MIKE 321 FM Scientific Doc PDF

MIKE 21 & MIKE 3 Flow Model FM
Hydrodynamic and Transport Module

Scientific Documentation
MIKE 2017
DHI headquarters
Agern Allé 5
DK-2970 Hørsholm
Denmark
+45 4516 9200 Telephone
+45 4516 9333 Support
+45 4516 9292 Telefax
mike@dhigroup.com
www.mikepoweredbydhi.com
mike_321_fm_scientific_doc.docx / ORS/SIS / 2017-10-03 - © DHI

PLEASE NOTE
COPYRIGHT This document refers to proprietary computer software, which is

protected by copyright. All rights are reserved. Copying or other
reproduction of this manual or the related programmes is
prohibited without prior written consent of DHI. For details please
refer to your ‘DHI Software Licence Agreement’.
LIMITED LIABILITY The liability of DHI is limited as specified in Section III of your
‘DHI Software Licence Agreement’:
‘IN NO EVENT SHALL DHI OR ITS REPRESENTATIVES
(AGENTS AND SUPPLIERS) BE LIABLE FOR ANY DAMAGES
WHATSOEVER INCLUDING, WITHOUT LIMITATION,
SPECIAL, INDIRECT, INCIDENTAL OR CONSEQUENTIAL
DAMAGES OR DAMAGES FOR LOSS OF BUSINESS PROFITS
OR SAVINGS, BUSINESS INTERRUPTION, LOSS OF
BUSINESS INFORMATION OR OTHER PECUNIARY LOSS
ARISING OUT OF THE USE OF OR THE INABILITY TO USE
THIS DHI SOFTWARE PRODUCT, EVEN IF DHI HAS BEEN
ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. THIS
LIMITATION SHALL APPLY TO CLAIMS OF PERSONAL
INJURY TO THE EXTENT PERMITTED BY LAW. SOME
COUNTRIES OR STATES DO NOT ALLOW THE EXCLUSION
OR LIMITATION OF LIABILITY FOR CONSEQUENTIAL,
SPECIAL, INDIRECT, INCIDENTAL DAMAGES AND,
ACCORDINGLY, SOME PORTIONS OF THESE LIMITATIONS
MAY NOT APPLY TO YOU. BY YOUR OPENING OF THIS
SEALED PACKAGE OR INSTALLING OR USING THE
SOFTWARE, YOU HAVE ACCEPTED THAT THE ABOVE
LIMITATIONS OR THE MAXIMUM LEGALLY APPLICABLE
SUBSET OF THESE LIMITATIONS APPLY TO YOUR
PURCHASE OF THIS SOFTWARE.’
MIKE 2017
CONTENTS
MIKE 21 & MIKE 3 Flow Model FM
Hydrodynamic and Transport Module
Scientific Documentation
1 Introduction ...................................................................................................................... 1
2 Governing Equations ....................................................................................................... 2

2.1 3D Governing Equations in Cartesian Coordinates ............................................................................. 2
2.1.1 Shallow water equations ...................................................................................................................... 2
2.1.2 Transport equations for salt and temperature ...................................................................................... 4
2.1.3 Transport equation for a scalar quantity............................................................................................... 5
2.1.4 Turbulence model................................................................................................................................. 5
2.1.5 Governing equations in Cartesian and sigma coordinates .................................................................. 8
2.2 3D Governing Equations in Spherical and Sigma Coordinates ......................................................... 10
2.3 2D Governing Equations in Cartesian Coordinates ........................................................................... 12
2.3.1 Shallow water equations .................................................................................................................... 12
2.3.2 Transport equations for salt and temperature .................................................................................... 12
2.3.3 Transport equations for a scalar quantity ........................................................................................... 13
2.4 2D Governing Equations in Spherical Coordinates ............................................................................ 13
2.5 Bottom Stress ..................................................................................................................................... 14
2.6 Wind Stress ........................................................................................................................................ 15
2.7 Ice Coverage ...................................................................................................................................... 16
2.8 Tidal Potential ..................................................................................................................................... 17
2.9 Wave Radiation .................................................................................................................................. 18
2.10 Heat Exchange ................................................................................................................................... 18
2.10.1 Vaporisation ....................................................................................................................................... 19
2.10.2 Convection ......................................................................................................................................... 21
2.10.3 Short wave radiation........................................................................................................................... 21
2.10.4 Long wave radiation ........................................................................................................................... 24
3 Numerical Solution......................................................................................................... 26
3.1 Spatial Discretization .......................................................................................................................... 26
3.1.1 Vertical Mesh ...................................................................................................................................... 28
3.1.2 Shallow water equations .................................................................................................................... 31
3.1.3 Transport equations ........................................................................................................................... 34
3.2 Time Integration ................................................................................................................................. 35
3.3 Boundary Conditions .......................................................................................................................... 36
3.3.1 Closed boundaries ............................................................................................................................. 36
3.3.2 Open boundaries ................................................................................................................................ 37
3.3.3 Flooding and drying ............................................................................................................................ 37
4 Infiltration and Leakage ................................................................................................. 39

4.1 Net Infiltration Rates ........................................................................................................................... 39
4.2 Constant Infiltration with Capacity ...................................................................................................... 40
5 Jet Sources ..................................................................................................................... 42

5.1 Nearfield Calculations ........................................................................................................................ 42
i
5.2 End of Nearfield region....................................................................................................................... 44
5.3 Nearfield-Farfield model coupling ...................................................................................................... 46
6 Validation ........................................................................................................................ 48
6.1 Dam-break Flow through Sharp Bend................................................................................................ 48
6.1.1 Physical experiments ......................................................................................................................... 48
6.1.2 Numerical experiments....................................................................................................................... 49
6.1.3 Results ............................................................................................................................................... 50
6.2 Jet Source .......................................................................................................................................... 52
7 References ...................................................................................................................... 57
ii
Introduction
1 Introduction
This document presents the scientific background for the new MIKE 21 & MIKE 3 Flow
Model FM1 modelling system developed by DHI Water & Environment. The objective is to
provide the user with a detailed description of the flow and transport model equations,
numerical discretization and solution methods. Also, model validation is discussed in this
document.
The MIKE 21 & MIKE 3 Flow Model FM is based on a flexible mesh approach and it has
been developed for applications within oceanographic, coastal and estuarine
environments. The modelling system may also be applied for studies of overland flooding.
The system is based on the numerical solution of the two/three-dimensional

incompressible Reynolds averaged Navier-Stokes equations invoking the assumptions of
Boussinesq and of hydrostatic pressure. Thus, the model consists of continuity,
momentum, temperature, salinity and density equations and it is closed by a turbulent
closure scheme. For the 3D model the free surface is taken into account using a sigma
coordinate transformation approach.
The spatial discretization of the primitive equations is performed using a cell-centred finite
volume method. The spatial domain is discretized by subdivision of the continuum into
non-overlapping elements/cells. In the horizontal plane, an unstructured grid is used,
while in the vertical domain in the 3D model a structured mesh is used. In the 2D model,
the elements can be triangles or quadrilateral elements. In the 3D model, the elements
can be prisms or bricks, whose horizontal faces are triangles and quadrilateral elements,
respectively.
1 Including the MIKE 21 Flow Model FM (two-dimensional flow) and MIKE 3 Flow Model FM (three-
dimensional flow)
© DHI - MIKE 21 & MIKE 3 Flow Model FM - Hydrodynamic and Transport Module 1
Governing Equations
2 Governing Equations
2.1 3D Governing Equations in Cartesian Coordinates
2.1.1 Shallow water equations
The model is based on the solution of the three-dimensional incompressible Reynolds

averaged Navier-Stokes equations, subject to the assumptions of Boussinesq and of
hydrostatic pressure.
The local continuity equation is written as
u v w
  S (2.1)
x y z
and the two horizontal momentum equations for the x- and y-component, respectively
u u 2 vu wu  1 pa

    fv  g  
t x y z x  0 x
(2.2)
g   1  sxx sxy    u 
0  z
dz      Fu   t   us S
x  0h  x y  z  z 
v v 2 uv wv  1 pa

     fu  g  
t y x z y  0 y
(2.3)
g   1  s yx s yy    v 
0  z
dz      Fv   t   vs S
y  0h  x y  z  z 
where t is the time; x, y and z are the Cartesian coordinates;  is the surface elevation;
d is the still water depth; h    d is the total water depth; u, v and w are the velocity
components in the x, y and z direction; f  2 sin  is the Coriolis parameter (  is the
angular rate of revolution and  the geographic latitude); g is the gravitational
acceleration;  is the density of water; sxx , sxy , s yx and s yy are components of the
radiation stress tensor;  t is the vertical turbulent (or eddy) viscosity; p a is the
atmospheric pressure;  o is the reference density of water. S is the magnitude of the
discharge due to point sources and u s , vs  is the velocity by which the water is
discharged into the ambient water. The horizontal stress terms are described using a
gradient-stress relation, which is simplified to
  u     u v  
Fu  2A    A  
x  x  y   y x  
(2.4)
Governing Equations
   u v     v 
Fv  A  
x   y x   y  y 
2A (2.5)
where A is the horizontal eddy viscosity.
The surface and bottom boundary condition for u, v and w are
At z   :
    u v 
 
(2.6)
1
u v  w  0,  ,    sx , sy
t x y  z z   0 t
At z  d :
d d  u v 
 
(2.7)
1
u v  w  0,  ,    bx , by
x y  z z   0 t
where  sx , sy  and  bx , by  are the x and y components of the surface wind and
bottom stresses.
The total water depth, h, can be obtained from the kinematic boundary condition at the
surface, once the velocity field is known from the momentum and continuity equations.
However, a more robust equation is obtained by vertical integration of the local continuity
equation
h hu hv  
   hS  P  E (2.8)
t x y
 
where P and E are precipitation and evaporation rates, respectively, and u and v are
the depth-averaged velocities
 
hu   udz , hv   vdz (2.9)
d d
The fluid is assumed to be incompressible. Hence, the density,  , does not depend on
the pressure, but only on the temperature, T, and the salinity, s, via the equation of state
   (T , s) (2.10)
Here the UNESCO equation of state is used (see UNESCO, 1981).
Governing Equations
2.1.2 Transport equations for salt and temperature
The transports of temperature, T, and salinity, s, follow the general transport-diffusion

equations as
T uT vT wT   T  

    FT   Dv   H  Ts S (2.11)
t x y z z  z 
s us vs ws   s 

    Fs   Dv   s s S (2.12)
t x y z z  z 

where Dv is the vertical turbulent (eddy) diffusion coefficient. H is a source term due to
heat exchange with the atmosphere. T s and s s are the temperature and the salinity of
the source. F are the horizontal diffusion terms defined by
     
FT ,F s     
 Dh    Dh T , s  (2.13)
 x  x  y  y 
where D h is the horizontal diffusion coefficient. The diffusion coefficients can be related
to the eddy viscosity
A t
Dh  and Dv  (2.14)
T T
where  T is the Prandtl number. In many applications a constant Prandtl number can be
used (see Rodi (1984)).
The surface and bottom boundary conditions for the temperature are
At z   :
T Qn (2.15)
Dh   T p Pˆ  Te Eˆ
z  0 c p
At z  d :
(2.16)
T
0
z
where Q n is the surface net heat flux and c p  4217 J /( kg  K ) is the specific heat of

the water. A detailed description for determination of H and Q n is given in Section 2.10.
Governing Equations
The surface and bottom boundary conditions for the salinity are
At z   :
(2.17)
s
0
z
At z  d :
(2.18)
s
0
z
When heat exchange from the atmosphere is included, the evaporation is defined as
 q
  v qv  0
E    0lv (2.19)
0 qv  0

where q v is the latent heat flux and l v  2.5  10 6 is the latent heat of vaporisation of
water.
2.1.3 Transport equation for a scalar quantity
The conservation equation for a scalar quantity is given by
C uC vC wC   C 

    FC   Dv   k p C  Cs S (2.20)
t x y z z  z 
where C is the concentration of the scalar quantity, k p is the linear decay rate of the
scalar quantity, C s is the concentration of the scalar quantity at the source and Dv is the
vertical diffusion coefficient. FC is the horizontal diffusion term defined by
      
FC    Dh    Dh C (2.21)
 x  x  y  y 
where D h is the horizontal diffusion coefficient.
2.1.4 Turbulence model
The turbulence is modelled using an eddy viscosity concept. The eddy viscosity is often
described separately for the vertical and the horizontal transport. Here several turbulence
models can be applied: a constant viscosity, a vertically parabolic viscosity and a
standard k- model (Rodi, 1984). In many numerical simulations the small-scale
turbulence cannot be resolved with the chosen spatial resolution. This kind of turbulence
can be approximated using sub-grid scale models.
Governing Equations
Vertical eddy viscosity

