Pole, Lettau (1979) deduced the nondimensional gradi-
Traditionally, Ricr is assumed to be 0.200.25 (Busch
ent functions
1973, Businger 1973). But Mahrt (1981) and Heine-
mann and Rose (1990) have reported that a larger value
φm (ζ) = (1 + 4.5ζ)3 / 4
(81)
is sometimes indicated. Woods (1969) may have ex-
plained this range of values by demonstrating that there
and
is a hysteresis in Ricr. He concluded that a turbulent
flow becomes laminar when Ri exceeds 1, but a lami-
φh (ζ) = φm (ζ)2 = (1 + 4.5ζ)3 / 2
(82)
nar flow does not become turbulent until Ri falls below
0.25 (see also Plate 1971, p. 76). In his observations at
which are also depicted in Figure 5. Lettau's φh func-
South Pole, however, Lettau (1979) frequently found
tion predicts a very steep scalar gradient with increas-
turbulence to exist even when Ri exceeded unity; he
ing ζ that has not been confirmed by independent ob-
thus concluded that, in stable conditions, there was no
servations.
critical Richardson number. Monin and Yaglom (1971,
Although Lettau (1979) developed his functions es-
p. 440 f.) also argued that there seems to be no critical
pecially to treat very stable conditions, the log-linear
Richardson number in stable conditions. In light of this
form (eq 80) is also often applied in very stable condi-
controversy, I show in Table 1 what the four formula-
tions, though it has been tested only for 0 ≤ ζ <1 (e.g.,
tions for φm and φh depicted in Figure 5 predict as the
Dyer 1974, Hicks 1976, Yaglom 1977). As an alterna-
behavior of Ri in very stable conditions.
tive to the log-linear form for 0 ≤ ζ ≤10, Holtslag and
de Bruin (1988) and Beljaars and Holtslag (1991) pro-
Table 1. Predicted behavior of the Deacon and
posed the function
Richardson numbers in very stable conditions accord-
ing to the four gradient formulation shown in Figures
4 and 5.
φm (ζ) = φh (ζ)
limit ζ → ∞
φm, φh
(83)
Dm
Dh
Ri
= 1 + 0.7ζ + 0.75ζ(6 - 0.35ζ) exp(-0.35ζ).
1 + 5ζ
1/
0
0
5=
0.20
1 + 7ζ
1/
I will refer to this as the Dutch formulation.
0
0
7=
0.14
In Figure 5, we see that there is little difference
φm = (1 + 4.5ζ)3/4
among the four functions suggested for φm and φh when
ζ
1/
1/2
4
0 ≤ ζ ≤ 0.5. Thus, because the log-linear form is much
φh = (1 +
4.5ζ)3/2
simpler mathematically than Lettau's (1979) or the
1 + 0.7 ζ + 0.75ζ(6 0.35ζ) exp(0.35ζ)
0
0
1.4
Dutch formulation, it might be preferable for weakly
stable conditions. Some other considerations, however,
might help us decide which of the four formulations is
Lettau (1979; see also Viswanadham [1979, 1982])
best in more stable conditions.
described two other parameters that characterize
The gradient Richardson number is
surface-layer profiles, the Deacon numbers for wind
speed and potential temperature, Dm and Dh, respec-
Θ / dz
g
tively. These quantify profile curvature. For wind speed
Ri =
.
(84)
(
U / dz)
2
Tv
-z(d 2U / dz 2 )
Dm ≡
.
(87)
From eq 78 and the definition of ζ (eq 58) we can write
dU / dz
this as
With eq 78a, it is easy to show that
(
)
g t* / kz φh (ζ) ζ φh (ζ)
ζ dφm (ζ)
Dm (ζ) = 1 -
.
(88)
Ri =
=
φm (ζ) dζ
.
(85)
(
)
Tv u / kz 2 φ2 (ζ) φ2 (ζ)
m
m
*
Similarly, for potential temperature
In stable conditions, turbulence ceases when the
-z(d 2Θ / dz 2 )
Dh ≡
Richardson number exceeds a critical value, Ricr. Thus,
(89)
dΘ / dz
we should expect accurate profile functions to yield this
critical value through eq 85; that is
which, with eq 78b, becomes
lim Ri = Ricr .
ζ dφh (ζ)
(86)
Dh (ζ) = 1 -
ζ→∞
.
(90)
φh (ζ) dζ
13