Appendix Resolving the upper-ocean warm layer improves the

advertisement
1
Appendix
2
Resolving the upper-ocean warm layer improves the simulation of the
3
Madden-Julian Oscillation
4
Wan-Ling Tseng1,2, Ben-Jei Tsuang3, Noel S. Keenlyside4, Huang-Hsiung Hsu1, &
5
Chia-Ying Tu1
6
1
Research Center for Environmental Changes, Academia Sinica, Taipei, Taiwan.
7
2
GEOMAR | Helmholtz-Zentrum für Ozeanforschung, Kiel, Germany.
8
3
National Chung-Hsing University, Taichung, Taiwan.
9
4
Geophysical Institute and Bjerknes Centre, University of Bergen, Bergen, Norway.
10
Corresponding author: W.-L. Tseng, Research Center for Environmental Changes,
11
Academia Sinica, Taipei, 115, Taiwan. (wtseng@gate.sinica.edu.tw)
12
TEL:886-2-2652-5174
13
FAX:886-2-2783-3584
14
15
16
The one-column ocean model, Snow-Ice-Thermocline (SIT), is based on Tsuang
17
et al. (2001) that follows the turbulent kinetic energy (TKE) approach of Gaspar et al.
18
(1990). SIT parameterizes ice-formation, the warm-layer effect and the cool-skin
19
effect, and introduces a surface effective thickness (β„Žπ‘’ ) to improve the simulation of
20
upper ocean temperature (Tu and Tsuang 2005; Tsuang et al. 2009). The model has
21
been verified at a tropical ocean site (Tu and Tsuang 2005), in the South China Sea
22
(Lan et al. 2010), and in the Caspian Sea (Tsuang et al. 2001). The melt and formation
23
of snow and ice above a water column has been introduced (Tsuang et al. 2001).
24
However, these parts are not utilized in the experiments here.
25
SIT determines water temperature, salinity, and u, v currents (denoted as
26
variable X) in each depth according to the energy, salinity and momentum budget in a
27
one-column model (e.g., Gaspar et al., 1990) as:
28
πœ•π‘‹Μ…
πœ•π‘§
=−
Μ…Μ…Μ…Μ…Μ…Μ…Μ…
πœ•π‘‹′𝑀′
πœ•π‘§
(1)
29
where X ' w' (positive upward) is the vertical flux of scalar X, a transport property. It
30
can be temperature (T) (K), horizontal velocities (u, v) (m s-1) or salinity (S) (practical
31
salinity ‰). Variable z is the height (positive upward). The over bar represents a time
32
averaged value. Furthermore, the penetration of solar radiation is parameterized
33
according to a nine-band equation (Paulson and Simpson 1981), where the absorption
34
coefficients of solar radiation are set according to (Fairall et al. 1996).
35
To simulate the warm layer the vertical resolution of the water column needs to
36
be fine enough to resolve variations in the upper 10 m of the ocean. Below the cool
37
skin, the vertical flux X ' w' is parameterized using the classical K approach as:
38
πœ•π‘‹Μ…
Μ…Μ…Μ…Μ…Μ…Μ…
𝑋′𝑀′ = −(π‘˜ + 𝜈) πœ•π‘§
39
where k and ν are eddy and molecular diffusion coefficients (m2 s-1), respectively.
40
Furthermore k and ν are designated as km and νm for momentum, and as kh and νh for
41
temperature and salinity. The surface net heat flux (latent heat flux + sensible heat
42
flux + net long wave radiation) are used as the upper boundary condition for heat Eq;
43
the wind stress is used as the upper boundary condition for momentum Eq; the net
44
fresh water salinity flux, (evaporation - precipitation - river inflow) multiply by
45
surface salinity, is used as the upper boundary condition for salinity Eq.
(2)
46
To simulate the cool skin effect, the eddy diffusion coefficient for heat, kh,
47
within the cool skin and the eddy diffusion coefficient for momentum, km, within the
48
viscous layer are set to zero: molecular transport is the only mechanism for vertical
49
diffusion of heat and momentum in the cool skin and in the viscous layer, respectively
50
(Hasse 1971; Grassl 1976; Wu 1985). The molecular diffusion coefficient for
51
momentum, πœˆπ‘š , is set at 1.20×10-6 m2 s-1, and that for heat, πœˆβ„Ž , is set at 1.34×10-7 m2
52
s-1, according to (Paulson and Simpson 1981). The thickness of the cool skin δ is
53
determined by (Saunders 1967) as:
πœ†πœˆπ‘š
54
𝛿=
55
where λ is a dimensionless constant and 𝑒∗ is the friction velocity of water (m/s). SIT
56
determines λ according to (Artale et al. 2002). Below the cool skin and the viscous
57
layer, eddy diffusivity is determined according to a TKE-mixing length approach
58
(Gaspar et al. 1990).
𝑒∗
(3)
59
To correct bulk SST computed using conventional discretization to skin SST we
60
introduce a surface effective thickness (β„Žπ‘’ ) (Tu 2006; Tu and Tsuang 2014). The
61
effective thickness is a function of the surface layer, which is added to the top of the
62
uppermost numerical layer of the conventional discretization (Fig. A1). This surface
63
layer has a physical thickness d of 0.25 h1, i.e., d = 0.25 h1. The net heat flux absorbed
64
within the surface layer is G0+Rsn[F(z0)-F(z0-d)]+G0,1. To determine the upper skin
65
temperature T0 (not the column-mean temperature) of the surface layer, T0 is
66
parameterized as:
67
 w cw he
68
Where he is the effective thickness (m) for heat of the surface layer. Note that the first
69
term on the right-hand side of the above equation is a cooling term since the non-solar
70
surface heat flux G0 is usually upward (Saunders 1967; Dalu and Purini 1982;
71
Soloviev and Schlüssel 1994; Fairall et al. 1996). The second term is a heating term
ο‚ΆT0
ο€½ G0  Rsn F z0  ο€­ F z0 ο€­ d   G0,1
ο‚Άt
(4)
72
due to the absorption of shortwave solar radiation (Fairall et al. 1996). The third term
73
is the vertical heat flux due to molecular (if within the skin layer) or eddy (if below
74
the skin layer) diffusivity. Eq. (4) is proposed by this study to calculate the skin
75
temperature T0. The effective thickness is a function of heat conductivity. It is derived
76
analytically to reproduce the diurnal fluctuation of skin temperature if the fluctuation
77
can be described as a cosine function in time (Tsuang et al. 2009). The effective
78
thickness is derived to be:
79