The eddy viscosity derived from the log-law is calculated by
 zd zd
2
 t  U  h c1  c2    (2.22)
 h  h  
 
where U  max Us ,Ub  and c1 and c 2 are two constants. Us and U b are the
friction velocities associated with the surface and bottom stresses, c1  0.41 and
c 2  0.41 give the standard parabolic profile.
In applications with stratification the effects of buoyancy can be included explicitly. This is
done through the introduction of a Richardson number dependent damping of the eddy
viscosity coefficient, when a stable stratification occurs. The damping is a generalisation
of the Munk-Anderson formulation (Munk and Anderson, 1948)
 t   t* (1  aRi ) b (2.23)
where  t* is the undamped eddy viscosity and Ri is the local gradient Richardson number
1
g    u   v  
2 2
Ri       (2.24)
 0 z   z   z  
a  10 and b  0.5 are empirical constants.
In the k- model the eddy-viscosity is derived from turbulence parameters k and  as
k2
 t  c (2.25)

where k is the turbulent kinetic energy per unit mass (TKE),  is the dissipation of TKE
and c  is an empirical constant.
The turbulent kinetic energy, k, and the dissipation of TKE,  , are obtained from the
following transport equations
k uk vk wk    k 

    Fk   t   P  B 
z   k z 
(2.26)
t x y z
 u v w

   
t x y z
(2.27)
     
F   t   c1 P  c3 B  c2  
z    z  k
where the shear production, P, and the buoyancy production, B, are given as
Governing Equations
 xz u  yz v   u  2  v  2 
P   t       (2.28)
 0 z  0 z   z   z  
 
t 2
B N (2.29)
t
with the Brunt-Väisälä frequency, N, defined by
g 
N2   (2.30)
 0 z
t is the turbulent Prandtl number and  k ,   , c1 , c 2 and c3 are empirical
constants. F are the horizontal diffusion terms defined by
      
( Fk , F )    Dh    Dh (k ,  ) (2.31)
 x  x  y  y 
The horizontal diffusion coefficients are given by Dh  A /  k and Dh  A /   ,

respectively.
Several carefully calibrated empirical coefficients enter the k-e turbulence model. The
empirical constants are listed in (2.47) (see Rodi, 1984).
Table 2.1 Empirical constants in the k- model.
c c1 c 2 c3 t k 
0.09 1.44 1.92 0 0.9 1.0 1.3
At the surface the boundary conditions for the turbulent kinetic energy and its rate of
dissipation depend on the wind shear, Us
At z = :
1
k U2s
c (2.32)
U3s
 for U s  0
zb
k

k c 3/ 2
0 for Us  0 (2.33)

z ah
where  =0.4 is the von Kármán constant, a  0.07 is and empirical constant and z s
is the distance from the surface where the boundary condition is imposed. At the seabed
the boundary conditions are
Governing Equations
At z  d :
1 U3b (2.34)
k U 2b 
c zb
where z b is the distance from the bottom where the boundary condition is imposed.
Horizontal eddy viscosity

In many applications a constant eddy viscosity can be used for the horizontal eddy
viscosity. Alternatively, Smagorinsky (1963) proposed to express sub-grid scale
transports by an effective eddy viscosity related to a characteristic length scale. The
subgrid scale eddy viscosity is given by
A  cs2l 2 2Sij Sij (2.35)
where cs is a constant, l is a characteristic length and the deformation rate is given by
1  u u j 
S ij   i   (i, j  1,2) (2.36)
2  x j xi 

2.1.5 Governing equations in Cartesian and sigma coordinates
The equations are solved using a vertical -transformation
z  zb
 , x  x, y  y (2.37)
h
where  varies between 0 at the bottom and 1 at the surface. The coordinate
transformation implies relations such as
 1 
 (2.38)
z h 
      1  d h    1  d h   
 ,        ,        (2.39)
 x y   x h  x x   y  h  y y   
In this new coordinate system the governing equations are given as
h hu hv h

    hS (2.40)
t x  y  
Governing Equations
hu hu 2 hvu hu  h pa

    fvh  gh  
t x y  x  0 x
(2.41)
hg   1  s s     v u 
0  z x
dz   xx  xy   hFu 
 0  x y 

  h  
  hus S
hv huv hv 2 hv  h pa

     fuh  gh  
t x y   y   0 y 
(2.42)
hg   1  s s     v v 
0  z y
dz   yx  yy   hFv 
 0  x y 

  h  
  hvs S
hT huT hvT hT

   
t x y  
(2.43)
  Dv T 
hFT     hH  hTs S
  h  
hs hus hvs hs   Dv s 

    hFs     hs s S (2.44)
t x y     h  
hk huk hvk hk

   
t x y  
(2.45)
1    t k 
hFk     h( P  B   )
h    k  
h hu hv h

   
t x y  
(2.46)
1    t   
hF     h  c1 P  c3 B  c2  
h       k
hC huC hvC hC   Dv C 

    hFC     hk p C  hC s S (2.47)
t x  y     h  
The modified vertical velocity is defined by
1 d d  h h h 
  w  u v     u v 
y 
(2.48)
h x y  t x
The modified vertical velocity is the velocity across a level of constant

. The horizontal diffusion terms are defined as
 u     u v  
hFu   2hA    hA    (2.49)
x  x  y   y x  
   u v     v 
hFv   hA      2hA 
 
x   y x   y  y 
(2.50)
Governing Equations
h( FT , Fs , Fk , F , Fc ) 
        (2.51)
  hDh    hDh   (T , s, k ,  , C )
 x  x  y  y  
The boundary condition at the free surface and at the bottom are given as follows
At  =1:
 u v 
 sx , sy 
h (2.52)
  0,  ,  
     0 t
At  =0:
 u v 
 
h (2.53)
  0,  ,   bx , by
     0 t
The equation for determination of the water depth is not changed by the coordinate
transformation. Hence, it is identical to Eq. (2.6).
2.2 3D Governing Equations in Spherical and Sigma Coordinates

In spherical coordinates the independent variables are the longitude,  , and the latitude,
 . The horizontal velocity field (u,v) is defined by
d d
u  R cos  vR (2.54)
dt dt
where R is the radius of the earth.
In this coordinate system the governing equations are given as (all superscripts indicating
the horizontal coordinate in the new coordinate system are dropped in the following for
notational convenience)
h 1  hu hv cos   h

    hS
t R cos      
(2.55)
hu 1  hu 2 hvu cos   hu  u 

      f  tan   vh 
t R cos        R 
1   1 pa g   1  s s  
 gh  
R cos     0   0  z 
dz   xx  cos  xy   
 0     
(2.56)
   v u 
hFu     hus S
  h  
Governing Equations
hv 1  huv hv 2 cos   hv  u 

       f  tan   uh 
t R cos        R 
1   1 pa g   1  1 s yx s yy  
 gh  
R    0   0  z 
dz  
 0  cos  
 
  
(2.57)
   v v 
hFv     hvs S
  h  
hT 1  huT hvT cos   hT

     
t R cos     
(2.58)
  Dv T 
hFT     hH  hTs S
  h  
hs 1  hus hvs cos   h s

     
t R cos     
(2.59)
  Dv s 
hFs     hss S
  h  
hk 1  huk hvk cos   hk

     
t R cos      
(2.60)
1    t k 
hFk     h( P  B   )
h    k  
h 1  hu hv cos   h

     
t R cos     
(2.61)
1    t   
hF     h  c1 P  c3 B  c2  
h       k
hC 1  huC hvC cos   hC

     
t R cos     
(2.62)
  Dv C 
hFC     hk pC  hCs S
  h  
The modified vertical velocity in spherical coordinates is defined by
1 u d v d  h u h v h 
 w        
h R cos   R y  t R cos   R   (2.63)
The equation determining the water depth in spherical coordinates is given as
h 1  hu hv cos  

     hS
t R cos   
(2.64)
 
Governing Equations
2.3 2D Governing Equations in Cartesian Coordinates
Integration of the horizontal momentum equations and the continuity equation over depth
h    d the following two-dimensional shallow water equations are obtained
h hu hv
   hS (2.65)
t x y
hu hu 2 hvu  h pa

   fvh  gh  
t x y x  0 x
gh 2   sx  bx 1  sxx sxy 
    
2  0 x  0  0  0  x
(2.66)
y 
 
 hTxx    hTxy   hus S
x y
hv huv hv 2  h pa

    fuh  gh  
t x y y  0 y
gh 2   sy  by 1  s yx s yy 
    
2  0 y  0  0  0  x
(2.67)
y 
 
x
 hTxy    hTyy   hvs S
y
The overbar indicates a depth average value. For example, u and v are the depth-
averaged velocities defined by
 
hu   udz , hv   vdz (2.68)
d d
The lateral stresses Tij include viscous friction, turbulent friction and differential
advection. They are estimated using an eddy viscosity formulation based on of the depth
average velocity gradients
u  u v  v
Txx  2 A , Txy  A  , T yy  2 A (2.69)
x  y x  y
2.3.2 Transport equations for salt and temperature
Integrating the transport equations for salt and temperature over depth the following two-
dimensional transport equations are obtained
hT hu T hv T 

   hFT  hH  hTs S (2.70)
t x y
Governing Equations
hs hu s hv s

   hFs  hs s S (2.71)
t x y
where T and s is the depth average temperature and salinity.
2.3.3 Transport equations for a scalar quantity
Integrating the transport equations for a scalar quantity over depth the following two-
dimensional transport equations are obtained
hC hu C hv C

   hFC  hk p C  hC s S (2.72)
t x y
where C is the depth average scalar quantity.
2.4 2D Governing Equations in Spherical Coordinates

In spherical coordinates the independent variables are the longitude,  ,and the latitude,
 . The horizontal velocity field (u,v) is defined by
d d
u  R cos  vR (2.73)
dt dt
where R is the radius of the earth.
In spherical coordinates the governing equation can be written
h 1  hu hv cos  

     0
t R cos   
(2.74)
 
hu 1  hu 2 hvu cos    u 

      f  tan   vh
t R cos       R 
1   h pa gh 2  1  sxx s  
  gh      cos  xy    (2.75)
R cos     0  2  0   0     
 sx  bx  
   hTxx    hTxy   hus S
 0  0 x y
hv 1  huv hv 2 cos    u 

       f  tan   uh
t R cos       R 
1   h pa gh 2  1  1 s yx s yy  
  gh     
R    0  2  0   0  cos     
(2.76)
 sy  by  
   hTxy    hTyy   hvs S
 0  0 x y
Governing Equations
hT 1  hu T hv T cos   

     hFT  hH  hTs S (2.77)
t R cos     
hs 1  hu s hv s cos  

     hFs  hs s S
R cos   
(2.78)
t  
hC 1  hu C hv C cos  

     hFC  hk p C  hCs S
R cos   
(2.79)
t  
2.5 Bottom Stress


The bottom stress,  b  ( bx , by ) , is determined by a quadratic friction law

b  
 c f ub ub (2.80)
0

where c f is the drag coefficient and ub  (ub , vb ) is the flow velocity above the bottom.
The friction velocity associated with the bottom stress is given by
2
U  b  c f ub (2.81)

For two-dimensional calculations u b is the depth-average velocity and the drag coefficient
can be determined from the Chezy number, C , or the Manning number, M
g
cf  (2.82)
C2
g
cf 
Mh  1/ 6 2
(2.83)

For three-dimensional calculations u b is the velocity at a distance z b above the sea
bed and the drag coefficient is determined by assuming a logarithmic profile between the
seabed and a point z b above the seabed
1
cf  2
 1  z b   (2.84)
 ln  
   z 
  0 
where  =0.4 is the von Kármán constant and z 0 is the bed roughness length scale.
When the boundary surface is rough, z 0 , depends on the roughness height, k s
z 0  mk s (2.85)
Governing Equations
where m is approximately 1/30.
Note, that the Manning number can be estimated from the bed roughness length using
the following
25.4
M (2.86)
k s1 / 6
The wave induced bed resistance can be determined from
2
 u fc 
c f    (2.87)
 ub 
where Ufc is the friction velocity calculated by considering the conditions in the wave
boundary layer. For a detailed description of the wave induced bed resistance, see
Fredsøe (1984) and Jones et.al. (2014).
2.6 Wind Stress


