been recently reported as the results of dissolution processes observed both in geological sites and in laboratory experiments in bulk flows [24].

Figure 2. (a) Thawing surface depth h(x,y)h(x,y), defined from the phase field as the points where ϕ(x,y,h)=1/2\phi(x,y,h)=1/2, for run C at time t=150t=150 days. The average position of the surface is H=1.6H=1.6 m. (b) Vertical temperature profile T‾(z)\overline{T}(z) (red line) and phase field profile ϕ‾(z)\overline{\phi}(z) (blue line) at time t=150t=150 days. The black dotted line represents the average position of the interface defined as ϕ‾(−H)=1/2\overline{\phi}(-H)=1/2. Results from run C.

In order to compare quantitatively the conductive and convective processes, we compute the vertical profiles of temperature field T‾(z)\overline{T}(z) and of phase field ϕ‾(z)\overline{\phi}(z), where the overbar represents the average over horizontal planes. Figure 2b shows one example of these profiles for the run C. In the temperature profile we recognize the presence of a boundary layer close to the upper surface at temperature T1T_1, followed by a non-monotonic temperature pattern, due to the presence of thermal plumes which produce a local temperature maximum. Below this maximum, the temperature decreases sharply, while the phase field profile ϕ‾(z)\overline{\phi}(z) increases. The region in which 0<ϕ‾<10<\overline{\phi}<1 corresponds to the thawing surface with the coexistence, at a given depth, of zones which are liquid (ϕ=0\phi=0) with others still frozen (ϕ=1\phi=1). The depth at which ϕ‾(z)=1/2\overline{\phi}(z)=1/2 is used to define the average vertical position H(t)H(t) of the water-ice interface (dotted line in Figure 2).

The displacement of the interface H(t)H(t) as a function of time is shown in Figure 3a. The convective case with lowest permeability (run A, green line) is initially below the critical Rayleigh-Darcy number and therefore in a first phase it follows the diffusive Stefan solution (grey line). At about t=15t=15 days convection sets in and the displacement of the interface accelerates with respect to the diffusive case. For larger values of the permeability (run B, red line and run C, blue line) the convective motion develops since the beginning. The effect of convection on the thawing process is evident: after 60 days (vertical dashed line) the interface has displaced, in the case of the highest permeability, of about 0.7 m from the starting position, that is more than three times what is predicted in the absence of convection (grey line).

Figure 3. (a) Average displacement of the water-ice interface as a function of time for the three runs in the convective regime A (green line), B (red line) and C (blue line) and for run D in the diffusive regime (grey line). H0=0.33 mH_0 = 0.33\,m is the initial position of the interface at t=0t = 0. The grey area represents the width of the thawing layer (0<ϕˉ(z)<10 < \bar{\phi}(z) < 1) for run C. The vertical dashed line is a guide for the t=60t = 60 days. Inset: displacement rescaled with the characteristic velocity U0U_0 in the last part of the evolution. (b) Nusselt number NuNu as a function of the Rayleigh-Darcy number RaRa for run A (green squares), B (red circles), C (blue triangles) and D (grey inverted triangles). The black line represent the linear fit Nu=0.016Ra+const.Nu = 0.016Ra + const. Inset: The same data in semi-logarithmic scales for the lower range of RaRa.

At long times, the convection-driven motion of the interface attains a linear law H∝tH\propto t, instead of the diffusive growth predicted by the Stefan model. This can be understood from the mathematical model governing the process. From the Darcy equation (1) one can define a characteristic velocity U0=Δρkg/(μϕ)U_0=\Delta\rho kg/(\mu\phi) which represents the typical speed of thermal plumes. By rescaling the different trajectories of H(t)H(t) in Figure 3a with U0U_0 one obtains the collapse of the linear behavior H(t)=αU0tH(t)=\alpha U_0t with α≃0.0055\alpha\simeq0.0055, as shown in the inset of Figure 3a. Hence, by changing the permeability value (or the other parameters in U0U_0) one obtains different velocities of the thawing interface in the convective regime. We remark that the dimensionless coefficient α\alpha is here much smaller than the one observed in Rayleigh-Taylor convection in porous media [23] since a significant fraction of the energy is used to thaw the ice phase. Together with HH, we find that also the extension of the layer where 0<ϕˉ(z)<10 < \bar{\phi}(z) < 1 grows with time, as shown by the grey area in Figure 3a for run C. This region corresponds to the coexistence of zones that are already liquid (ϕ=0\phi= 0) with others still frozen (ϕ=1\phi= 1) at a given depth, i.e. the vertical extension of the thawing surface shown in Figure 2a.