οƒΆ 
οƒΆοƒΉ  
οƒΆ 
οƒΆοƒΉ

οƒ· 
οƒ·οƒΊ οƒͺ 
οƒ· 
οƒ·οƒΊ
οƒͺ
k οƒͺ
d οƒ·  d οƒ·οƒΊ οƒͺ 
d οƒ·  d οƒ·οƒΊ

he ο‚Ί
1 ο€­ exp ο€­
cos
 exp ο€­
sin

 οƒͺ
2k οƒ·  2k οƒ· οƒΊ οƒͺ 
2k οƒ·  2k οƒ· οƒΊ

οƒ· 
οƒ·οƒΊ οƒͺ 
οƒ· 
οƒ·οƒΊ
οƒͺ
 οƒΈ     
 οƒΈ   


80
where  is the angular velocity of the earth with respect to the sun (=2/86400 s-1).
81
Note that the upper temperature T0 of the skin layer is the so-called sea surface
82
temperature (SST). Once SST is determined, we can determine heat fluxes between
83
the atmosphere and ocean for the next model timestep. Then, the error due to incorrect
84
usage of T1 for T0 to determine heat exchange between the atmosphere and ocean in
85
the conventional approach is corrected.
2
2
(5)
86
Overall, SIT simulates the SST and upper ocean temperature variations, including
87
the cool-skin and warm-layer mechanisms. In the finest resolution experiments, SIT
88
has 42 vertical layers, and with 12 in the upper 10m. Simulated water temperatures
89
are at surface, and grid cells with center at depth of 0.05mm, 1 m, 2 m, 3 m, 4 m, 5 m,
90
6 m, 7 m, 8 m, 9 m, 10 m, 16.8 m, 29.5 m, 43.6 m, 59.3 m, 76.9 m, 96.8 m, 119.4 m,
91
145.3 m, 174.9 m, 208.9 m, 248.3 m, 293.8 m, 346.8 m, 408.4 m, 480.2 m, 564.3 m,
92
662.6 m, 779.7 m, 913.1 m, 1072 m, 1258.8 m, 1478.6 m, 1737.3m, 2042m, 2401m,
93
2824.4m, 3323.6m, 3912.4m, 4607.1m and the ocean seabed. The resolution in the
94
upper 10 m is very fine in order to capture the upper ocean warm layer, and there is a
95
layer at 0.05 mm and resolving a corresponding effective thickness for reproducing
96
the cool skin of the ocean surface. For the C-17m we deleted layer from 0.05mm to 10
97
m and C-59m we deleted layer from 0.05mm to 43.6 m.
98
99
In addition, a nudging technique is used to correct the bias of calculated ocean
temperature and salinity (denoted as X) at layers deeper than 10 m depth as:
100
∗,𝑛+1
∗,𝑛+1
𝑛+1
𝑛+1
π‘‹π‘˜,𝑐
= π‘‹π‘˜,𝑐
+ 𝛽(π‘‹π‘˜,π‘œ
− π‘‹π‘˜,𝑐
)
101
∗,𝑛+1
𝑛+1
where π‘‹π‘˜,𝑐
is calculated X at depth k at timestep n+1; π‘‹π‘˜,𝑐
is calculated π‘‹π‘˜,𝑐 by
102
𝑛+1
Eq. (1) at timestep n+1; π‘‹π‘˜,π‘œ
is observed π‘‹π‘˜,𝑐 at timestep n+1; 𝛽 is a relaxation
103
factor. It should be within 0 and 1. When setting 𝛽 at 0, the calculated X is determined
104
by Eq. 1 only; when setting 𝛽 at 1, the calculated X is restored back to observed X
105
every time step. The relaxation factor is parameterized as:
106
𝛽 = 0.5 𝜏
(6)
βˆ†π‘‘
(7)
107
Where 𝜏 is the timescale for nudging; βˆ†π‘‘ is the time step. To account for neglected
108
horizontal processes, the ocean is weakly nudged with a 30-day time scale for depths
109
within 10-100 m (i.e., τ = 30 𝑑), and 1-day time scale for depths > 100 m (i.e., τ =
110
1 𝑑) to the observed climatological ocean temperature; there is no nudging within the
111
upper 10-m depth.
112
Reference
113
Artale V, Iudicone D, Santoleri R, Rupolo V, Marullo S, D'Ortenzio F (2002) Role of
114
surface fluxes in ocean general circulation models using satellite sea surface
115
temperature: Validation of and sensitivity to the forcing frequency of the
116
Mediterranean thermohaline circulation. J Geophys Res 107 (C8):29-21-29-24