In areas not covered by ice the surface stress,  s  ( sx , sy ) , is determined by the winds
above the surface. The stress is given by the following empirical relation
 s   a cd u w u w (2.88)

where  a is the density of air, c d is the drag coefficient of air, and u w  (u w , vw ) is the
wind speed 10 m above the sea surface. The friction velocity associated with the surface
stress is given by
a c f uw
2
Us  (2.89)
0
The drag coefficient can either be a constant value or depend on the wind speed. The
empirical formula proposed by Wu (1980, 1994) is used for the parameterisation of the
drag coefficient.
c a w10  wa

 c  ca
c f  c a  b w10  wa  wa  w10  wb (2.90)
 wb  wa
cb w10  wb
where ca, cb, wa and wb are empirical factors and w10 is the wind velocity 10 m above the
sea surface. The default values for the empirical factors are ca = 1.255·10-3, cb =
2.425·10-3, wa = 7 m/s and wb = 25 m/s. These give generally good results for open sea
applications. Field measurements of the drag coefficient collected over lakes indicate that
the drag coefficient is larger than open ocean data. For a detailed description of the drag
coefficient see Geernaert and Plant (1990).
Governing Equations
2.7 Ice Coverage

It is possible to take into account the effects of ice coverage on the flow field.
In areas where the sea is covered by ice the wind stress is excluded. Instead, the surface

stress is caused by the ice roughness. The surface stress,  s  ( sx ,  sy ) , is determined
by a quadratic friction law

s  
 c f us us (2.91)
0

where c f is the drag coefficient and u s  (u s , v s ) is the flow velocity below the surface.
The friction velocity associated with the surface stress is given by
2
U s  c f u s (2.92)

For two-dimensional calculations u s is the depth-average velocity and the drag coefficient
can be determined from the Manning number, M
g
cf 
Mh  1/ 6 2
(2.93)
The Manning number is estimated from the bed roughness length using the following
25.4
M (2.94)
k s1 / 6

For three-dimensional calculations u s is the velocity at a distance z s below the surface
and the drag coefficient is determined by assuming a logarithmic profile between the
surface and a point z b below the surface
1
cf  2
 1  z s   (2.95)
 ln  
   z 
  0 
where  =0.4 is the von Kármán constant and z 0 is the bed roughness length scale.
When the boundary surface is rough, z 0 , depends on the roughness height, k s
z 0  mk s (2.96)
where m is approximately 1/30.
If ice thickness is specified, the water level is supressed by ice / water of the ice
thickness, where ice = 971 kg/m3 and water is the actual density of the water.
Governing Equations
2.8 Tidal Potential

The tidal potential is a force, generated by the variations in gravity due to the relative
motion of the earth, the moon and the sun that act throughout the computational domain.
The forcing is expanded in frequency space and the potential considered as the sum of a
number of terms each representing different tidal constituents. The forcing is
implemented as a so-called equilibrium tide, which can be seen as the elevation that
theoretically would occur, provided the earth was covered with water. The forcing enters
the momentum equations (e.g. (2.66) or (2.75)) as an additional term representing the
gradient of the equilibrium tidal elevations, such that the elevation  can be seen as the
sum of the actual elevation and the equilibrium tidal potential.
  ACTUAL T (2.97)
The equilibrium tidal potential T is given as
t
T   ei H i f i Li cos(2  bi  i0 x ) (2.98)
i Ti
where T is the equilibrium tidal potential, i refers to constituent number (note that the
constituents here are numbered sequentially), ei is a correction for earth tides based on
Love numbers, Hi is the amplitude, fi is a nodal factor, Li is given below, t is time, Ti is the
period of the constituent, bi is the phase and x is the longitude of the actual position.
The phase b is based on the motion of the moon and the sun relative to the earth and can
be given by
bi  (i1  i0 )s  (i2  i0 )h  i3 p  i4 N  i5 ps  ui sin( N ) (2.99)
where i0 is the species, i1 to i5 are Doodson numbers, u is a nodal modulation factor (see
Table 2.3) and the astronomical arguments s, h, p, N and ps are given in Table 2.2.
Table 2.2 Astronomical arguments (Pugh, 1987)
Mean longitude of the moon s 277.02+481267.89T+0.0011T2

Mean longitude of the sun h 280.19+36000.77T+0.0003T 2
Longitude of lunar perigee p 334.39+4069.04T-0.0103T2
Longitude of lunar ascending node N 259.16-1934.14T+0.0021T2
Longitude of perihelion ps 281.22+1.72T+0.0005T2
In Table 2.2 the time, T, is in Julian century from January 1 1900 UTC, thus T = (365(y –
1900) + (d – 1) + i)/36525 and i = int (y-1901)/4), y is year and d is day number
L depends on species number i0 and latitude y as
i0 = 0 L  3sin 2 ( y )  1
i0 = 1 L  sin(2 y)
i0 = 2 L  cos2 ( y)
Governing Equations
The nodal factor fi represents modulations to the harmonic analysis and can for some
constituents be given as shown in Table 2.3.
Table 2.3 Nodal modulation terms (Pugh, 1987)
fi ui
Mm 1.000 - 0.130 cos(N) 0
Mf 1.043 + 0.414 cos(N) -23.7 sin(N)
Q1, O1 1.009 + 0.187 cos(N) 10.8 sin(N)
K1 1.006 + 0.115 cos(N) -8.9 sin(N)
2N2, 2, 2, N2, M2 1.000 - 0.037 cos(N) -2.1 sin(N)
K2 1.024 + 0.286 cos(N) -17.7 sin(N)
2.9 Wave Radiation

The second order stresses due to breaking of short period waves can be included in the
simulation. The radiation stresses act as driving forces for the mean flow and can be
used to calculate wave induced flow. For 3D simulations a simple approach is used. Here
a uniform variation is used for the vertical variation in radiation stress.
2.10 Heat Exchange

The heat exchange with the atmosphere is calculated on basis of the four physical
processes
• Latent heat flux (or the heat loss due to vaporisation)

• Sensible heat flux (or the heat flux due to convection)
• Net short wave radiation
• Net long wave radiation
Latent and sensible heat fluxes and long-wave radiation are assumed to occur at the
surface. The absorption profile for the short-wave flux is approximated using Beer’s law.
The attenuation of the light intensity is described through the modified Beer's law as
I (d )  1   I 0 e d (2.100)
where I (d ) is the intensity at depth d below the surface; I 0 is the intensity just below
the water surface;  is a quantity that takes into account that a fraction of light energy
(the infrared) is absorbed near the surface;  is the light extinction coefficient. Typical
values for  and  are 0.2-0.6 and 0.5-1.4 m-1, respectively.  and  are user-
specified constants. The default values are   0.3 and   1.0 m 1 . The fraction of the
light energy that is absorbed near the surface is I 0 . The net short-wave radiation,
q sr ,net , is attenuated as described by the modified Beer's law. Hence the surface net
heat flux is given by
Governing Equations
Q n  qv  qc  qsr ,net  qlr ,net (2.101)

For three-dimensional calculations the source term H is given by
  q 1   e  (  z )  qsr,net 1    e  (  z )
H   sr ,net  (2.102)
z  0c p 
 0c p

For two-dimensional calculations the source term H is given by
 qv  qc  q sr ,net  qlr ,net

H (2.103)
0c p
The calculation of the latent heat flux, sensible heat flux, net short wave radiation, and net
long wave radiation as described in the following sections.
In areas covered by ice the heat exchange is excluded.
2.10.1 Vaporisation
Dalton’s law yields the following relationship for the vaporative heat loss (or latent flux),
see Sahlberg, 1984
qv  LCe (a1  b1W2m )Qwater  Qair  (2.104)
where L  2.5  106 J / kg is the latent heat vaporisation (in the literature
L  2.5  106  2300 Twater is commonly used); Ce  1.32  10 3 is the moisture transfer
coefficient (or Dalton number); W 2 m is the wind speed 2 m above the sea surface; Q water
Qair is the water vapour density in the
is the water vapour density close to the surface;
atmosphere; a1 and b1 are user specified constants. The default values are a1  0.5
and b1  0.9 .
Measurements of Q water and Qair are not directly available but the vapour density can
be related to the vapour pressure as
0.2167
Qi  ei (2.105)
Ti  Tk
in which subscript i refers to both water and air. The vapour pressure close to the sea,
e water , can be expressed in terms of the water temperature assuming that the air close to
the surface is saturated and has the same temperature as the water
1 1 
ewater  6.11e K    (2.106)
 Tk Twater  Tk 
Governing Equations
where K  5418 K and TK  273.15  K is the temperature at 0 C. Similarly the

vapour pressure of the air, eair , can be expressed in terms of the air temperature and the
relative humidity, R
1 1 
eair  R  6.11e K    (2.107)
 Tk Tair  Tk 
Replacing Q water and Qair with these expressions the latent heat can be written as
qv   Pv  a1  bW
1 2m  
  1 1   1 1  
 exp  K     R  exp  K    
   Tk Twater  Tk      Tk Tair  Tk    (2.108)
 Twater  Tk Tair  Tk 
 
 
 
where all constants have been included in a new latent constant Pv  4370 J  K / m 3 .
During cooling of the surface the latent heat loss has a major effect with typical values up
to 100 W/m2.
The wind speed, W 2, 2 m above the sea surface is calculated from the from the wind
speed, W 10, 10 m above the sea surface using the following formula:
Assuming a logarithmic profile the wind speed, u(z), at a distance z above the sea
surface is given by
u*  z
u( z)  log   (2.109)
  zo 
where u* is the wind friction velocity, z0 is the sea roughness and  =0.4 is von
Karman's constant. u* and z0 are given by
z0  zCharnocku*2 / g (2.110)
u ( z )
u* 
 z (2.111)
log  
 z0 
where z Charnock is the Charnock parameter. The default value is zCharnock  0.014. The
wind speed, W2, 2 m above the sea surface is then calculated from the from the wind
speed, W10, 10m above the sea surface by first solving Eq. (2.114) and Eq. (2.115)
iteratively for z0 with z=10m and u(z)=W10. Then W2 is given by
Governing Equations
2
log  
W2  W10  zo  W10  0.5m / s
 10  (2.112)
log  
 z0 
W2  W10 W10  0.5m / s
The heat loss due to vaporization occurs both by wind driven forced convection by and
free convection. The effect of free convection is taken into account by the parameter a1 in
Eq. (2.104). The free convection is also taken into account by introducing a critical wind
speed Wcritical so that the wind speed used in Eq. (2.112) is obtained as
W10=max(W10,Wcritical) . The default value for the critical wind speed is 2 m/s.
2.10.2 Convection
The sensible heat flux, qc (W / m 2 ) , (or the heat flux due to convection) depends on the
type of boundary layer between the sea surface and the atmosphere. Generally this
boundary layer is turbulent implying the following relationship
  air cair cheatingW10 (Tair  Twater ) Tair  T


qc   (2.113)
  air cair ccoolingW10 (Tair  Twater ) Tair  T

where  air is the air density 1.225 kg/m3; cair  1007 J /( kg  K ) is the specific heat
of air; cheating  0.0011 and ccooling  0.0011 , respectively, is the sensible transfer
coefficient (or Stanton number) for heating and cooling (see Kantha and Clayson, 2000);
W10 is the wind speed 10 m above the sea surface; Twater is the temperature at the sea
surface; Tair is the temperature of the air.
The convective heat flux typically varies between 0 and 100 W/m2.
The heat loss due to convection occurs both by wind driven forced convection by and free
convection. The free convection is taken into account by introducing a critical wind speed
Wcritical so that the wind speed used in Eq. (2.113) is obtained as W10=max(W10,Wcritical) .
The default value for the critical wind speed is 2 m/s.
2.10.3 Short wave radiation
Radiation from the sun consists of electromagnetic waves with wave lengths varying from
1,000 to 30,000 Å. Most of this is absorbed in the ozone layer, leaving only a fraction of
the energy to reach the surface of the Earth. Furthermore, the spectrum changes when
sunrays pass through the atmosphere. Most of the infrared and ultraviolet compound is
absorbed such that the solar radiation on the Earth mainly consists of light with wave
lengths between 4,000 and 9,000 Å. This radiation is normally termed short wave
radiation. The intensity depends on the distance to the sun, declination angle and
latitude, extraterrestrial radiation and the cloudiness and amount of water vapour in the
atmosphere (see Iqbal, 1983)
Governing Equations
The eccentricity in the solar orbit, E 0 , is given by
2
r 
E0   0   1.000110  0.034221cos()  0.001280 sin( )
r (2.114)
 0.000719 cos(2)  0.000077 sin( 2)
where r0 is the mean distance to the sun, r is the actual distance and the day angle
 (rad ) is defined by
2 (d n  1)
 (2.115)
365
and d n is the Julian day of the year.
The daily rotation of the Earth around the polar axes contributes to changes in the solar
radiation. The seasonal radiation is governed by the declination angle,  (rad ) , which
can be expressed by
  0.006918  0.399912 cos( )  0.07257sin( ) 

