Material derivative ($D/Dt$)

The material derivative follows a moving continuum parcel:

The Eulerian time derivative measures a continuum field's rate of change at a fixed Eulerian coordinate. For a velocity field $\mathbf{u}$, the material derivative instead follows a fluid element and satisfies $Df/Dt={\partial }_{t}f+\mathbf{u}\cdot \nabla f$. A nonuniform steady flow has zero Eulerian time derivative of velocity but can have nonzero material acceleration.

For a smoothly moving volume whose boundary has normal velocity $v_b\cdot n$, the derivative of an integral equals the local time-change integral plus the boundary-transport flux. The boundary term comes from the thin layer swept by the moving surface during a short time interval. Taking $f=1$ gives $d|V|/dt=\int v_b\cdot ndS$. If the boundary moves with the fluid, this combines with the divergence theorem to give the integral of the material derivative plus $f$ times the velocity divergence.

Convective acceleration is the spatial part $(\mathbf{u}\cdot \nabla )\mathbf{u}$ of the material derivative of fluid velocity. It can remain nonzero in steady flow, because a moving particle passes through positions with different velocities. The identity $(\mathbf{u}\cdot \nabla )\mathbf{u}=\nabla (|\mathbf{u}{|}^2/2)-\mathbf{u}\times (\nabla \times \mathbf{u})$ relates it to vorticity.

For a velocity field $\mathbf{u}=U(r)e_{\varphi }$ in spherical coordinates, the convective acceleration is $(\mathbf{u}\cdot \nabla )\mathbf{u}=-U(r{)}^2(e_r+\cot \theta e_{\theta })/r$, away from the polar axis. Each streamline is a circle of cylindrical radius $r\sin \theta$, so this is centripetal acceleration of magnitude $U^2/(r\sin \theta )$ directed toward the polar axis. The terms containing $U^{\prime }(r)$ cancel because a particle on such a streamline stays at fixed $r$ and $\theta$.

The material acceleration is the rate of change of velocity following a continuum parcel:

A scalar potential whose negative spatial gradient is the material acceleration. For a smooth global potential, the spatial curl of the material acceleration vanishes. On a simply connected domain the converse holds under the usual smoothness assumptions. Its physical dimension is squared length divided by squared time.

A homogeneous motion has an acceleration potential exactly when $A=\ddot{F}{F}^{-1}$ is a symmetric matrix. Necessity follows because $A$ is the spatial gradient of the material acceleration, which must be a negative Hessian matrix. For sufficiency, put $\mathbf{y}=\mathbf{x}-\mathbf{c}(t)$ and take $\xi =-{\mathbf{y}}^{T}A\mathbf{y}/2-\ddot{\mathbf{c}}\cdot \mathbf{y}+g(t)$. Direct spatial differentiation gives $-\nabla \xi =A\mathbf{y}+\ddot{\mathbf{c}}$. Equivalently $F^T\dot{F}-{\dot{F}}^{T}F$ is constant: its time derivative is $F^T\ddot{F}-{\ddot{F}}^{T}F$.

An Eulerian coordinate labels a fixed position in space at which continuum fields are observed.

A Lagrangian coordinate labels a material particle and remains attached to it as the continuum moves.

An Eulerian time average holds spatial position fixed while averaging a time-dependent field:

A Lagrangian trajectory follows one marked fluid particle and solves

A fluid element is an infinitesimal material parcel followed along a Lagrangian trajectory.

A material volume is a region containing the same fluid particles at every time: its boundary moves with the fluid velocity field. Its mass is fixed by mass conservation, while its volume can change in compressible motion. Applying the Reynolds transport theorem to its volume gives $d|V(t)|/dt={\int }_{V(t)}\nabla \cdot \mathbf{u}dV$. A material volume element is the infinitesimal version.

A material volume element contains the same fluid particles as it moves and deforms. If its edges transform with the local velocity gradient, its volume changes at a rate equal to the trace of that gradient, giving $D(\delta V)/Dt=(\nabla \cdot \mathbf{u})\delta V$. Mass conservation states $D(\rho \delta V)/Dt=0$, not that its volume is always fixed. These two identities give the material form of the continuity equation, $D\rho /Dt=-\rho \nabla \cdot \mathbf{u}$.

Stokes drift is the difference between the mean velocity of fluid particles and the Eulerian mean velocity at fixed positions. It appears at second order because particles sample the oscillatory velocity at displaced positions.

For prescribed one-dimensional velocity $u=u_0\sin (kx-\omega t)$, let $\alpha =k{X}_0$, $\theta =\alpha -\omega t$ and expand the Lagrangian trajectory as $x=X_0+u_{0}a+u_0^{2}b$. The Taylor expansion of $\dot{x}=u(x,t)$ gives $\dot{a}=\sin \theta$ and $\dot{b}=ka\cos \theta$. Initial position $X_0$ requires $a=(\cos \theta -\cos \alpha )/\omega$ andThe secular velocity $k{u}_0^2/(2\omega )$ is Stokes drift, even though the Eulerian time average vanishes. This fixed-time expansion requires $|k{u}_0/\omega |\ll 1$ and is not uniform once the secular displacement changes the phase by order unity.

Two plane internal gravity waves can each have zero Stokes drift but give a nonzero drift through interference. The displacement of one wave need not be perpendicular to the other wave's wave vector. For upward waves with wave vectors $(\kappa ,-\kappa )$ and $(-\kappa ,-\kappa )$ and vertical amplitudes $W$, the cross-wave drift is $(0,-2\kappa W^2\cos (2\kappa x)/\omega )$ with $\omega =N/\sqrt{2}$. It has zero spatially averaged vertical flux. It cannot on its own satisfy an impermeable oscillating boundary: the second-order Eulerian mean velocity supplies the necessary normal correction there.

A fluid displacement measured from an initial position must vanish at the initial time. Integrating a harmonic velocity therefore introduces phase-dependent constants. Mean-orbit labels absorb those constants and leave purely oscillatory displacements. The instantaneous displacement correction to velocity can differ under the two reference conventions, but its period mean gives the same leading Stokes drift.