Outline • Announcements • Where were we? • Measuring unsaturated flow • Soil water diffusivity Soil Physics 2010 Announcements • Homework 4 due March 3 • Excel Solver demo on course website Soil Physics 2010 Where were we? Ks is pretty easy. K(q) is slow, and hard to control. • Apply water at steady q < Ks • Wait till outflow = inflow • Measure q and/or y across a “test interval” Soil Physics 2010 • • • • • Prevent evaporation Water evenly, no disturbance Tall column, or tension at bottom Tensiometer can change flow Measure q with gamma-rays How do we measure K(q) in the lab? K(q) is slow, and hard to control. Other methods: • Centrifuge • Evaporation • One-step • Multi-step As q decreases: Soil Physics 2010 Slower Harder to control More uncertainty How do we measure K(q) in the field? • Instantaneous profile • Various others • Best solved with inverse methods The “forward problem”: Given the parameters and boundary conditions, simulate what happened (or will happen). The “inverse problem: Given the data and the boundary conditions, estimate the parameter values. Requires way more computer resources than a simple statistical fit. (The Excel Solver solves an inverse problem.) Soil Physics 2010 What’s the equation for K(q)? Most estimation methods start with the WRC [q (y)], because it’s our best estimate of the pore size distribution Capillary tube approach Cut-and-rejoin approach 1950s – 1960s Soil Physics 2010 Estimating K(q) from q(y) Known volume of pores with a known radius ym r Soil Physics 2010 q How do you average conductivities? Network Parallel Serial (like bundle of tubes) Z=∞ 20 > Z > 2 N K Arith f i K i i 1 Soil Physics 2010 K Geom f i K i i 1 Z=2 n K harm 1 N fi i 1 K i Beyond averages Effective Medium Theory: Network What value of conductor, if it replaced every other conductor in the system, would give an equivalent system conductivity? N 2 < Z < 50 N i 1 Soil Physics 2010 i 1 Ki K fi 0 * Ki 2K Ki K fi 0 Ki Z 1 K * 2 * * What is the equation for K(q)? Examples (many, but not all, found in Hillel): K q K sq q K q K s K S Ks S K y n n 22.5 y n Ks K y n 1 ay K y Ks y 1 y c …but none of them works (Brooks & Corey, 1964) Soil Physics 2010 Ks n Widely used models: take your pick q Burdine: K S K s S 2 y y 0 0 2 r dq 2 r dq q Mualem: K S K s van Genuchten’s q(y) model combined with Mualem’s K(S) model: Soil Physics 2010 y 2 dq 0 r S 2 y r dq 0 2 K S K s S 1 1 S 2 m 1m Hydraulic diffusivity Different forms of Richards’ equation have different advantages and disadvantages. q h K q t x x “Mixed form”: both q and h q q D q t x x “q ” form h h c h K h t x x “h” form Soil Physics 2010 “The same equations have the same solutions” Richard Feynman h h c h K h t x x q c(h), also called c(q), h “differential water capacity” h q h q So c h t h t t Soil Physics 2010 “The same equations have the same solutions” Richard Feynman q q D q t x x K q K q D q q cq h “hydraulic diffusivity” q h q h So D q K q K q x q x x Soil Physics 2010 h(q), K(q), c(q) and D(q) 1.E+01 1.E+04 Sand 1.E+03 0 clay 0.1 0.2 0.3 0.4 0.5 0.6 1.E-01 loam 1.E+02 1.E-03 1.E+01 Sand clay 1.E+00 0 1.E-01 0.1 hq 0.2 0.3 0.4 0.5 0.6 1.E-05 loam q c q h 1.E-07 1.E+02 1.E-02 1.E+02 1.E+00 0 0.1 0.2 0.3 0.4 0.5 0.6 1.E+00 1.E-02 0 1.E-04 0.1 0.2 0.3 0.4 0.5 0.6 1.E-02 1.E-06 Sand 1.E-08 clay loam 1.E-10 Sand 1.E-04 clay K q D q cq loam 1.E-12 1.E-14 Soil 1.E-16Physics 2010 K(q) 1.E-06 1.E-08 So what’s the point? “The same equations have the same solutions” Richard Feynman C C D 2 t x Diffusion equation T T DT 2 t x Heat flow equation 2 2 Extremely well studied equations q q q D q Dq 2 t x x x 2 Soil Physics 2010 Hydraulic diffusivity equation if D(q) constant in x Cost / benefit analysis for the hydraulic diffusivity equation: Cost: assumptions of • No hysteresis • Horizontal only • c(q) and K(q) (and thereby D(q)) don’t change in x or t Benefit: an equation that q q Dq 2 t x • Has only 1 variable (q) that changes in x and t • Has only 1 function (D) that needs to be measured or estimated • Has centuries of mathematical history Soil Physics 2010 2 What good is math history? A cool trick: the Boltzmann transformation q q Given Dq 2 , t x 2 notice that D has units of L2/t, which is characteristic of the diffusion equation. Introduce a new variable B B dq dq D q 2 2 dB dB 2 Soil Physics 2010 ≡ x/t1/2. Then • q varies only in B • ODE, not PDE 1 function in 1 unknown Boltzmann variable B B dq dq D q 2 2 dB dB 2 ≡ x/t1/2 Horizontal infiltration Bruce & Klute setup Hydraulic diffusivity experiment q Soil Physics 2010 B What use is it? • From this (easy) experiment we get D(q) • From a water retention curve (also fairly easy) we get c(q) • Combining them, we get K(q), which is way hard to measure. q Soil Physics 2010 B