0.006758cos(2)  0.000907sin(2)  (2.116)
0.002697 cos(3)  0.00148sin(3)
The day length, n d , varies with  . For a given latitude,  , (positive on the northern
hemisphere) the day length is given by
arccos  tan( ) tan(  ) 

24
nd  (2.117)

and the sunrise angle,  sr (rad ) , and the sunset angle  ss (rad ) are
sr  arccos tan( ) tan( ) and ss  sr (2.118)
The intensity of short wave radiation on the surface parallel to the surface of the Earth
changes with the angle of incidence. The highest intensity is in zenith and the lowest
during sunrise and sunset. Integrated over one day the extraterrestrial intensity,
H 0 ( MJ / m 2 / day) , in short wave radiation on the surface can be derived as
q sc E 0 cos   cos  sin  sr    sr cos  sr 

24
H0  (2.119)

where q sc  4.9212 ( MJ / m 2 / h) is the solar constant.
For determination of daily radiation under cloudy skies, H ( MJ / m 2 / day) , the following
relation is used
Governing Equations
H n
 a 2  b2 (2.120)
H0 nd
in whichn is the number of sunshine hours and n d is the maximum number of sunshine
hours. a2 and b2 are user specified constants. The default values are a 2  0.295 and
b2  0.371 . The user-specified clearness coefficient corresponds to n / nd . Thus the
solar radiation, qs (W / m2 ) , can be expressed as
 H  106
qs    q0  a3  b3 cos i   (2.121)
 H0  3600
where
 
a3  0.4090  0.5016 sin   sr   (2.122)
 3
 
b3  0.6609  0.4767 sin   sr   (2.123)
 3
The extraterrestrial intensity, q0 (MJ / m 2 / h) and the hour angle  i is given by
 
q 0  q sc E 0  sin  sin    cos  cos  cos i 
24
(2.124)
  
  E 
12  t displaceme nt  LS  L E   t  t local 
4
i  (2.125)
12  60 60 
t displacem ent is the displacement hours due to summer time and the time meridian LS is
the standard longitude for the time zone. t displacem ent and LS are user specified
e  0 (h) and L S  0 (deg) . L E is the
constants. The default values are t displacem nt
local longitude in degrees. Et (s) is the discrepancy in time due to solar orbit and is
varying during the year. It is given by
 0.000075  0.001868 cos()  0.032077 sin( ) 

Et     229.18 (2.126)
  0.014615cos(2)  0.04089 sin( 2) 
Finally, t local is the local time in hours.
Solar radiation that impinges on the sea surface does not all penetrate the water surface.
Parts are reflected back and are lost unless they are backscattered from the surrounding
atmosphere. This reflection of solar energy is termed the albedo. The amount of energy,
which is lost due to albedo, depends on the angle of incidence and angle of refraction.
For a smooth sea the reflection can be expressed as
Governing Equations
1  sin 2 (i  r ) tan 2 (i  r ) 
   2  
2  sin (i  r ) tan 2 (i  r ) 
(2.127)
where i is the angle of incidence, r the refraction angle and  the reflection coefficient,
which typically varies from 5 to 40 %.  can be approximated using
 altitude
 5 0.48 altitude  5

 30  altitude
  0.48  0.05 5  altitude  30 (2.128)
 25
0.05 altitude  30


where the altitude in degrees is given by
 180 
altitude  90   arccos(sin ( ) sin(  )  cos( ) cos( ) cos( i ))  (2.129)
  
Thus the net short wave radiation, q s ,net (W / m 2 ) , can possibly be expressed as
qsr ,net  1    qs (2.130)
The net short wave radiation, qsr,net, can be calculated using empirical formulae as
described above. Alternatively, the net short wave radiation can be calculated using Eq.
(2.130) where the solar radiation, qs, is specified by the user or the net short wave
radiation, qsr,net, can be given by the user.
2.10.4 Long wave radiation
A body or a surface emits electromagnetic energy at all wavelengths of the spectrum.

The long wave radiation consists of waves with wavelengths between 9,000 and 25,000
Å. The radiation in this interval is termed infrared radiation and is emitted from the
atmosphere and the sea surface. The long wave emittance from the surface to the
atmosphere minus the long wave radiation from the atmosphere to the sea surface is
called the net long wave radiation and is dependent on the cloudiness, the air
temperature, the vapour pressure in the air and the relative humidity. The net outgoing
2
long wave radiation, qlr ,net (W / m ) , is given by Brunt’s equation (See Lind and
Falkenmark, 1972)
 
qlr ,net   sb Tair  TK 4 a  b ed  c  d n 
 (2.131)
 nd 
where e d is the vapour pressure at dew point temperature measured in mb; n is the
number of sunshine hours, n d is the maximum number of sunshine hours;
 sb  5.6697  10 8 W /( m 2  K 4 ) is Stefan Boltzman's constant; Tair (C ) is the air
temperature. The coefficients a, b, c and d are given as
Governing Equations
a  0.56; b  0.077mb½ ; c  0.10; d  .90 (2.132)
The vapour pressure is determined as
ed  10  R esaturated (2.133)
where R is the relative humidity and the saturated vapour pressure, esaturated (kPa) , with
100 % relative humidity in the interval from –51 to 52 C can be estimated by
esaturated  3.38639 
7.38 10 3
 Tair  0.8072   1.9  105 1.8  Tair  48  1.316  103
8
 (2.134)
The net long wave radiation, qlr,net, can be calculated using empirical formulae as
described above. Alternatively, the net long wave radiation can be calculated as
qlr ,net  qar ,net  qbr (2.135)
where the net incident atmospheric radiation, qar,net, is specified by the user and the back
radiation, qbr, is given by
qbr  (1  r ) sbTK4 (2.136)
where r=0.03 is the reflection coefficient and ε=0.985 is the emissivity factor of the
atmosphere. The net long wave radiation can also be specified by the user.
Numerical Solution
3 Numerical Solution
3.1 Spatial Discretization

The discretization in solution domain is performed using a finite volume method. The
spatial domain is discretized by subdivision of the continuum into non-overlapping
cells/elements.
In the two-dimensional case the elements can be arbitrarily shaped polygons, however,
here only triangles and quadrilateral elements are considered.
In the three-dimensional case a layered mesh is used: in the horizontal domain an

unstructured mesh is used while in the vertical domain a structured mesh is used (see
Figure 3.1). The vertical mesh is based on either sigma coordinates or combined sigma/z-
level coordinates. For the hybrid sigma/z-level mesh sigma coordinates are used from the
free surface to a specified depth and z-level coordinates are used below. The different
types of vertical mesh are illustrated in Figure 3.2. The elements in the sigma domain and
the z-level domain can be prisms with either a 3-sided or 4-sided polygonal base. Hence,
the horizontal faces are either triangles or quadrilateral element. The elements are
perfectly vertical and all layers have identical topology.
Figure 3.1 Principle of meshing for the three-dimensional case
Numerical Solution
Figure 3.2 Illustrations of the different vertical grids. Upper: sigma mesh, Lower: combined
sigma/z-level mesh with simple bathymetry adjustment. The red line shows the
interface between the z-level domain and the sigma-level domain
The most important advantage using sigma coordinates is their ability to accurately
represent the bathymetry and provide consistent resolution near the bed. However, sigma
coordinates can suffer from significant errors in the horizontal pressure gradients,
advection and mixing terms in areas with sharp topographic changes (steep slopes).
These errors can give rise to unrealistic flows.
The use of z-level coordinates allows a simple calculation of the horizontal pressure
gradients, advection and mixing terms, but the disadvantages are their inaccuracy in
representing the bathymetry and that the stair-step representation of the bathymetry can
result in unrealistic flow velocities near the bottom.
Numerical Solution
3.1.1 Vertical Mesh
For the vertical discretization both a standard sigma mesh and a combined sigma/z-level
mesh can be used. For the hybrid sigma/z-level mesh sigma coordinates are used from
the free surface to a specified depth, zσ, and z-level coordinates are used below. At least
one sigma layer is needed to allow changes in the surface elevation.
Sigma
In the sigma domain a constant number of layers, Nσ, are used and each sigma layer is a
fixed fraction of the total depth of the sigma layer, hσ, where ℎ𝜎 = 𝜂 − max⁡(𝑧𝑏 , 𝑧𝜎 ). The
discretization in the sigma domain is given by a number of discrete σ-levels {𝜎𝑖 ,⁡⁡⁡𝑖 =
1, (𝑁𝜎 + 1)}.⁡Here σ varies from 𝜎1 = 0 at the bottom interface of the lowest sigma layer
to 𝜎𝑁𝜎 +1 = 1 at the free surface.
Variable sigma coordinates can be obtained using a discrete formulation of the general
vertical coordinate (s-coordinate) system proposed by Song and Haidvogel (1994). First
an equidistant discretization in a s-coordinate system (-1≤ s ≤0) is defined
𝑁𝜎 + 1 − 𝑖
𝑠𝑖 = − ⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡𝑖 = 1, (𝑁𝜎 + 1) (3.1)
𝑁𝜎
The discrete sigma coordinates can then be determined by
 i  1   c si  1   c csi  i  1, N  1 (3.2)
where
1 𝜃
tanh (𝜃 (𝑠 + 2)) − tanh⁡(2 )
sinh(𝜃𝑠) (3.3)
𝑐(𝑠) = (1 − 𝑏) +𝑏
sinh(𝜃) 𝜃
2tanh⁡(2 )
Here σc is a weighting factor between the equidistant distribution and the stretch
distribution, θ is the surface control parameter and b is the bottom control parameter. The
range for the weighting factor is 0<σc≤1 where the value 1 corresponds to equidistant
distribution and 0 corresponds to stretched distribution. A small value of σc can result in
linear instability. The range of the surface control parameter is 0<θ≤20 and the range of
the bottom control parameter is 0≤b≤1. If θ<<1 and b=0 an equidistant vertical resolution
is obtained. By increasing the value of the θ, the highest resolution is achieved near the
surface. If θ>0 and b=1 a high resolution is obtained both near the surface and near the
bottom.
Examples of a mesh using variable vertical discretization are shown in Figure 3.3 and
Figure 3.4.
Numerical Solution
Figure 3.3 Example of vertical distribution using layer thickness distribution. Number of layers:
10, thickness of layers 1 to 10: .025, 0.075, 0.1, 0.01, 0.02, 0.02, 0.1, 0.1, 0.075,
0.025
Figure 3.4 Example of vertical distribution using variable distribution. Number of layers: 10, σc =
0.1, θ = 5, b = 1
Combined sigma/z-level
In the z-level domain the discretization is given by a number of discrete z-levels {𝑧𝑖 ,⁡⁡⁡𝑖 =
1, (𝑁𝑧 + 1)},⁡where Nz is the number of layers in the z-level domain. z1 is the minimum z-
level and 𝑧𝑁𝑧 +1 is the maximum z-level, which is equal to the sigma depth, zσ. The
corresponding layer thickness is given by
Δ𝑧𝑖 = 𝑧𝑖+1 − 𝑧𝑖 ⁡⁡⁡⁡⁡⁡⁡⁡⁡𝑖 = 1, 𝑁𝑧 (3.4)
The discretization is illustrated in Figure 3.5 and Figure 3.6.
Using standard z-level discretization the bottom depth is rounded to the nearest z-level.
Hence, for a cell in the horizontal mesh with the cell-averaged depth, zb, the cells in the
corresponding column in the z-domain are included if the following criteria is satisfied
(zi+1 − zi )/2 ≥ 𝑧𝑏 ⁡⁡⁡⁡𝑖 = 1, 𝑁𝑧 (3.5)
Numerical Solution
The cell-averaged depth, zb, is calculated as the mean value of the depth at the vortices
of each cell. For the standard z-level discretization the minimum depth is given by z1. Too
take into account the correct depth for the case where the bottom depth is below the
minimum z-level (𝑧1 > 𝑧𝑏 ) a bottom fitted approach is used. Here, a correction factor, f1,
for the layer thickness in the bottom cell is introduced. The correction factor is used in the
calculation of the volume and face integrals. The correction factor for the bottom cell is
calculated by
(𝑧2 − 𝑧𝑏 )
𝑓1 = (3.6)
∆𝑧1
The corrected layer thickness is given by ∆𝑧1∗ = 𝑓1 ∆𝑧1. The simple bathymetry
adjustment approach is illustrated in Figure 3.5.
For a more accurate representation of the bottom depth an advanced bathymetry

adjustment approach can be used. For a cell in the horizontal mesh with the cell-
averaged depth, zb, the cells in the corresponding column in the z-domain are included if
the following criteria is satisfied
zi+1 > 𝑧𝑏 ⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡𝑖 = 1, 𝑁𝑧 (3.7)
A correction factor, fi, is introduced for the layer thickness
(𝑧𝑖+1 − 𝑧𝑏 ) 𝑧𝑚𝑖𝑛
𝑓𝑖 = 𝑚𝑎𝑥 ( , )⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡𝑧𝑖 < 𝑧𝑏 < 𝑧𝑖+1 ⁡⁡𝑜𝑟⁡⁡𝑧1 > 𝑧𝑏
∆𝑧𝑖 ∆𝑧𝑖 (3.8)
𝑓𝑖 = 1⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡𝑧1 ≥ 𝑧𝑏
A minimum layer thickness, ∆𝑧𝑚𝑖𝑛 , is introduced to avoid very small values of the
correction factor. The correction factor is used in the calculation of the volume and face
integrals. The corrected layer thicknesses are given by {∆𝑧𝑖∗ = 𝑓𝑖 ∆𝑧𝑖 , 𝑖 = 1, 𝑁𝑧 }.⁡The
advanced bathymetry adjustment approach is illustrated in Figure 3.6.
Figure 3.5 Simple bathymetry adjustment approach
Numerical Solution
Figure 3.6 Advanced bathymetry adjustment approach
The integral form of the system of shallow water equations can in general form be written
U
   F (U )  S (U ) (3.9)
t
where U is the vector of conserved variables, F is the flux vector function and S is the
vector of source terms.
In Cartesian coordinates the system of 2D shallow water equations can be written
U   Fx  Fx    Fy  Fy 
I V I V
  S (3.10)
t x y
where the superscripts I and V denote the inviscid (convective) and viscous fluxes,
respectively and where
Numerical Solution
h 
U   hu  ,
 
 hv 
 
 
 hu  0 
    
1  u 
Fx I   hu 2  g (h 2  d 2 )  , FxV   hA  2  
 2    x  
 huv  
   u v  
 hA   
  y x  
 
  0 
 hv   
    u v  
Fy I   hvu  , Fy   hA  y  x  
V
 2 1     (3.11)
 hv  g ( h 2
 d 2
)    v  
 2 a  hA  2  
  x  
 
 
0 
 d h pa gh  1  s xx sxy  
2
 g  fvh     
 x  0 x 2  0 x  0  x y  
   
S  sx  bx  hus 
 0 0 
 
 g d  fuh  h pa  gh   1  s yx  s yy  
2
 y  0 y 2  0 y  0  x y  
 
  sy  by 
   hvs
 0 0 
 
In Cartesian coordinates the system of 3D shallow water equations can be written
U FxI Fy FI FxV Fy FV

I V
      S (3.12)
t x y  x y 
where the superscripts I and V denote the inviscid (convective) and viscous fluxes,
respectively and where
Numerical Solution
h 
U   hu  ,
 
 hv 
 
 
 hu  0 
  
1  u  
Fx I   hu 2  g (h 2  d 2 )  , FxV   hA  2  
 2    x  
 huv    u v  
 
 hA    
  y x  
 
0 
 hv   
    u v  
Fy   hvu
I
 , Fy
V
  hA    
 hv 2  1 g (h 2  d 2 )    y x  
 2    v   (3.13)
 hA  2  
  x  
 
0 
 h   
 u 
F   h u  , F V
I
 t
  h  
 h v   v 
 t 
h  
 
 
0 
 d h pa hg   1  s s  
S   g
 x
 fvh 
0
 
x  0 z x
dz   xx
 0  x
 xy   hus 
y  
 
 g d  fuh  h pa hg   1  s yx s 
 yy   hvs 
y   0  z y
 
 0  x
dz
 y 0 y  
Integrating Eq. (3.9) over the ith cell and using Gauss’s theorem to rewrite the flux
integral gives
U
 Ai t
d    ( F  n) ds   S (U )d 
i Ai
(3.14)
where Ai is the area/volume of the cell  is the integration variable defined on Ai , i

is the boundary of the ith cell and ds is the integration variable along the boundary. n is
the unit outward normal vector along the boundary. Evaluating the area/volume integrals
by a one-point quadrature rule, the quadrature point being the centroid of the cell, and
evaluating the boundary intergral using a mid-point quadrature rule, Eq. (3.14) can be
written
Numerical Solution
Ui 1 NS
t

Ai
 F  n 
j
j  Si (3.15)
Here U i and S i , respectively, are average values of U and S over the ith cell and stored
at the cell centre, NS is the number of sides of the cell, n j is the unit outward normal
vector at the jth side and  j the length/area of the jth interface.
Both a first order and a second order scheme can be applied for the spatial discretization.
For the 2D case an approximate Riemann solver (Roe’s scheme, see Roe, 1981) is used
to calculate the convective fluxes at the interface of the cells. Using the Roe’s scheme the
dependent variables to the left and to the right of an interface have to be estimated.
Second-order spatial accuracy is achieved by employing a linear gradient-reconstruction
technique. The average gradients are estimated using the approach by Jawahar and
Kamath, 2000. To avoid numerical oscillations a second order TVD slope limiter (Van
Leer limiter, see Hirch, 1990 and Darwish, 2003) is used.
For the 3D case an approximate Riemann solver (Roe’s scheme, see Roe, 1981) is used
to calculate the convective fluxes at the vertical interface of the cells (x’y’-plane). Using
the Roe’s scheme the dependent variables to the left and to the right of an interface have
to be estimated. Second-order spatial accuracy is achieved by employing a linear
gradient-reconstruction technique. The average gradients are estimated using the
approach by Jawahar and Kamath, 2000. To avoid numerical oscillations a second order
TVD slope limiter (Van Leer limiter, see Hirch, 1990 and Darwish, 2003) is used. The
convective fluxes at the horizontal interfaces (vertical line) are derived using first order
upwinding for the low order scheme. For the higher order scheme the fluxes are
approximated by the mean value of the fluxes calculated based on the cell values above
and below the interface for the higher order scheme.
3.1.3 Transport equations
The transport equations arise in the salt and temperature model, the turbulence model
and the generic transport model. They all share the form of Equation Eq. (2.20) in
Cartesian coordinates. For the 2D case the integral form of the transport equation can be
given by Eq. (3.9) where
U  hC
F I   huC , hvC 
 C C  (3.16)
F V   hDh , hDh
 x y 
S   hk pC  hCs S .
For the 3D case the integral form of the transport equation can be given by Eq. (3.9)
where
Numerical Solution
U  hC
F I  huC , hvC , hC 

 C C D C  (3.17)
F V   hDh  , hDh  , h h
 x y h  
S   hk pC  hCs S .
The discrete finite volume form of the transport equation is given by Eq. (3.15). As for the
shallow water equations both a first order and a second order scheme can be applied for
the spatial discretization.
In 2D the low order approximation uses simple first order upwinding, i.e., element
average values in the upwinding direction are used as values at the boundaries. The
higher order version approximates gradients to obtain second order accurate values at
the boundaries. Values in the upwinding direction are used. To provide stability and
minimize oscillatory effects, a TVD-MUSCL limiter is applied (see Hirch, 1990, and
Darwish, 2003).
In 3D the low order version uses simple first order upwinding. The higher order version
approximates horizontal gradients to obtain second order accurate values at the
horizontal boundaries. Values in the upwinding direction are used. To provide stability
and minimize oscillatory effects, an ENO (Essentially Non-Oscillatory) type procedure is
applied to limit the horizontal gradients. In the vertical direction a 3rd order ENO
procedure is used to obtain the vertical face values (Shu, 1997).
3.2 Time Integration

Consider the general form of the equations
U
 G U  (3.18)
t
For 2D simulations, there are two methods of time integration for both the shallow water
equations and the transport equations: A low order method and a higher order method.
The low order method is a first order explicit Euler method
Un1  Un  t G(Un ) (3.19)
where  t is the time step interval. The higher order method uses a second order Runge
Kutta method on the form:
U n  1  U n  12 t G (U n )
2
(3.20)
U n 1  U n  t G (U n  1 )
2
For 3D simulations the time integration is semi-implicit. The horizontal terms are treated
implicitly and the vertical terms are treated implicitly or partly explicitly and partly
implicitly. Consider the equations in the general semi-implicit form.
Numerical Solution
U
 Gh (U )  Gv ( BU )  Gh (U )  GvI (U )  GvV (U ) (3.21)
t
where the h and v subscripts refer to horizontal and vertical terms, respectively, and the
superscripts refer to invicid and viscous terms, respectively. As for 2D simulations, there
is a lower order and a higher order time integration method.
The low order method used for the 3D shallow water equations can written as
Un1  12 t Gv (Un 1 )  Gv (Un )   Un  t Gh (Un ) (3.22)
The horizontal terms are integrated using a first order explicit Euler method and the
vertical terms using a second order implicit trapezoidal rule. The higher order method can
be written
U n 1 2  14 t Gv (U n 1 2 )  Gv (U n )   U n  12 t Gh (U n )
(3.23)
U n 1  12 t  Gv (U n 1 )  Gv (U n )   U n  t Gh (U n 1 2 )
The horizontal terms are integrated using a second order Runge Kutta method and the
vertical terms using a second order implicit trapezoidal rule.
The low order method used for the 3D transport equation can written as
Un1  12 t  GvV (Un1 )  GvV (Un )   Un  t Gh (Un )  t GvI (Un ) (3.24)
The horizontal terms and the vertical convective terms are integrated using a first order
explicit Euler method and the vertical viscous terms are integrated using a second order
implicit trapezoidal rule. The higher order method can be written
U n 1 2  14 t GvV (U n 1 2 )  GvV (U n )  
U n  12 t Gh (U n )  12 t G vI (U n )
U n 1  12 t  GvV (U n 1 )  GvV (U n )  
(3.25)
U n  t Gh (U n 1 2 )  t GvI (U n 1/ 2 )
The horizontal terms and the vertical convective terms are integrated using a second
order Runge Kutta method and the vertical terms are integrated using a second order
implicit trapezoidal rule for the vertical terms.
3.3 Boundary Conditions
3.3.1 Closed boundaries
Along closed boundaries (land boundaries), normal fluxes are forced to zero for all
variables. For the momentum equations, this leads to full-slip along land boundaries. For
the shallow water equations, the no slip condition can also be applied where both the
normal and tangential velocity components are zero.
Numerical Solution
3.3.2 Open boundaries
For the shallow water equations a number of different boundary conditions can be applied
The flux, velocity and Flather boundary conditions are all imposed using a weak
approach. A ghost cell technique is applied where the primitive variables in the ghost cell
are specified. The water level is evaluated based on the value of the adjacent interior cell,
and the velocities are evaluated based on the boundary information. For a discharge
boundary, the transverse velocity is set to zero for inflow and passively advected for
outflow. The boundary flux is then calculated using an approximate Riemann solver.
The Flather (1976) condition is one of the most efficient open boundary conditions. It is
very efficient in connection with downscaling coarse model simulations to local areas (see
Oddo and Pinardi (2007)). The instabilities, which are often observed when imposing
stratified density at a water level boundary, can be avoided using Flather conditions
The level boundary is imposed using a strong approach based on the characteristic
theory (see e.g. Sleigh et al., 1998).
The discharge boundary condition is imposed using both a weak formulation using ghost
cell technique described above and a strong approach based on the characteristic theory
(see e.g. Sleigh et al., 1998).
Note that using the weak formulation for a discharge boundary the effective discharge
over the boundary may deviate from the specified discharge.
For transport equations, either a specified value or a zero gradient can be given. For
specified values, the boundary conditions are imposed by applying the specified
concentrations for calculation of the boundary flux. For a zero gradient condition, the
concentration at the boundary is assumed to be identical to the concentration at the
adjacent interior cell.
3.3.3 Flooding and drying
The approach for treatment of the moving boundaries problem (flooding and drying
fronts) is based on the work by Zhao et al. (1994) and Sleigh et al. (1998). When the
depths are small the problem is reformulated and only when the depths are very small the
elements/cells are removed from the calculation. The reformulation is made by setting the
momentum fluxes to zero and only taking the mass fluxes into consideration.
The depth in each element/cell is monitored and the elements are classified as dry,
partially dry or wet. Also the element faces are monitored to identify flooded boundaries.
• An element face is defined as flooded if the following two criteria are satisfied:
Firstly, the water depth at one side of face must be less than a tolerance depth, hdry
, and the water depth at the other side of the face larger than a tolerance depth,
h flood . Secondly, the sum of the still water depth at the side for which the water
depth is less than hdry and the surface elevation at the other side must be larger
than zero.
Numerical Solution
• An element is dry if the water depth is less than a tolerance depth, hdry , and no of
the element faces are flooded boundaries. The element is removed from the
calculation.
• An element is partially dry if the water depth is larger than hdry and less than a
tolerance depth, hwet , or when the depth is less than the hdry and one of the
element faces is a flooded boundary. The momentum fluxes are set to zero and only
the mass fluxes are calculated.
• An element is wet if the water depth is greater than hwet . Both the mass fluxes and
the momentum fluxes are calculated.
The wetting depth, hwet , must be larger than the drying depth, hdry , and flooding depth,
h flood , must satisfy
hdry  h flood  hwet (3.26)
The default values are hdry  0.005 m , h flood  0.05 m and hwet  0.1m .
Note, that for very small values of the tolerance depth, hwet , unrealistically high flow
velocities can occur in the simulation and give cause to stability problems.
Infiltration and Leakage
4 Infiltration and Leakage