117
Dalu G, Purini R (1982) The diurnal thermocline due to buoyant convection.
118
Quarterly Journal of the Royal Meteorological Society 108 (458):929-935
119
Fairall C, Bradley EF, Godfrey J, Wick G, Edson JB, Young G (1996) Cool-skin and
120
warm-layer effects on sea surface temperature. Journal of Geophysical
121
research 101 (C1):1295-1308
122
Gaspar P, Gregoris Y, Lefevre J-M (1990) A simple eddy kinetic energy model for
123
simulations of the oceanic vertical mixing: Tests at station Papa and long-term
124
upper ocean study site. J Geophys Res 95 (C9):16179-16193
125
Grassl H (1976) The dependence of the measured cool skin of the ocean on wind
126
stress and total heat flux. Boundary-Layer Meteorology 10 (4):465-474
127
Hasse L (1971) The sea surface temperature deviation and the heat flow at the sea-air
128
interface. Boundary-Layer Meteorology 1 (3):368-379
129
Lan Y-Y, Tsuang B-J, Tu C-Y, Wu T-Y, Chen Y-L, Hsieh C-I (2010) Observation
130
and simulation of meteorology and surface energy components over the South
131
China Sea in summers of 2004 and 2006. Terrestrial, Atmospheric and
132
Oceanic Sciences 21 (2):325-342
133
134
135
136
Paulson CA, Simpson JJ (1981) The temperature difference across the cool skin of the
ocean. J Geophys Res 86 (C11):11044-11054
Saunders PM (1967) The temperature at the ocean-air interface. Journal of the
Atmospheric Sciences 24 (3):269-273
137
Soloviev AV, Schlüssel P (1994) Parameterization of the cool skin of the ocean and of
138
the air-ocean gas transfer on the basis of modeling surface renewal. Journal of
139
Physical Oceanography 24 (6):1339-1346
140
141
Tsuang B-J, Tu C-Y, Arpe K (2001) Lake parameterization for climate models.
Max-Planck-Institute for Meteorology Rept 316:72pp
142
Tsuang B-J, Tu C-Y, Tsai J-L, Dracup JA, Arpe K, Meyers T (2009) A more accurate
143
scheme for calculating Earths-skin temperature. Climate Dynamics 32
144
(2-3):251-272
145
146
147
148
149
150
151
152
Tu C-Y (2006) Sea Surfce Temperature Simulation in Climate Model. National
Chung Hsing University,
Tu C-Y, Tsuang B-J (2005) Cool-skin simulation by a one-column ocean model.
Geophysical research letters 32 (22)
Tu C-Y, Tsuang B-J (2014) Numerical discretization for sea surface temperature
simulation in a turbulent kinetic energy ocean model.
Wu J (1985) On the cool skin of the ocean. Boundary-Layer Meteorology 31
(2):203-207
153
154
FigureA1. Left figure is the schema of the surface effective thickness β„Žπ‘’ of an ideal
155
surface with depth d. 𝑇0 is the skin temperature and 𝑇1 is the averaged temperature
156
of the uppermost layer of the water with a thickness of β„Ž1 . Note that the shaded area
157
denotes the energy stored from the surface to depth d, which is close to the
158
rectangular area. The mean temperature thus calculated from the rectangular area, the
159
open circle beneath 𝑇0 , is representative of the skin temperature. 𝐺𝑖,𝑗 denotes the
160
energy flux between layer i and j. Right figure is the schematic of the conventional
161
discretization with a thickness of β„Ž1 for the uppermost layer, representing only the
162
bulk SST.
163
Download