The enhancement of the heat transport due to the convection with respect to the thermal diffusion is quantified by the Nusselt number NuNu, defined as the ratio of the total (convective plus diffusive) heat flux to the diffusive one:

Nu=⟨uzT⟩Hκ(Tmax⁡−TM)+1(5)Nu = \frac{\langle u_z T\rangle H}{\kappa(T_{\max} - T_M)} + 1 \tag*{(5)}

where ⟨⋯ ⟩\langle\cdots\rangle indicates the average over the convective region of depth HH and uzu_z is the vertical velocity. Figure 3b shows the evolution of the Nusselt number (5) as a function of the Rayleigh-Darcy number (4) for the four runs. We recall that since the thickness of the active layer HH grows with time, so does the Rayleigh-Darcy number and therefore each simulation spans a range of RaRa. Given the small density differences of water between TMT_M and Tmax⁡T_{\max}, the Rayleigh-Darcy number remains moderate for all the permeability values investigated here. As shown in the inset of Figure 3b, the Nusselt number in the run D at the lowest permeability is always Nu=1Nu = 1, because it remains in the diffusive regime (Ra<RacRa < Ra_c). For the other three simulations we see that, for sufficiently large values of the Rayleigh-Darcy number, NuNu increases with RaRa following a linear law. This can be understood by a scaling argument based on the definitions of the two dimensionless numbers. Indeed, from (4) we see that, at fixed values of the fluid parameters, RaRa is proportional to the thickness HH of the fluid layer. From Eq. (5), assuming that the correlation between the vertical velocity and the temperature field can be simply expressed as ⟨uzT⟩=βU0ΔT\langle u_z T\rangle= \beta U_0\Delta T (where β\beta is a dimensionless constant) we have the prediction of a linear scaling Nu=βRaNu = \beta Ra, as indeed observed in Figure 3b, with β≃0.016\beta\simeq0.016. We remark that, in spite of the different thawing speed for the different permeability values, Figure 3b indicates a universal RaRa–NuNu relation, almost independent of the permeability kk. This suggests that the correlation between the vertical velocity and the temperature, which enters in Eq. (5), is not affected by the permeability of the porous medium.

To conclude, we have studied how convection accelerates the pace of permafrost thawing compared to the conduction process in gravel-dominated soils. With the onset of convection, the ice-water interface deepens linearly in time, in clear contrast with the diffusive law expected from the conductive case. Another consequence of convection is the formation of a complex thawing interface, charaterized by spatial patterns, that can be used as a diagnostic for the presence of convection in the active layer. Remarkably, similar patterns have been observe in dissolution processes in bulk flows [24]. Further investigations are needed to clarify whether the observed thawing surface can be explained within the same theoretical framework.

Convective flows have been already mentioned in Arctic field studies [25, 26, 27], thus confirming the possibility of this process in natural conditions. According to the present study, the onset of convection is mostly controlled by the soil permeability. Indeed in saturated soils, large permeability values are needed to reach the condition for the fluid density stratification to become unstable and undergo convective motions. Using typical water parameters, such a condition is met for instance in rocky and sandy-gravel soils [28, 29, 30]. Fire events in arctic and subarctic ecosystems have also been suggested to increase the connectivity of soil pores, i.e. soil permeability, thus allowing for advective fluid flows that enhance post-fire permafrost degradation [30, 31]. Instead, if the active layer is characterized by a soil matrix with low permeability the onset of convection is usually prevented.

Despite using a simplified representation of the active layer system, where for instance soil characteristics are kept constant in time and across soil depths, we showed the importance of considering convective processes in cold soils as their effects can already manifest on short time scales of the order of one month. Hence, the effects of convection can already be felt within one summer period, when the ground surface temperature can be much higher than 4 ∘C4\,^{\circ}\mathrm{C}. These results suggest that climate models accounting for only heat conduction may underestimate the permafrost thawing depth in most permeable soils, where soil water convection in the active layer can occur. In these cases, the framework proposed here may serve to obtain useful parameterisations to be implemented in climate models.

