1. Transport and mixing 1.1 The material derivative LetV() x, t be the velocity of a fluid at the pointx = ()x,, y z and timet . Consider also some scalar fieldχ()x, t such as the temperature or density. We are interested not only in the partial derivative ofχ with respect to time,∂χ ∂⁄ t , but also in the time derivative following the motion of the fluid,dχ ⁄ dt . The latter is the so-called material derivative: dχ ⁄dt = lim [] χ(x+ V() x, t δt, t+ δ t )– χ()x, t δ⁄ t δt → 0 . (1.1) = (∂⁄ ∂t + V ⋅ ∇ )χ (),, ∇() ∂ ,,∂ ∂ In Cartesian coordinates,V = u v w ,= x y z and dχ ⁄dt = ∂ χ ∂⁄ t + u∂χ ∂⁄ x +v∂χ ∂⁄ y + w∂χ ∂⁄ z . (1.2) We will also be using spherical coordinates. The notation()u,, v w is conventional for the (eastward, northward, radially outward) components of the velocity field. In addition to the radial distance from the origin,r , we use the symbolθ for latitude (–π ⁄ 2 at the south pole,π ⁄ 2 at the north pole) andλ for longitude (ranging for zero to2π ). The gradient operator in these coordi- nates is ∇= λˆ ()r cosθ –1(∂⁄ ∂λ )+ θˆ r–1()∂⁄ ∂θ + rˆ()∂⁄ ∂r , (1.3) so that d⁄ dt = ∂⁄ ∂t +()r cosθ –1u()∂⁄ ∂λ +r–1v()∂⁄ ∂θ + w()∂⁄ ∂r . (1.4) The material derivative of a vector, such as the velocityV itself, is defined just as in 1.1. But care is required when considering the material derivative of a component of a vector if the unit vectors of one’s coordinate system are position dependent, as they are in spherical coordi- nates. For example, the radial component of the material derivative of the velocity is not equal to the material derivative of the radial component of the velocity; rather, dV d drˆ rˆ ⋅ ------- = -----()rˆ ⋅ V – V ⋅ ----- . (1.5) dt dt dt The second term on the rhs is sometimes referred to as the metric term. After obtaining expressions fordrˆ ⁄ dt , etc., one arrives at the spherical-coordinate expression for the acceleration of a fluid parcel (see Problem 1.1): -1.2- du λˆ ------ – uvtanθ ⁄ r+ uw⁄ r dt dV dv 2 ------- = + θˆ ------ + u tanθ ⁄ r+ vw⁄ r . (1.6) dt dt dw 2 2 + rˆ ------- – ()u + v ⁄ r . dt Another case of interest is the material derivative of an infinitesimal material line segment. Suppose thatx andx + δx are infinitesimally close to each other, with the vectorδx pointing from one point to the other, and letδx()t be the evolution of this vector assuming that its end- points move with the flow. Then, from the figure below,dδx ⁄dt = () δx ⋅ ∇ V . To make sure that we understand this notation, suppose that the line segment is oriented vertically at the time in question. Then, in Cartesian coordinates, dδx⁄ dt = δz()∂ u ∂⁄ z ; dδy⁄ dt = δz()∂ v ∂⁄ z ; dδz⁄ dt = δz()∂ w ∂⁄ z . We refer to the third term as stretching, as it lengthens or shortens the line segment, and the other two terms as tilting, as they change the orientation of the line segment without changing its length. This equation and the terminology will be important when we discuss vorticity dynamics. V( x + δx, t )δt δx()t δx()t + δt V( x, t )δt The fractional rate of the change of the length of a infinitesimal vector pointing in the x- direction, moving with the fluid, is∂u ∂⁄ x and similarly for y and z. Therefore, the fractional rate of change in the volume of an infinitesimal piece of fluid per unit time is ()1 ⁄ δV dδV⁄ dt = ()1 ⁄ δxδyδz d()δxδyδz ⁄ dt = ()1 ⁄ δx dδx⁄ dt +()1 ⁄ δy dδy⁄ dt + ()1 ⁄ δz dδz⁄ dt (1.7) = ∂u ∂⁄ x +∂v ∂⁄ y + ∂w ∂⁄ z in Cartesian coordinates, or, more generally, ()1 ⁄ δV dδV⁄ dt = ∇ ⋅ V . (1.8) With this physical interpretation in mind, the expression for the divergence in spherical coordinates can be seen to take the form -1.3- 2 –1 ∂u –1∂[]()cosθ v –2∂()r w ∇ ⋅ V = ()r cosθ ------ +()r cosθ --------------------------- + r -----------------, (1.9) ∂λ ∂θ ∂r usingδV∝ r2 cosθdrdθdλ . 1.2 Conservation of mass Ifρ()x, t is the density of the fluid, the statement that mass is neither created nor destroyed can be written in two equivalent forms: ∂ρ ------ = –∇⋅ () ρV (1.10) ∂t dρ ------ = –ρ∇ ⋅ V . (1.11) dt The first (flux) form is more convenient when considering a fixed volume in space, since, using Gauss’s Theorem, it immediately yields ∂ ρdV = – ρV⋅ nˆ dA , (1.12) ∂t ∫∫ ∫ ∫∫ V S where V is any volume, S its surface, andnˆ a unit vector normal to S and directed outwards. The latter (advective) form is more convenient when following the fluid, and expresses the fact that the divergence∇ ⋅ V is equal to the fractional change, per unit time, in the density of an infinitesimal parcel of fluid: d()ρδV ⁄ dt =0 ⇒ () 1 ρ⁄ dρ ⁄ dt + ()1 ⁄ δV dδV⁄ dt = 0 . For any other scalar fieldχ with source S per unit mass, we can write the equation in either advective form, dχ ------ = S , (1.13) dt or flux form, ∂()ρχ -------------- + ∇⋅ () ρVχ = ρS . (1.14) ∂t 1.3 Eulerian vs. Lagrangian descriptions of fluid flow Consider a point moving with the velocity fieldV . We refer to such a point as a fluid par- ticle. Ifa is the position of this particle at timet0 , its position at any later time is t x() a, t = ∫ V() x() a, t′ , t′ dt′ . (1.15) t0 -1.4- Rather than consider fields such asV andχ as functions ofx andt (the Eulerian description), χ((), , ) χ (), one can instead consider these fields to be functions ofa andt :x a t t = L a t . The χ ⁄ ∂χ ∂⁄ material derivatived dt is then simplyL t . This Lagrangian description is occasionally convenient because of its closer analogy with the classical mechanics of point masses. While we will be working with the Eulerian equations of motion exclusively, it will often be important, if not critical, to maintain a Lagrangian perspective, to think not only about fields evolving inx and t , but also about the trajectories of fluid particles and how quantities of interest evolve following the flow. Simple Eulerian flows do not necessarily produce simple particle trajectories. Given the velocity fieldV , the advection of a passive tracer is formally a linear problem: ∂χ ∂⁄ t = –V ⋅ ∇χ . Yet this linear problem can also be solved by first solving the nonlinear prob- lem of computing trajectories:d x ⁄ dt = V() x, t . The latter problem is identical in form to that of a general dynamical system, except that here the “phase space” of the system is three-dimen- sional physical space. Autonomous systems (steady flow, in this context) with three degrees of freedom can produce chaotic, extremely complex, trajectories. Non-autonomous systems (unsteady flows) in two dimensions can possess a similar level of complexity. In contrast, auton- omous systems in two dimensions cannot produce trajectories more complicated than limit cycles, that is, simple closed streamlines. Consider a 2-dimensional flow with Cartesian components()u, v that is non-divergent in the x-y plane. Such a flow is efficiently described in terms of a streamfunction,ψ , where u = –∂ψ ∂⁄ y andv = ∂ψ ∂⁄ x . Note also that this is a Hamiltonian system withψ as the Hamiltonian and x and y as conjugate variables. We analyze 2-dimensional flows more exten- sively in Chapter 4. 1.4 Diffusion and friction In the atmosphere there is a tremendous disparity between the scales of motion that con- tain the bulk of the energy in the flow, on which our interest will be focused, and scales at which molecular diffusion and viscosity become significant. If we consider some fieldχ that is advected by the flowV as well as diffused with the constant kinematic diffusivityD , then ∂χ ∂⁄ t = –V ⋅ ∇χ + D∇2χ . (1.16) If the characteristic scale of the velocities in the flow isU , then for variations inχ on the scaleL , the characteristic ratio of diffusion to advection is measured by the Peclet number, V ⋅ ∇χ UL ------------------- ∼ -------- ≡ Pe . (1.17) D∇2χ D The time required to diffuse away a feature of scaleL in the absence of flow isL2 ⁄ D ; Pe is the ratio of this diffusive time scale to the advective timeLU⁄ . Ifχ is one component of the velocity -1.5- field itself, so thatD is replaced by the kinematic viscosityν , then the corresponding ratio is UL ν⁄ , the Reynolds number. The order of magnitude ofν orD in the atmosphere near the earth’s surface is 10–5 m2 ⁄ s . Large-scale flows in the earth’s atmosphere typically have velocity scales of 10 m⁄ s and vertical scales larger than1 km , leading to Peclet or Reynolds numbers at least of the order of109 . A weak gust of wind near the surface with a speed of1 m⁄ s and a modest ver- tical scale of10 m still has advection dominating diffusion or viscosity by a factor of106 . Only on scales of a millimeter or so do the latter come into play directly. Molecular diffusivities in the ocean are even smaller (see Problem 1.4). One should not assume on this basis that molecular diffusion and dissipation can be ignored. The atmosphere is a forced, dissipative system through which energy is flowing. The generation of kinetic energy integrated over the depth of the atmosphere is estimated to be 2 - 3 W⁄ m2 , averaged over the globe. Dissipation by molecular viscosity must balance this gen- eration in a steady state. The kinetic energy per unit mass of the atmosphere is roughly 102 m2 ⁄ s2 on average, and the total mass of an atmospheric column is104 kg⁄ m2 , so one might estimate that this energy, if not constantly replenished, would be dissipated (converted into heat) within a few days.
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages11 Page
-
File Size-