+ All Categories
Home > Science > DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Date post: 05-Jul-2015
Category:
Upload: delftsoftwaredays
View: 392 times
Download: 13 times
Share this document with a friend
Description:
SWAN Advanced Course
38
SWAN Advanced Course 2. Setting up a SWAN computation Delft Software Days 28 October 2014, Delft
Transcript
Page 1: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

SWAN Advanced Course2. Setting up a SWAN computation

Delft Software Days28 October 2014, Delft

Page 2: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Modelling aspects

Setting up a SWAN wave model

• Define the problem, is SWAN the proper tool?

• Define the area of interest (x and y limits) and resolution (dx, dy)

• Define frequency ( ) and direction ( ) limits and resolution (d , d )

• Define the time scales. Stationairy or non-stationairy?

• Define the boundary conditions

• Carry out sensitivity tests

• Calibration and verification: compare results with measurements

2

Page 3: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

!** model version swan-dcsm-j13-v1!*************************HEADING************************PROJ 'SWAN-DCSM' '001'!********************MODEL INPUT*************************SET NAUTICALSET LEVEL=0MODE NONSTCOORDINATES SPHERICAL CCMCGRID REGULAR -12. 48. 0. 21. 16. 420 480 CIRCLE 45 0.03 0.6INPGRID BOTTOM REGULAR -15. 43. 0. 560 420 0.050 0.050 EXC -9999.READINP BOTTOM 1. '../INP/exc03.BOT' idla=3 FREEINPGRID wind REGULAR -12 48 0.0 210 240 0.10000 0.06667 EXC -99. NONSTAT 20110905.0000 60 MIN 20110907.0300READINP wind 1. SERIES '../fews_wind.inp' idla=3 FREE!************************************* BOUNDARY CONDITIONS **************************************BOUND NEST '../INP/SPECTRA.BND' OPENINIT HOTSTART SING '../INP/DCSMA00b_hot_2011090612'!****************************************** PHYSICS *********************************************GEN3 KOMENWCAP KOMEN delta=0FRIC JONSWAP 0.038BREA CONST 1.0 0.73!************************************ NUMERICS ***************************************PROP BSBTNUM ACCUR npnts=98 NONSTAT mxitns=20!************************************ OUTPUT ***************************************POINTS 'P1' FILE '../INP/POINTS.PNT'BLOCK 'COMPGRID' NOHEAD 'SWAN.MAT' LAYOUT 3 XP YP HSIG HSWELL TMM10 TPS DIR DSPR WATLEV BOTLEV OUT 20110906.1200 2 HRSPECOUT 'P1' SPEC1D ABS 'SPEC_P1.SP1' OUT 20110906.1200 2 HRTABLE 'P1' HEADER 'POINTS.TAB' TIME XP YP DEP HSIG HSWELL TMM10 TM02 TPS DIR DSPR WIND WATLEV OUT 20110906.1200 1 HRTEST 1 0 POINTS XY 3 52 2.5 52 PAR 'TEST.PAR'COMPUTE NONSTAT 20110906.1200 60 MIN 20110906.2000STOP

general

grids

numerics

boundary conditions & initial state

physics

output

compute

E

N

E

Nnautical cartesian

3

Page 4: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Modelling aspects

4

Page 5: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

SWAN grids

• Computational grids:• Geographical space (x,y; cartesian or spherical):

>rectangular>curvilinear>unstructured

• Spectral space ( , ):> logarithmic (frequencies; flow, fhigh, msc)>regular (directions; circle or sector; mdc)

• Resolutions ( x, y, , )

• Resolve relevant spatial and temporal details• Resolution bottom grid ~ resolution computational grid

5

Page 6: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Modelling aspects

6

Page 7: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Modelling aspects

CGRID REGULAR -12. 48. 0. 21. 16. 420 480

Spherical coordinateslatitudelongitude

7

Page 8: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

3. SWAN model set up

Grid 1 (DCSM)RectangularArea:1500 km x 1700 kmCell size:3.6 km x 3.6 km

Grid 2 (ZUNO)CurvilinearArea:770 km x 750 kmCell size:200 m - 2 km x200 m - 2 km

8/19

Page 9: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

9

Page 10: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

SWAN grids: Resolution -space

low ~ 0.5 p

high ~ 3 p

E( )

prognostic part diagnostic part

-4low high

wind sea /swell

10

Page 11: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Swell 15 s -> 0.07 HzWind sea 3 s -> 0.33 Hz

CGRID REGULAR -12. 48. 0. 21. 16. 420 480 CIRCLE 45 0.03 0.6

11

Page 12: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

12

Page 13: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

13

Page 14: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

SWAN grids: Directional spreading

ms[-]

one sideddirectional

spreading [°]Type

1 37.5

4 24.9 wind sea

15 14.2

60 7.3

100 5.7 swell

800 2

( ) A cosmD

Directional distribution of incident wave energy D( )

14

Page 15: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

SWAN grids: Directional spreading

-0,05

0

0,05

0,1

0,15

0,2

0,25

0,3

0,35

0,4

-90 -80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90Direction

Swellm = 100

wind seam = 4

The value of is chosen based on the nature of the wave field• swell has little directional spread, so smaller bin sizes necessary• wind sea has large directional spread, so larger bin sizes possible

15

Page 16: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Side boundary effects

Directional spreading causes boundary effects: ‘shadow zone’Disturbed zone approximately equal to half-power width of directionalspread from upwave corner point

x-axis

y-axis

computational grid

dir.spread

16

Page 17: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Side boundary effects

Disturbed zone

See also exercise 05

17

Page 18: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Time scales

1000 km cdeep 1.56 T 22 m/sgroup velocity 0.5 * c 10 m/s 36 km/h20 – 30 hoursNon-stationairy

Wave boundary conditions every 6 hoursWind every 1 hourHydrodynamics every 1 hour

Tests with dt= 15 min / 30 min / 60 min gave similar resultsHalf the time step does not double the computational time

18

Page 19: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Wave boundary conditions• what data is available and how to schematize this

Can be obtained from:

• nesting (WAM, Wavewatch, SWAN)

• 1D / 2D spectra (measured or computed)

• parametric spectra in terms of Hm0 and Tp or Tm01

Beware of boundary value problems!

- for curvilinear grids use POINT and SPECOUT instead of NGRID andNESTOUT

1) SPECOUT 'PZ' SPEC2D ABS 'RES/SPEC_PZ.SP2' OUT 20120104.0000 1 HR

2) BOUND NEST 'RES/SPEC_PZ.SP2' OPEN

19

Page 20: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Boundary value problem

increase in wave height

strong initial decrease of mean wave period

20

Page 21: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Boundary value problem

growth above peak frequency

explains increase in waveheight and decay of meanperiod (NOT in peak period)

21

See also exercise 04b

Page 22: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

22

Page 23: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

23

Page 24: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Physics

- Wave growth by wind- White capping- Quadruplets- Bottom friction- Triads- Wave breaking- Diffraction- Obstacles

24

Page 25: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Numerics

- Propagation scheme (default 1st order upwind)

- Accuracy ACCUR - relative change in Hm0, Tm01- percentage wet grid points- maximum number of iterations

STOPC - absolute change in Hm0, Tm01- relative change in Hm0, Tm01- curvature in Hm0, Tm01

- Numerical schemes for refraction (DIRIMPL) and freq shift (SIGIMPL)default cdd and css=0.5;cdd=0 (central, no diffusion) is more accurate but less stablecdd=1 (1st order upwind) more diffusive, preferable with large gradients

25

Page 26: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Convergence criterium

Hs

DHSIGN

90%-conv. crit.

default 98%-conv. crit.

Hs

Example: 2003 experiment NCEX(Levi Gorrell)

26

Page 27: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Output

27

Page 28: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Additional SWAN output grids

• Types of sets of output points• Frame/Group: set of output location in box

• Curve: set of output locations along a curve

• Ray: set of output locations along depth/bottom contour

• Points: set of isolated output locations

• Ngrid: set of output locations for a nested grid

• Data file output• Table: integral wave parameters

• Specout: 1D/2D energy density spectra

• Nestout: 2D energy density spectra along boundary of nested grid

• Block: spatial distributions of integral wave parameters (also DHSIGN, DRTM01)

• Test: test output (output per iteration, spectral distribution of source terms, etc.)