Acknowledgements. We acknowledge HPC CINECA for computing resources (INFN-CINECA Grant INFN23-FieldTurb). M.M., and A.P. acknowledge the Italian National Biodiversity Future Center (NBFC) National Recovery and Resilience Plan (NRRP; mission 4, component 2, investment 1.4 of the Ministry of University and Research, funded by the European Union - NextGenerationEU; project code CN00000033).

References

References

  1. [1]Jan Hjort, Dmitry Streletskiy, Guy Doré, Qingbai Wu, Kevin Bjella, and Miska Luoto, “Impacts of permafrost degradation on infrastructure,” Nat. Rev. Earth Environ. 3, 24–38 (2022).DOI
  2. [2]Olli Karjalainen, Juha Aalto, Miska Luoto, Sebastian Westermann, Vladimir E Romanovsky, Frederick E Nelson, Bernd Etzelmüller, and Jan Hjort, “Circumpolar permafrost maps and geohazard indices for near-future infrastructure risk assessments,” Sci. Data 6, 1–16 (2019).DOI
  3. [3]Jan Hjort, Olli Karjalainen, Juha Aalto, Sebastian Westermann, Vladimir E Romanovsky, Frederick E Nelson, Bernd Etzelmüller, and Miska Luoto, “Degrading permafrost puts Arctic infrastructure at risk by mid-century,” Nat. Commun. 9, 5147 (2018).DOI
  4. [4]Nicola Colombo, Franco Salerno, Stephan Gruber, Michele Freppaz, Mark Williams, Simona Fratianni, and Marco Giardino, “Impacts of permafrost degradation on inorganic chemistry of surface fresh water,” Glob. Plan. Change 162, 69–83 (2018).DOI
  5. [5]SE Chadburn, EJ Burke, PM Cox, P Friedlingstein, Gustaf Hugelius, and S Westermann, “An observation-based constraint on permafrost loss as a function of global warming,” Nat. Clim. Chang. 7, 340–344 (2017).DOI
  6. [6]Jaroslav Obu, Sebastian Westermann, Annett Bartsch, Nikolai Berdnikov, Hanne H Christiansen, Avirmed Dashtseren, Reynald Delaloye, Bo Elberling, Bernd Etzelmüller, Alexander Kholodov, et al., “Northern Hemisphere permafrost map based on TTOP modelling for 2000–2016 at 1 km² scale,” Earth-Science Rev. 193, 299–316 (2019).DOI
  7. [7]R Dankers, EJ Burke, and J Price, “Simulation of permafrost and seasonal thaw depth in the JULES land surface scheme,” The Cryosphere 5, 773–790 (2011).
  8. [8]R Wania, I Ross, and IC Prentice, “Integrating peatlands and permafrost into a dynamic global vegetation model: 1. evaluation and sensitivity of physical land surface processes,” Global Biogeochemical Cycles 23 (2009).DOI
  9. [9]Matthieu Guimberteau, Dan Zhu, Fabienne Maignan, Ye Huang, Chao Yue, Sarah Dénéle-Clédec, Catherine Ottlé, Albert Jornet-Puig, Ana Bastos, Pierre Laurent, et al., “ORCHIDEE-MICT (v8. 4.1), a land surface model for the high latitudes: model description and validation,” Geosci. Model Dev. 11, 121–163 (2018).DOI
  10. [10]David M Lawrence, Andrew G Slater, Vladimir E Romanovsky, and Dmitry Nicolsky, “Sensitivity of a model projection of near-surface permafrost degradation to soil column depth and representation of soil organic matter,” J. Geophys. Res.: Earth Surface 113 (2008).DOI
  11. [11]Christian G Andresen, David M Lawrence, Cathy J Wilson, A David McGuire, Charles Koven, Kevin Schaefer, Elchin Jafarov, Shushi Peng, Xiaodong Chen, Isabelle Gouttevin, et al., “Soil moisture and hydrology projections of the permafrost region - a model intercomparison,” The Cryosphere 14, 445–459 (2020).DOI
  12. [12]Christoph Beckermann, H-J Diepers, Ingo Steinbach, Alain Karma, and Xinglin Tong, “Modeling melt convection in phase-field simulations of solidification,” J. Comput. Phys. 154, 468–496 (1999).DOI