The effect of infiltration and leakage at the surface zone may be important in cases of
flooding scenarios on otherwise dry land. It is possible to account for this in one of two
ways: by Net infiltration rates or by constant infiltration with capacity.
j-1 j j+1
Surface zone
Infiltration
Qi
Infiltration zone
Ql
Leakage
Figure 4.1 Illustration of infiltration process
4.1 Net Infiltration Rates

The net infiltration rate is defined directly. This will act as a simple sink in each element in
the overall domain area.
The one-dimensional vertical continuity equation is solved at each hydrodynamic time

step after the two-dimensional horizontal flow equations have been solved. The
calculation of the new water depth in the free surface zone for each horizontal element is
found by
𝐻(𝑗) = ⁡𝐻(𝑗) − ⁡ 𝑉𝑖𝑛𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑖𝑜𝑛 (𝑗) /𝐴(𝑗) (4.1)
Where 𝑉𝑖𝑛𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑖𝑜𝑛 (𝑗) is the infiltrated volume in element (𝑗) and A(j) the area of the
element.
If 𝐻(𝑗) becomes marked as dry then element (𝑗) will be taken out of the two-dimensional
horizontal flow calculations and no infiltration can occur until the element is flooded again.
In summary: when using Net infiltration rate an unsaturated zone is never specified and
thus has no capacity limits, so the specified infiltration rates will always be fully
effectuated as long as there is enough water available in the element.
4.2 Constant Infiltration with Capacity

Constant infiltration with capacity describes the infiltration from the free surface zone to
the unsaturated zone and from the unsaturated zone to the saturated zone by a simplified
model. The model assumes the following:
• The unsaturated zone is modelled as an infiltration zone with constant porosity over
the full depth of the zone.
• The flow between the free surface zone and the infiltration zone is based on a
constant flow rate, i.e. 𝑉𝑖𝑛𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑖𝑜𝑛 = 𝑄𝑖 ∙ ∆𝑡 where 𝑄𝑖 is the prescribed flow rate.
• The flow between the saturated and unsaturated zone is modelled as a leakage 𝑄𝑙
having a constant flow rate, i.e. 𝑉𝑙𝑒𝑎𝑘𝑎𝑔𝑒 = 𝑄𝑙 ∙ ∆𝑡.
The simplified model described above is solved through a one-dimensional continuity

equation. Feedback from the infiltration and leakage to the two-dimensional horizontal
hydrodynamic calculations is based solely on changes to the depth of the free surface
zone – the water depth.
Note that the infiltration flow cannot exceed the amount of water available in the free
surface water zone nor the difference between the water capacity of the infiltration zone
and the actual amount of water stored there. It is possible that the infiltration flow
completely drains the free surface zone from water and thus creates a dried-out point in
the two-dimensional horizontal flow calculations.
The one-dimensional vertical continuity equation is solved at each hydrodynamic time

step after the two-dimensional horizontal flow equations have been solved. The solution
proceeds in the following way:
1. Calculation of the volume from leakage flow in each horizontal element – 𝑉𝑙𝑒𝑎𝑘𝑎𝑔𝑒 (𝑗)
𝑉𝑙𝑒𝑎𝑘𝑎𝑔𝑒 (𝑗) = ⁡ 𝑄𝑙 (𝑗) ∙ ∆⁡𝑡 ∙ 𝐴(𝑗) (4.2)
𝑉𝑙𝑒𝑎𝑘𝑎𝑔𝑒 (𝑗) = min⁡(𝑉𝑙𝑒𝑎𝑘𝑎𝑔𝑒 (𝑗), 𝑉i(𝑗)) (4.3)
𝑉i(𝑗) ∶= ⁡𝑉i(𝑗) − 𝑉𝑙𝑒𝑎𝑘𝑎𝑔𝑒 (𝑗) (4.4)
Where 𝑉i(𝑗)⁡is the total amount of water in the infiltration zone and 𝑄𝑙 (𝑗) is the
leakage flow rate.
2. Calculation of the volume from infiltration flow in each horizontal element –

𝑉𝑖𝑛𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑖𝑜𝑛 (𝑗)
𝑉𝑖𝑛𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑖𝑜𝑛 (𝑗) = 𝑄𝑖 (𝑗) ∙ ∆⁡𝑡 ∙ 𝐴(𝑗) (4.5)
𝑉𝑖𝑛𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑖𝑜𝑛 (𝑗) = min (𝑉𝑖𝑛𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑖𝑜𝑛 (𝑗)⁡, 𝑆𝐶𝑖 (𝑗) − 𝑉i(𝑗), 𝐻(𝑗) ∙ 𝐴(𝑗) (4.6)
𝑉i(𝑗) ∶= ⁡𝑉i(𝑗) + ⁡ 𝑉𝑖𝑛𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑖𝑜𝑛 ⁡(𝑗) (4.7)
Where 𝑄𝑖 (𝑗) is the infiltration rate, 𝑆𝐶𝑖 (𝑗) is the water storage capacity and 𝐻(𝑗) the
depth of the free surface.
3. Calculation of the new water depth in the free surface zone for each horizontal
element
𝐻(𝑗) = ⁡𝐻(𝑗) − ⁡ 𝑉𝑖𝑛𝑓𝑖𝑙𝑡𝑟𝑎𝑡𝑖𝑜𝑛 (𝑗)/𝐴(𝑗) (4.8)
If 𝐻(𝑗) becomes marked as dry then element (j) will be taken out of the two-dimensional
horizontal flow calculations. The element can still leak but no infiltration can occur until
the element is flooded again.
The water storage capacity of the infiltration zone is calculated as
𝑆𝐶𝑖 (𝑗) = 𝑍𝑖 (𝑗) ∙ 𝐴(𝑗) ⁡ ∙ 𝛾(𝑗) (4.9)
Where 𝑍𝑖 (𝑗) is the depth of the infiltration zone and 𝛾(𝑗) is the porosity of the same zone.
In summary, when using Constant infiltration with capacity there can be situations where
the picture is altered and the rates are either only partially effectuated or not at all:
• If = ⁡𝐻(𝑗) < 𝐻𝑑𝑟𝑦 on the surface (dry surface) => infiltration rate is not effectuated
• If: the water volume in the infiltration zone reaches the full capacity => infiltration rate
is not effectuated
• If: the water volume is zero in the infiltration zone (the case in many initial conditions)
=> leakage rate is not effectuated
• Leakage volume must never eclipse the available water volume in the infiltration
zone, if so we utilise the available water volume in infiltration zone as leakage
volume
• Infiltration volume must never eclipse the available water volume on the surface, if
so we utilise the available water on the surface as infiltration volume
Jet Sources
5 Jet Sources
The simulation of jets/plumes is based on dynamic coupling of nearfield integrated jet
solution and the farfield hydrodynamic flow model (MIKE 3 Flow Model FM).
5.1 Nearfield Calculations

The near field solution is based on the integral jet model equations described by Jirka
(2004). It determines the steady state solution of the jet/plume by solving conservation
equations for flux and momentum, salinity and temperature (if included) under the given
ambient conditions.
The velocity profile and distribution of state parameters and scalar mass is assumed to
follow the Gaussian formulation. The jet model employs an entrainment closure approach
that distinguishes between the separate contributions of transverse shear and of
azimuthal shear mechanisms. It further contains a quadratic law turbulent drag force
mechanism (𝐹𝐷 ) as suggested by a number of recent detailed experimental investigations
on the dynamics of transverse jets into crossflow. The conservation principles for volume
(continuity), momentum components in the global directions, state parameters and scalar
mass, follow Jirka (2004), lead to the equations below:
𝑑𝑄
=𝐸 (5.1)
𝑑𝑠
𝑑𝑀𝑥
= 𝐸𝑢𝑎 + 𝐹𝐷 √1 − 𝑐𝑜𝑠 2 𝜃𝑐𝑜𝑠 2 𝜎 (5.2)
𝑑𝑠
𝑑𝑀𝑦 𝑐𝑜𝑠 2 𝜃𝑠𝑖𝑛𝜎𝑐𝑜𝑠𝜎

= −𝐹𝐷 (5.3)
𝑑𝑠 √1 − 𝑐𝑜𝑠 2 𝜃𝑐𝑜𝑠 2 𝜎
𝑑𝑀𝑧 𝑠𝑖𝑛𝜃𝑐𝑜𝑠𝜃𝑐𝑜𝑠𝜎
= 𝜋𝜆2 𝑏 2 𝑔𝑐′ − 𝐹𝐷 (5.4)
𝑑𝑠 √1 − 𝑐𝑜𝑠 2 𝜃𝑐𝑜𝑠 2 𝜎
Figure 5.1 Nearfield jet integral model definition sketch
Jet Sources
Where s is the axial distance along the jet trajectory and E is the rate of entrainment, and
b is the characteristic width of the jet, which is defined as the jet radius, where the jet
excess velocity is 𝑒 −1 = 37%. The centerline density is contained in the definition of
centerline buoyancy 𝑔𝑐′ and is calculated by the UNESCO equation of state, as function of
salinity (S) and temperature (T):
𝜌𝐽𝑒𝑡 = 𝜌𝑈𝑁𝐸𝑆𝐶𝑂 (𝑆, 𝑇) (5.5)
If sediments are present inside the jet, and their dynamics are included in the
calculations, then the jet density will be corrected for the presence of sediments. This can
be activated by defining the source in MT module and activating the MT-HD feedback.
𝜌𝐽𝑒𝑡 = (1 − 𝐶𝑠𝑒𝑑 )𝜌𝐽𝑒𝑡 + 𝐶𝑠𝑒𝑑 𝜌𝑠𝑒𝑑 (5.6)
𝐶𝑠𝑒𝑑 is the volumetric sediment concentration derived from the sediment concentration
provided by the user for the Jet source in MT module, and 𝜌𝑠𝑒𝑑 is the sediment density
provided by the user for the MT-HD feedback in MT module.
The buoyant acceleration is then defined as below, where 𝜌𝑎 is the ambient density, and
𝜌𝑟𝑒𝑓 is the reference density calculated by the reference salinity and temperatures
provided by the user in HD module.
𝜌𝐽𝑒𝑡 − 𝜌𝑎
𝑔𝑐′ = 𝑔 (5.7)
𝜌𝑟𝑒𝑓
The two important physical processes influencing the jet trajectory and dilution rates are
the entrainment rate (𝐸) and the ambient drag force (𝐹𝐷 ). The entrainment rate is
calculated as being proportional to the streamwise contribution of the jet centerline
velocity (𝑢𝑐 ) plus the azimuthal contribution from the transverse component of the
ambient velocity (𝑢𝑎 √1 − 𝑐𝑜𝑠 2 𝜃𝑐𝑜𝑠 2 𝜎).
𝑠𝑖𝑛𝜃 𝑢𝑎 𝑐𝑜𝑠𝜃𝑐𝑜𝑠𝜎
𝐸 = 2𝜋𝑏𝑢𝑐 (𝛼1 + 𝛼2 + 𝛼3 )
𝐹𝑙2 𝑢𝑐 + 𝑢𝑎
(5.8)
+ 2𝜋𝑏𝑢𝑎 √1 − 𝑐𝑜𝑠 2 𝜃𝑐𝑜𝑠 2 𝜎𝛼4 |𝑐𝑜𝑠𝜃𝑐𝑜𝑠𝜎|
𝐹𝑙 is the local densimetric Froude number, and is defined as:
𝑢𝑐
𝐹𝑙 = (5.9)
√𝑔𝑐′ 𝑏
The first term in the streamwise part of the entrainment function represents the “pure jet”
effects, the second term adds the effect of “pure plume” and the third term is for “pure
wake”. The four coefficients defining the entrainment rate are given the empirical values
suggested by Jirka (2004):
𝛼1 = 0.055⁡⁡⁡,⁡⁡⁡𝛼2 = 0.6⁡⁡⁡,⁡⁡⁡𝛼3 = 0.055⁡⁡⁡,⁡⁡⁡𝛼4 = 0.5

(5.10)
Deflection of the jet is a consequence of the pressure drag exerted on it by the cross flow
(𝐹𝐷 ) and of the entrainment by the jet of laterally moving fluid from the crossflow ( 𝐸𝑢𝑎 ).
The drag force is parametrized as a quadratic law force mechanism (Jirka, 2004):
1
𝐹𝐷 = 𝐶 2√2𝑏𝑢𝑎2 (1 − 𝑐𝑜𝑠 2 𝜃𝑐𝑜𝑠 2 𝜎) (5.11)
2 𝐷
Jet Sources
The jet diameter is calculated as 2√2𝑏 and 𝐶𝐷 is the drag coefficient as function of
velocity ratio between jet and the ambient, following Chan et al. (1976).
Calculations of the jet trajectory are discretized based on the incremental distance along
the jet trajectory (𝑑𝑠). Following a recommendation from Lee and Cheung (1990), the
spatial discretization of jet trajectory is calculated as below:
𝑑𝑠 = 𝑑𝑡(𝑢𝑐 + 𝑢𝑎 𝑐𝑜𝑠𝜃𝑐𝑜𝑠𝜎)
(5.12)
Where
0.1𝐷
𝑑𝑡 = ⁡
𝑢𝑐
However, the value of 𝑑𝑡 is set to have cut-off values of 0.001 seconds and 1.0 seconds.
𝐷 is the initial jet diameter.
This jet (nearfield) model calculates the jet trajectory and dilution until it reaches the end
of nearfield region. This is done at each HD time-step in the background flow model.
Although the HD time-steps can be much smaller than the time it takes for the trajectory
to reach the end of nearfield region, it is assumed that the temporal variations in the
background flow (ambient) are slower than the time it takes for the jet to go from
discharge point to the point of farfield release.
5.2 End of Nearfield region

In general, where the jet loses its driving characteristics over the ambient flow
(momentum and buoyancy), it has reached the end of its nearfield region and its volume
and scalar mass can be transferred and dispersed by the ambient flow into the Farfield
region. This can happen under different circumstances:
• Jet in cross-flow: The jet momentum M is combination of its initial momentum at the
diffuser, buoyancy and the ambient flow induced (co- or opposing) momentum Ma . It
loses its driving characteristics over the ambient flow when the excess momentum
becomes small, and close to the ambient flow momentum. This can be considered
as the end of the nearfield region and the release into the Farfield model by following
the condition: M − Ma < ε. Ma , where epsilon ε is left as a user-defined/calibration
parameter, with default value of 1%.
Jet Sources
Figure 5.2 Illustration of the jet velocity field and the contribution from the ambient flow
(Modified image from Jirka, 2004)
• Jet in stagnant environment: Under stagnant conditions, the contribution from the
ambient currents to the jet momentum is zero, and the Nearfield region can extend
until the point where the jet loses its own momentum due to dilution and buoyancy.
Considering the modeling/numerical limitations, a minimum value for the jet excess
velocity can be defined (gamma γ) to mark the end of Nearfield region and the
release into Farfield model. The default value for gamma is set to 1 cm/s.
• Jet in stratified stagnant environment: Non-horizontal jets (positively buoyant jets

pointed downward or negatively buoyant jets pointed upward) in stagnant
stratification create a complex situation. Depending on the stratification gradient and
the jet initial momentum, the jet can either be trapped in a layer where its density
equals the ambient density, or it overshoots and experiences a reversed buoyancy.
In latter case, the jet experiences a lateral collapse in form of an internal density
current formation in opposite direction and it ends up trapped in a terminal density
level. These complex processes are not included in the integral jet model formulation
and – in the absence of an adequate transition cut-off- it predicts (unrealistically) an
infinite number of oscillations about the terminal level. Therefore, the second
buoyancy reversal in the jet calculations is considered as the end of Nearfield
calculations and the release into the Farfield model.
• Jet in strong opposing flow: Jet integral models cannot be expected to hold for
flow situations in which boundary layer behavior is no longer maintained. The
boundary layer approximation implies a pressure within the jet equal to that in the
outside ambient. This is violated whenever the jet exhibits strong curvature such as
going into strong opposing ambient current. Therefore, the jet nearfield solution
stops and releases into Farfield model as soon as it experiences a strong opposing
flow.
The other criterion that ends the nearfield calculations is when the jet reaches the bottom,
surface or a lateral boundary. The dynamics of the jet approaching a solid boundary or
water surface are not yet included in the nearfield calculations of the MIKE jet module.
Among the impacts are variations in entrainment rates at the vicinity of the boundary. The
nearfield calculations continue un-influenced until the jet reaches the boundary, and there
it releases into the Farfield model.
Jet Sources
5.3 Nearfield-Farfield model coupling

The coupling between the Nearfield and Farfield model concerns both the reading of
ambient conditions as an input for the Nearfield jet model, and the release of jet
discharge into the Farfield flow model at the end of the Nearfield region.
The ambient flow conditions can be determined either as the local flow conditions at the
jet location or as the upstream ambient flow conditions. The upstream option can be used
to avoid unrealistic feedback between the jet solution and the ambient flow in cases with
dominant advection effects on the released material from the ambient flow. For the
upstream ambient flow condition, the conditions are obtained at a point defined by
distance from the jet location in the upstream flow direction. The distance is the maximum
of the characteristic length determined from the mesh and a user-specified minimum
upstream distance. The characteristic length is here determined as 2.3 times the square
root of the local element area at the initial release point.
At the end of Nearfield region (determined by any of conditions described in previous

section), the final jet discharge (which is diluted) is released into the Farfield flow model
at the corresponding point in space. This will assure that the right volume with right
dilution is released into the Farfield model at the right location. The discharge is
distributed into several sources over a plane area corresponding to the final jet diameter
and with a Gaussian distribution for volume and other scalars. The number of sources
depend on the mesh resolution in the Farfield model.
The increased jet discharge at the end of the Nearfield region (due to entrainment) is
inserted into the Farfield flow model at each Hydrodynamic time-step, which then impacts
the hydrodynamic solution at the next time step. The shallow-water equations being
solved in the Farfield flow model may not be able to correctly handle the insertion and
acceleration of such relatively large volume inserted into the domain (this depends as
well on local mesh resolution and water depth). The resulting flow field therefore might
not appear realistic. As a partial remedy, and to help the solution, at each of the release
point sources there will be added a forcing (momentum flux) to the momentum equation,
in the direction of jet release into the ambient domain, calculated as:
𝐹𝑥 = 𝑀 cos 𝜃 cos 𝜎, 𝐹𝑦 = 𝑀 cos 𝜃 sin 𝜎 (5.13)
The increased jet discharge and (consequently) its dilution at the release point is a result
of entrainment along its trajectory. Conservation of mass and volume in the Farfield flow
model then requires removing this excess mass and volume that have been inserted at
the release point. This has been done by introducing entrainment sinks along the
centerline of the jet trajectory (see Figure 5.3). The number of sinks depend on the mesh
resolution in the Farfield model. This method follows the Distributed Entrainment Sink
Approach (DESA) proposed by Choi and Lee (2007).
Jet Sources
Figure 5.3 Illustration of positioning of entrainment sinks along the trajectory in a 3D domain,
and the distributed source points at the release location
Similar to the problem at the release point, at the sink locations, subtraction of volume
inside the domain may result in dubious flow fields near the sinks (depending on local
mesh resolution). Therefore, following the same reasoning used for the release point, at
each sink point, a forcing (momentum flux), calculated as the product of sink rate and the
ambient flow velocity, is added to the momentum balance with an opposing direction.
This cannot be effective for the vertical velocities induced by the sinks, where their impact
becomes more visible in vertical jets.
(5.14)
𝑓𝑥 = −𝑞𝑠𝑖𝑛𝑘 𝑢𝑎_𝑥 , 𝑓𝑦 = −𝑞𝑠𝑖𝑛𝑘 𝑢𝑎_𝑦
Validation
6 Validation
The new finite-volume model has been successfully tested in a number of basic, idealised
situations for which computed results can be compared with analytical solutions or
information from the literature. The model has also been applied and tested in more
natural geophysical conditions; ocean scale, inner shelves, estuaries, lakes and overland,
which are more realistic and complicated than academic and laboratory tests. A detailed
validation report is under preparation.
This chapter presents a comparison between numerical model results and laboratory
measurements for a dam-break flow in an L-shaped channel.
Additional information on model validation and applications can be found here:
http://www.mikepoweredbydhi.com/download/product-documentation
6.1 Dam-break Flow through Sharp Bend

The physical model to be studied combines a square-shaped upstream reservoir and an
L-shaped channel. The flow will be essentially two-dimensional in the reservoir and at the
angle between the two reaches of the L-shaped channel. However, there are numerical
and experimental evidences that the flow will be mostly unidimensional in both rectilinear
reaches. Two characteristics or the dam-break flow are of special interest, namely
• The "damping effect" of the corner

• The upstream-moving hydraulic jump which forms at the corner
The multiple reflections of the expansion wave in the reservoir will also offer an
opportunity to test the 2D capabilities of the numerical models. As the flow in the reservoir
will remain subcritical with relatively small-amplitude waves, computations could be
checked for excessive numerical dissipation.
6.1.1 Physical experiments
A comprehensive experimental study of a dam-break flow in a channel with a 90 bend

has been reported by Frazão and Zech (2002, 1999a, 1999b). The channel is made of a
3.92 and a 2.92 metre long and 0.495 metre wide rectilinear reaches connected at right
angle by a 0.495 x 0.495 m square element. The channel slope is equal to zero. A
guillotine-type gate connects this L-shaped channel to a 2.44 x 2.39 m (nearly) square
reservoir. The reservoir bottom level is 33 cm lower that the channel bed level. At the
downstream boundary a chute is placed. See the enclosed figure for details.
Frazão and Zech performed measurements for both dry bed and wet bed condition. Here
comparisons are made for the case where the water in the reservoir is initially at rest, with
the free surface 20 cm above the channel bed level, i.e. the water depth in the reservoir is
53 cm. The channel bed is initially dry. The Manning coefficients evaluated through
steady-state flow experimentation are 0.0095 and 0.0195 s/m 1/3, respectively, for the bed
and the walls of the channel.
Validation
The water level was measured at six gauging points. The locations of the gauges are
shown in Figure 6.1 and the coordinates are listed in Table 6.1.
Figure 6.1 Set-up of the experiment by Frazão and Zech (2002)
Table 6.1 Location of the gauging points
Location x (m) y (m)
T1 1.19 1.20
T2 2.74 0.69
T3 4.24 0.69
T4 5.74 0.69
T5 6.56 1.51
T6 6.56 3.01
6.1.2 Numerical experiments
Simulations are performed using both the two-dimensional and the three-dimensional
shallow water equations.
An unstructured mesh is used containing 18311 triangular elements and 9537 nodes. The
minimum edge length is 0.01906 m and the maximum edge length is 0.06125 m. In the
3D simulation 10 layers is used for the vertical discretization. The time step is 0.002 s. At
the downstream boundary, a free outfall (absorbing) boundary condition is applied. The
wetting depth, flooding depth and drying depth are 0.002 m, 0.001 m and 0.0001 m,
respectively.
A constant Manning coefficient of 105.26 m 1/3/s is applied in the 2D simulations, while a

constant roughness height of 510-5 m is applied in the 3D simulation.
Validation
6.1.3 Results
In Figure 6.2 time series of calculated surface elevations at the six gauges locations are
compared to the measurements. In Figure 6.3 contour plots of the surface elevations are
shown at T = 1.6, 3.2 and 4.8 s (two-dimensional simulation).
In Figure 6.4 a vector plot and contour plots of the current speed at a vertical profile along
the centre line (from (x,y)=(5.7, 0.69) to (x,y)=(6.4, 0.69)) at T = 6.4 s is shown.
Figure 6.2 Time evolution of the water level at the six gauge locations. (blue) 3D calculation,
(black) 2D calculation and (red) Measurements by Frazão and Zech (1999a,b)
Validation
Figure 6.3 Contour plots of the surface elevation at T = 1.6 s (top), T = 3.2 s (middle) and T =
4.8 s (bottom).
Validation
Figure 6.4 Vector plot and contour plots of the current speed at a vertical profile along the centre
line at T = 6.4 s
6.2 Jet Source