28

Page 29: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

SWAN 1 Swan standard spectral file, version$ Data produced by SWAN version 40.91A$ Project: SWAN-DCSM ; run number: 001TIME time-dependent data

1 time coding optionLONLAT locations in longitude, latitude

2 number of locations3.00 52.002.50 52.00

QUANT9 number of quantities in table

Hsig Significant wave heightm unit

-9.000000 exception valueRTm01 Average relative wave periodsec unit

-9.000000 exception valueSwind wind source term (of var. dens.)m2/s unit

-9.000000 exception valueSwcap whitecapping dissipationm2/s unit-9.000000 exception value20110906.133000 date-time

0.4683E+01 0.6620E+01 0.6661E-03 -0.1137E-02 -0.1254E-04 0.0000E+00 0.0000E+00 0.0000E+00 0.4557E-030.4657E+01 0.6569E+01 0.6404E-03 -0.1155E-02 -0.9097E-05 0.0000E+00 0.0000E+00 0.0000E+00 0.4712E-03

20110906.150000 date-time0.4836E+01 0.7002E+01 0.6162E-03 -0.8178E-03 -0.1603E-04 0.0000E+00 0.0000E+00 0.0000E+00 0.3864E-030.4780E+01 0.6941E+01 0.6148E-03 -0.8210E-03 -0.1147E-04 0.0000E+00 0.0000E+00 0.0000E+00 0.4248E-03

20110906.163000 date-time0.4926E+01 0.7216E+01 0.5822E-03 -0.6814E-03 -0.1803E-04 0.0000E+00 0.0000E+00 0.0000E+00 0.3253E-030.4692E+01 0.7039E+01 0.4949E-03 -0.6359E-03 -0.1149E-04 0.0000E+00 0.0000E+00 0.0000E+00 0.3429E-03

20110906.180000 date-time0.4641E+01 0.7231E+01 0.4299E-03 -0.4395E-03 -0.1579E-04 0.0000E+00 0.0000E+00 0.0000E+00 0.2877E-030.4039E+01 0.6936E+01 0.2187E-03 -0.2861E-03 -0.7810E-05 0.0000E+00 0.0000E+00 0.0000E+00 0.1735E-03

20110906.193000 date-time0.4115E+01 0.7048E+01 0.2422E-03 -0.2502E-03 -0.1106E-04 0.0000E+00 0.0000E+00 0.0000E+00 0.1870E-030.3576E+01 0.6706E+01 0.1515E-03 -0.1639E-03 -0.5169E-05 0.0000E+00 0.0000E+00 0.0000E+00 0.1185E-03

For stationary runs the test output gives Hm0per iteration

For non stationary runs the test output does notinclude Hm0 per iteration, only per time step

29

Page 30: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Check convergence

30

Page 31: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Directions tend to converge more slowly due to weaker forcing

31

Page 32: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Sensitivity tests / calibration / validation

resolutiontime stepphysicsnumerics

Calibration: - Including realistic physics or fitting observations?for instance decrease bottom friction to compensatefor lack of proper non-linear interactions?

See also exercise 03: SWAN DCSM

32

Page 33: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

33

Page 34: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Accuracy of model results depends on:

• quality of model: input bottomcurrents / water levels / windswaves at boundary (Hs Tp, )

• physics (representation of)

• numerics

Modelling aspects

34

Page 35: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Modelling aspects

Setting up a wave model• Define the problem

• Which physical processes are important?• Is the model suitable to investigate the problem?

• Define the area of interest (x and y limits) and resolution (dx, dy)• What features (geographical or morphological) are important?• What are the dimensions of the feature of interest? Recommended to cover the

feature with at least 5 – 10 grid cells• At what location is data available to define boundary conditions?• Take care of boundary effects• Boundaries preferably parallel / perpendicular to contours

• Define frequency ( ) and direction ( ) limits and resolution (d , d )• What wave conditions are prevalent in the area?

• Define the boundary conditions• what data is available and how to schematize this• how many computations are needed / are feasible

• Calibration and verification: compare results with measurements• Sensitivity study numerics (model convergence, accuracy) and sensitivity study

physics (key parameters) to assess the uncertainties of the model predictions35

Page 36: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

Land-sea transitions

• Effects largest just after transition,negligible at longer distances;

• For extreme situations smallerdifferences effect of land-seatransitions can be neglected;

• For islands: dip in wind speeddecreases for very narrow islands

36

Page 37: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

1 km

Lake IJssel FL25

Hm0 obs = 0.32 mHm0 SWAN = 0.46 m

U10 = 18.3 m/s

Beware of land/water transition

37

Page 38: DSD-INT - SWAN Advanced Course - 02 - Setting up a SWAN computation

SWAN computations on flow grid

38


Recommended