  13. [13]Benjamin Favier, Jhaswantsing Purseed, and Laurent Duchemin, “Rayleigh–Bénard convection with a melting boundary,” J. Fluid Mech. 858, 437–473 (2019).DOI
  14. [14]Rui Yang, Kai Leong Chong, Hao-Ran Liu, Roberto Verzicco, and Detlef Lohse, “Abrupt transition from slow to fast melting of ice,” Phys. Rev. Fluids 7, 083503 (2022).DOI
  15. [15]Eric W Hester, Louis-Alexandre Couston, Benjamin Favier, Keaton J Burns, and Geoffrey M Vasil, “Improved phase-field models of melting and dissolution in multi-component flows,” Proc. Royal Soc. A 476, 20200508 (2020).
  16. [16]Andrew Rees and Ioan Pop, “Vertical free convective boundary-layer flow in a porous medium using a thermal nonequilibrium model,” J. Porous Media 3, 31 (2000).
  17. [17]Ziqi Wang, Enrico Calzavarini, Chao Sun, and Federico Toschi, “How the growth of ice depends on the fluid dynamics underneath,” Proc. Natl. Acad. Sciences 118, e2012870118 (2021).
  18. [18]E. R. Lapwood, “Convection of a fluid in a porous medium,” Math. Proc. Cambridge Phil. Soc. 44, 508 (1948).DOI
  19. [19]Nurzahan Banu and DAS Rees, “Onset of darcy–benard convection using a thermal non-equilibrium model,” Int. J. Heat Mass Trans. 45, 2221–2228 (2002).
  20. [20]Donald A Nield, Adrian Bejan, et al., Convection in porous media, Vol. 3 (Springer, 2006).
  21. [21]L.I. Rubenstein, The Stefan problem, Vol. 27 (American Mathematical Soc., 1971).
  22. [22]DR Hewitt, “Vigorous convection in porous media,” Proc. Royal Soc. A 476, 202000111 (2020).DOI
  23. [23]G Boffetta, M Borgnino, and S Musacchio, “Scaling of Rayleigh-Taylor mixing in porous media,” Phys. Rev. Fluids 5, 062501 (2020).arxiv.org/abs/2006.07175
  24. [24]Martin Chaigné, Sabrina Carpy, Marion Massé, Julien Derr, Sylvain Courrech du Pont, and Michael Berhanu, “Emergence of tip singularities in dissolution patterns,” Proc. Natl. Acad. Sciences 120, e2309379120 (2023).DOI
  25. [25]Julia Boike, Kurt Roth, and Pier Paul Overduin, “Thermal and hydrologic dynamics of the active layer at a continuous permafrost site (Tavmyr Peninsula, Siberia),” Water Resour. Res. 34, 355–363 (1998).DOI
  26. [26]Jens Weismüller, Ute Wollschläger, Julia Boike, Xicai Pan, Q Yu, and Kurt Roth, “Modeling the thermal dynamics of the active layer at two contrasting permafrost sites on svalbard and on the tibetan plateau,” The Cryosphere 5, 741–757 (2011).DOI
  27. [27]Jørgen Hollesen, Bo Elberling, and Per-Erik Jansson, “Future active layer dynamics and carbon dioxide production from thawing permafrost layers in Northeast Greenland,” Glob. Change Biol. 17, 911–926 (2011).DOI
  28. [28]Tom Gleeson, Leslie Smith, Nils Moosdorf, Jens Hartmann, Hans H Dürr, Andrew H Manning, Ludovicus PH van Beek, and A Mark Jellinek, “Mapping permeability over the surface of the Earth,” Geophys. Res. Lett. 38 (2011).DOI
  29. [29]Robert P Chapuis, “Predicting the saturated hydraulic conductivity of soils: a review,” Bull. Eng. Geol. Environ. 71, 401–434 (2012).DOI
  30. [30]Brian A Ebel, Joshua C Koch, and Michelle A Walwoord, “Soil physical, hydraulic, and thermal properties in interior Alaska, USA: Implications for hydrologic response to thawing permafrost conditions,” Water Resour. Res. 55, 4427–4447 (2019).DOI
  31. [31]Samuel C Zipper, Pierrick Lamontagne-Hallé, Jeffrey M McKenzie, and Adrian V Rocha, “Groundwater controls on postfire permafrost thaw: Water and energy balance effects,” J. Geophys. Res.: Earth Surface 123, 2677–2694 (2018).DOI

Paper details

vlnc:W_byp74e78
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents