Plant Hydraulics

Water loss during day-time transpiration drives plants to draw water from the soil by roots and transport it through the stem to leaves. The plant hydraulics code solves for the volumetric water content in the above ground plant material, $\theta_{l}$.

The volume flux of water $q$ (m/s) is given by Darcy's law as

\[\begin{align} q = - K(\psi) \frac{dh}{dz} \end{align}\]

where $h = \psi+z$ is the head (in meters), and $K$ is the conductivity (units of m/s). We approximate the flux between two depths, indexed with 2 at height $z_2$ and indexed with 1 at height $z_1<z_2$, using finite difference as

\[\begin{equation} q \approx -\frac{K_1(\psi_1) K_2(\psi_2)}{K_1(\psi_1)+K_2(\psi_2)} \times \bigg[\frac{\psi_2 - \psi_1}{z_2 - z_1} +1\bigg]. \end{equation}\]

The change of water volume (m$^3), V$, is then

\[\begin{equation} \frac{d V}{dt} = q_{roots}\sigma_{roots} - \tau \sigma_{leaf}, \end{equation}\]

where $\tau$ is a transpiration volume flux per unit emitting area, and $\sigma$ is the total emitting/conducting area.

This currently holds for a single pathway. To convert to fluxes from an entire surface, we can multiply by the number of individuals $N$. We can make use of the fact that $N\sigma/A$, where $A$ is the area of the ground those $N$ individuals are occupying, is the area index for that plant type. Following CLM, we incorporate a root and leaf area index (RAI, LAI) in order to model fluxes across an entire grid cell.

Then we have:

\[\begin{equation} \frac{d v}{dt} = q_{roots}RAI - \tau LAI, \end{equation}\]

where $v$ now represents the volume of water in that compartment (of a bulk plant) per unit ground area.

We also need to convert from the variable $v$ to $\psi$, in order to compute root extraction with the soil. To do so, we can convert $v$ to the volumetric water content, and from $\theta$ to $\psi$ using the retention curve. To convert, let the volume of water per area of ground be $V$, and $H_{canopy}$ the canopy height. Then,

\[\begin{equation} \theta=\frac{V}{A_{ground}} \times \frac{A_{ground}}{A_{leaf}} \times \frac{1}{H_{canopy}} = \frac{v}{H_{canopy} \times LAI }. \end{equation}\]

Substituting in the volumetric water content, we have

\[\begin{equation} \frac{d \theta}{dt} = \frac{q_{roots}RAI - \tau LAI}{H_{canopy} LAI}. \end{equation}\]

In the code, the root cross-sectional factor is implemented with a harmonic-mean coupling between RAI and LAI. This ensures that the flux goes to zero when either area index goes to zero.

We can also account for the distribution of roots as a function of depth. A quantity that is modeled in plant hydraulic models is the root fraction $P(z)$, satisfying $\int P(z) dz = 1$. Instead of having a single root at one discrete location, we can distribute the root system over different depths using $P(z)$. The total flux from roots between $z$ and $z+dz$ is given by

\[\begin{equation} dq_{roots}(z) = -P(z) q(z) dz. \end{equation}\]

Then the net root flux for the plant system would sum over this

\[\begin{equation} q_{roots} = -\int_{z_{min}}^{z_{sfc}} P(z) q(z) dz , \end{equation}\]

where $z_{min}$ is the minimum soil layer of the simulation.

The sink term of the soil is in terms of a volumetric fraction change, i.e. we need a volume of water per volume of soil per second. We can obtain this with

\[\begin{equation} S(z) = -(RAI) \frac{dq_{roots}(z)}{dz} = (RAI) P(z) q(z). \end{equation}\]

The sign change occurs in the expression for $S(z)$ because a positive value of $q_{roots}$ indicates flow from the soil to the plant. This is a sink term for the soil.

In order to close the set of equations, the user will have to specify $K(\psi)$ and a function $\psi(\theta)$. In our current implementation, we use a linear retention curve $\psi(\theta)$, and a Weibull permeability curve $K(\psi)$:

\[\begin{equation} K(\psi) = \begin{cases} K_{\rm sat}\exp{(\psi/\psi_{63})^c} & \psi <0 \\ K_{\rm sat} & \psi > 0 \end{cases} \end{equation}\]

and

\[\begin{equation} \psi(\theta) = \frac{1}{a}(\theta/\nu - 1). \end{equation}\]

The model needs the following parameters:

DriversSymbolUnitApproximate Value
Linear retention curve slopea1/m0.002
Saturated conductivity$K_{\rm sat}$m/s1e-7
Potential at 1/e loss in conductivity$\psi_{63}$m-400
Weibull exponential parameter$c$-4
Porosity$\nu$-1e-4