Fan (1967) did comprehensive series of laboratory tests, where a negatively buoyant jet
is discharged vertically down into a crossflow and into a stagnant stratified water tank.
Figure 6.5 shows a photograph of the negatively buoyant jet in crossflow test 40-8-D,
which is used to validate the MIKE jet model.
Figure 6.5 Photograph of test 40-8-D (Fan, 1967)
In Figure 6.6, the jet centreline trajectory (solid blue line) and the corresponding
characteristic width (dash blue line) calculated by the MIKE integral jet model are plotted
upon the lab observations reported by Fan (the background image) of test number 40-8D.
The trajectory and the general width of the jet follows the observations very well. The
Farfield results of the same simulation are shown in Figure 6.7, where the Nearfield
calculations also are indicated.
Validation
Figure 6.6 Comparison of observed (Fan, 1967) and model results of nearfield jet trajectory
Figure 6.7 Presentation of the Nearfield calculations over the Farfield flow model MIKE 3 FM
results
Besides the jet-in-crossflow tests, Fan did also tests where the negatively buoyant jet is
discharged into a stagnant stratified environment. These tests demonstrated the
complicated process of jet overshooting its neutral density level and experiencing a
reversed buoyancy and finally being trapped in a new density level. The jet integral
equations do not resolve all the physical details of this phenomenon, but can estimate the
trajectory path and dilution rates well.
In Figure 6.8, the jet centreline trajectory (solid blue line) and the corresponding
characteristic width b (dash blue line) calculated by the integral jet model are plotted upon
the lab observations of Fan (stratified tank test 1) and the CorJet model results (single-jet
module in CORMIX) as the background image (from Jirka 2004). The trajectory and the
general width of the jet follows both the CoreJet results and the observations very well.
Due to ambient stratification and the jet momentum, the jet overshoots its neutral density
layer and experiences a buoyancy reversal (BR). The nearfield calculations has ceased
after the second BR.
Validation
Figure 6.8 Comparison of MIKE Nearfield model results with observed visual plume (Fan 1967)
and CorJet (Jirka 2004)
Jirka (2004) compared the CoreJet model results to few other test cases from Fan’s
laboratory experiment. The same cases have been simulated by MIKE jet module and the
results are plotted in Figure 6.9 upon the graphs presented by Jirka (2004). The general
agreement with both the measured values and the CoreJet model results are satisfactory.
Figure 6.9 Comparison of MIKE Jet module results with integral model predictions of Jirka
(2004) with experimental data of Fan (1967)
Validation
In jets that contain sediments with high levels of concentration, the weight of sediment
should be included in the Nearfield calculations. By activating the HD/MT coupling
feature, the sediment sources defined in the MT module will be taken into account when
calculating the jet density in the Nearfield calculations.
Decrop et al. (2013) did a series of laboratory measurements of sediment mixture jets in
cross flow. In Figure 6.10 a photograph of the laboratory experiment (with strong cross-
flow) is shown. In Figure 6.11 and Figure 6.12 the results of MIKE jet module is compared
with Decrop’s measurements as well as Fisher (1979) and the Lagrangian model of Lee
and Chu (2003).
Figure 6.10 Image of the negatively buoyant sediment plume in crosss flow (Decrop, 2013)
Figure 6.11 Comparison of the MIKE Jet module results with the experimental data of Decrop
(2013) and other models
Validation
Figure 6.12 Comparison of the MIKE Jet module results with the experimental data of Decrop
(2013) and other models
References
7 References
/1/ Chan, D.T.L., Kennedy, J.F. and Lin, J.T., 1976. Entrainment and drag forces of
deflected jets. Journal of the Hydraulics Division, 102(5), pp.615-635.
/2/ Choi, K.W. and Lee, J.H., 2007. Distributed Entrainment Sink Approach
(DESA)-a New Method for Modelling Mixing and Transport in the Intermediate
Field.
/3/ Darwish M.S. and Moukalled F. (2003), TVD schemes for unstructured grids,
Int. J. of Heat and Mass Transfor, 46, 599-611)
/4/ Decrop, B., Mulder, T. De & Troch, P., 2013. Experimental investigation of
negatively buoyant sediment plumes resulting from Dredging operations.
CoastLab 2012, (3).
/5/ Fan, L.N., 1967. Turbulent buoyant jets into stratified or flowing ambient fluids.
/6/ Fischer, H.B., 1979. Mixing in Inland and Coastal Waters. Academic Press.
/7/ Fredsøe, J. (1984), Turbulent boundary layers in Combined Wave Current

Motion. J. Hydraulic Engineering, ASCE, Vol 110, No. HY8, pp. 1103-1120.
/8/ Geernaert G.L. and Plant W.L (1990), Surface Waves and fluxes, Volume 1 –
Current theory, Kluwer Academic Publishers, The Netherlands.
/9/ Hirsch, C. (1990). Numerical Computation of Internal and External Flows,

Volume 2: Computational Methods for Inviscid and Viscous Flows, Wiley.
/10/ Iqbal M. (1983). An Introduction to solar Radiation, Academic Press.
/11/ Jawahar P. and H. Kamath. (2000). A high-resolution procedure for Euler and
Navier-Stokes computations on unstructured grids, Journal Comp. Physics,
164, 165-203.
/12/ Jirka, G.H., 2004. Integral model for turbulent buoyant jets in unbounded
stratified flows. Part I: Single round jet. Environmental Fluid Mechanics, 4(1),
pp.1-56.
/13/ Jones, O., Zyserman, J.A. and Wu, Yushi (2014), Influence of Apparent
Roughness on Pipeline Design Conditions under Combined Waves and
Current, Proceedings of the ASME 2014 33rd International Conference on
Ocean, Offshore and Arctic Engineering.
/14/ Kantha and Clayson (2000). Small Scale Processes in Geophysical Fluid flows,
International Geophysics Series, Volume 67.
/15/ Lee, J.H. and Cheung, V.W., 1986. Inclined plane buoyant jet in stratified fluid.
Journal of Hydraulic Engineering, 112(7), pp.580-589.
/16/ Lee, J.H. and Cheung, V., 1990. Generalized Lagrangian model for buoyant jets
in current. Journal of Environmental Engineering, 116(6), pp.1085-1106.
/17/ Lee, J.H.W and Chu, V., 2003. Turbulent jets and Plumes: A Lagrangian
approach. Kluwer, pp378.
References
/18/ Lind & Falkenmark (1972), Hydrology: en inledning till vattenressursläran,

Studentlitteratur (in Swedish).
/19/ Munk, W., Anderson, E. (1948), Notes on the theory of the thermocline, Journal
of Marine Research, 7, 276-295.
/20/ Oddo P. and N. Pinardi (2007), Lateral open boundary conditions for nested
limited area models: A scale selective approach, Ocean Modelling 20 (2008)
134-156.
/21/ Pugh, D.T. (1987), Tides, surges and mean sea-level: a handbook for engineers
and scientists. Wiley, Chichester, 472pp
/22/ Rodi, W. (1984), Turbulence models and their applications in hydraulics, IAHR,
Delft, the Netherlands.
/23/ Rodi, W. (1980), Turbulence Models and Their Application in Hydraulics - A

State of the Art Review, Special IAHR Publication.
/24/ Roe, P. L. (1981), Approximate Riemann solvers, parameter vectors, and

difference-schemes, Journal of Computational Physics, 43, 357-372.
/25/ Sahlberg J. (1984). A hydrodynamic model for heat contents calculations on

lakes at the ice formation date, Document D4: 1984, Swedish council for
Building Research.
/26/ Shu C.W. (1997), Essentially Non-Oscillatory and Weighted Essenetially Non-
Oscillatory Schemes for Hyperbolic Conservation Laws, NASA/CR-97-206253,
ICASE Report No. 97-65, NASA Langley Research Center, pp. 83.
/27/ Sleigh, P.A., Gaskell, P.H., Bersins, M. and Wright, N.G. (1998), An
unstructured finite-volume algorithm for predicting flow in rivers and estuaries,
Computers & Fluids, Vol. 27, No. 4, 479-508.
/28/ Smagorinsky (1963), J. General Circulation Experiment with the Primitive

Equations, Monthly Weather Review, 91, No. 3, pp 99-164.
/29/ Soares Frazão, S. and Zech, Y. (2002), Dam-break in channel with 90 bend,
Journal of Hydraulic Engineering, ASCE, 2002, 128, No. 11, 956-968.
/30/ Soares Frazão, S. and Zech, Y. (1999a), Effects of a sharp bend on dam-break
flow, Proc., 28th IAHR Congress, Graz, Austria, Technical Univ. Graz, Graz,
Austria (CD-Rom).
/31/ Soares Frazão, S. and Zech, Y. (1999b), Dam-break flow through sharp bends
– Physical model and 2D Boltzmann model validation, Proc., CADAM Meeting
Wallingford, U.K., 2-3 March 1998, European Commission, Brussels, Belgium,
151-169.
/32/ UNESCO (1981), The practical salinity scale 1978 and the international
equation of state of seawater 1980, UNESCO technical papers in marine
science, 36, 1981.
/33/ Wu, Jin (1994), The sea surface is aerodynamically rough even under light
winds, Boundary layer Meteorology, 69, 149-158.
/34/ Wu, Jin (1980), Wind-stress Coefficients over sea surface and near neutral
conditions – A revisit, Journal of Physical. Oceanography, 10, 727-740.
References
/35/ Zhao, D.H., Shen, H.W., Tabios, G.Q., Tan, W.Y. and Lai, J.S. (1994), Finite-
volume two-dimensional unsteady-flow model for river basins, Journal of
Hydraulic Engineering, ASCE, 1994, 120, No. 7, 863-833.

MIKE 321 FM Scientific Doc PDF

Uploaded by

Copyright:

Available Formats

MIKE 321 FM Scientific Doc PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MIKE 321 FM Scientific Doc PDF

Uploaded by

Copyright:

Available Formats

MIKE 21 & MIKE 3 Flow Model FM

Hydrodynamic and Transport Module

mike_321_fm_scientific_doc.docx / ORS/SIS / 2017-10-03 - © DHI

COPYRIGHT This document refers to proprietary computer software, which is

2 Governing Equations ....................................................................................................... 2

4 Infiltration and Leakage ................................................................................................. 39

5 Jet Sources ..................................................................................................................... 42

The system is based on the numerical solution of the two/three-dimensional

2.1 3D Governing Equations in Cartesian Coordinates

2.1.1 Shallow water equations

The model is based on the solution of the three-dimensional incompressible Reynolds

The local continuity equation is written as

u u 2 vu wu  1 pa

v v 2 uv wv  1 pa

where A is the horizontal eddy viscosity.

The surface and bottom boundary condition for u, v and w are

Here the UNESCO equation of state is used (see UNESCO, 1981).

2.1.2 Transport equations for salt and temperature

The transports of temperature, T, and salinity, s, follow the general transport-diffusion

T uT vT wT   T  

s us vs ws   s 

2.1.3 Transport equation for a scalar quantity

The conservation equation for a scalar quantity is given by

C uC vC wC   C 

where D h is the horizontal diffusion coefficient.

2.1.4 Turbulence model

Vertical eddy viscosity

a  10 and b  0.5 are empirical constants.

k uk vk wk    k 

 u v w

with the Brunt-Väisälä frequency, N, defined by

The horizontal diffusion coefficients are given by Dh  A /  k and Dh  A /   ,

Table 2.1 Empirical constants in the k- model.

0 for Us  0 (2.33)

Horizontal eddy viscosity

A  cs2l 2 2Sij Sij (2.35)

where cs is a constant, l is a characteristic length and the deformation rate is given by

2.1.5 Governing equations in Cartesian and sigma coordinates

The equations are solved using a vertical -transformation

In this new coordinate system the governing equations are given as

h hu hv h

hu hu 2 hvu hu  h pa

hv huv hv 2 hv  h pa

hT huT hvT hT

hs hus hvs hs   Dv s 

hk huk hvk hk

h hu hv h

hC huC hvC hC   Dv C 

The modified vertical velocity is defined by

The modified vertical velocity is the velocity across a level of constant

2.2 3D Governing Equations in Spherical and Sigma Coordinates

where R is the radius of the earth.

h 1  hu hv cos   h

hu 1  hu 2 hvu cos   hu  u 

hv 1  huv hv 2 cos   hv  u 

hT 1  huT hvT cos   hT

hs 1  hus hvs cos   h s

hk 1  huk hvk cos   hk

h 1  hu hv cos   h