Presentation by Dano Roelvink (Deltares, IHE Delft) at the XBeach X (10th Year Anniversary) Conference, during Delft Software Days - Edition 2017. Wednesday, 1 November 2017, Delft.

1. 1
XBeach: Past, Present and Future
Dano Roelvink, IHE Delft, Deltares

2. 2
There was a time before XBeach...
• FIMP study, Long Island, NY
• Study carried out by Moffatt and Nichol and Delft Hydraulics for
NY District USACE
• Effect of breaching on water levels backbarrier modelled using
Delft3D
• Dune erosion and overwash not yet included properly
• Still, study showed reduced flood levels in back bays if actual
breach growth during flood was taken into account

3. 3
Shinnecock Inlet, 1938

4. 4
Cross-section Old Inlet
during 1938 storm
• Bloody thing would not
overtop!
• Dune had to be lowered
using SBeach :=(
• Missing:
Infragravity waves

5. 5
2004 hurricane season
Charley Frances Ivan Jeanne

6. 6
Motivation
• 2004 Hurricane season hit Florida coast 4 times
• Congress awarded multi-million project MORPHOS3D to
develop new physics-based model system to assess hurricane
impacts
• Scope: wind-surge-waves-nearshore processes-impacts
• Play ‘what-if?’ games around Corps of Engineers projects
Figure 1 Pre- and post hurricane Ivan, Perdido Key, Florida (source: USGS)

7. 7
Framework
• 3-year R&D Project granted through European Research Office
of ERDC to Delft consortium including UNESCO-IHE, Delft
Hydraulics and Delft University of Technology
• Objective:
‘the further development of MORPHOS3D through the inclusion
of modeling approaches recently developed by a consortium of
Delft institutes.’

8. 8
• no proprietary code used
• all developments public domain (GPL)
• easy to understand and transfer
• focus on physical processes rather than informatics, I/O
• emphasis on robustness
• focus on shallow water, swash, overwashing, breaching
• Swash and overwashing dominated by LF motions
Requirements development
environment
XBeach

9. 9
XBeach - approach
• open source code available for free on internet (xbeach.org)
• easy to use
• Short-wave averaged but long-wave resolving modeling of
waves, flow and morphology change in time-domain
• Swash and overwash motions
• Dune erosion, overwashing, breaching and full inundation
• Domain from outside surfzone to backbarrier
• Driven by boundary conditions from surge and spectral wave
models

10. 10
Principle sketch -
physics
20 s < T < 200 s
Ew(T)
u(T)
bound wave
motions dominated by
long waves not short
waves

11. 11

12. 12
Combined short/long wave
propagation and decay
Principle
test:
overwashing
by LF waves

13. 13
Institute For Water Education
UNESCO-IHE
WL | delft hydraulics
HF, LF and total Hrms HF, LF and total Urms
Measured (red) vs computed
mean transport
Initial, final measured and
computed profile

14. 14
Preparing
for a
reef bum
future...

15. 15
XBeach structure (Matlab)
while par.t<par.tstop;
% Wave boundary conditions
[par,s] = wave_bc (s,par,it);
% Flow boundary conditions
[s] = flow_bc (s,par,it);
% Wave timestep
[s] = wave_timestep (s,par);
% Flow timestep
[s,par] = flow_timestep (s,par);
% Suspended transport
[s]=transus(s,par);
% Bed level update
[s]=bed_update(s,par);
% Output
[it,s]=output(it,s,par);
end

16. 16
XBeach structure (Fortran 90/95)
do while(par%t<=par%tstop)
! Wave boundary conditions
call wave_bc (s,par,it);
! Flow boundary conditions
call flow_bc (s,par,it);
! Wave timestep
call wave_timestep(s,par)
! Flow timestep
call flow_timestep (s,par)
! Suspended transport
call transus(s,par)
! Bed level update
call bed_update(s,par)
! Output
call output(it,s,par)
enddo

17. 17
Code development and documentation styles
if (xmpi_isright .and. s%ny>0) then
! Fy(:,ny+1)=Fy(:,ny) ! only a dummy point in non mpi
s%Fx(:,s%ny+1)=s%Fx(:,s%ny) ! only a dummy point in non mpi
! ! where (Fy(:,ny+1)<0.d0) Fy(:,ny+1)=0.d0;
!Jaap + Bas: Don't do this before calling Dano :-)
endif
! cjaap: replaced par%hmin by par%eps
!s%hh = max(s%hh,par%eps)
! Robert: how is this mass conservative?
!where(wete2d==1)
! s%hh =max(s%hh,par%hmin)
! note: replace with new limiter?
!endwhere
if (xmpi_isright .and. s%ny>0) then
! Fy(:,ny+1)=Fy(:,ny) ! only a dummy point in non mpi
s%Fx(:,s%ny+1)=s%Fx(:,s%ny) ! only a dummy point in non mpi
! ! where (Fy(:,ny+1)<0.d0) Fy(:,ny+1)=0.d0;
!Jaap + Bas: Don't do this before calling Dano :-)
endif
! cjaap: replaced par%hmin by par%eps
!s%hh = max(s%hh,par%eps)
! Robert: how is this mass conservative?
!where(wete2d==1)
! s%hh =max(s%hh,par%hmin)
! note: replace with new limiter?
!endwhere

18. 18
Creative and meta-creative processes!
(c) Fedor Baart, 2008

19. 19
Testbank =>Testbed=>Skillbed
• Validation studies – 2DV lab
cases
• Controlled conditions
• Mostly large-scale tests
• Possibility to check
hydrodynamics, sediment
transport
• All test cases in automated
testbed
Boers_1C
Deltaflume2006_DP01
Deltaflume2006_DP02
Deltaflume2006_T01
Deltaflume2006_T01_zebra
Deltaflume2006_T02
Deltaflume2006_T03
Deltaflume2006_T04
DeltaflumeH298_T1
DeltaflumeH298_T2
DeltaflumeH298_T3
DeltaflumeLIP11D_1B
DeltaflumeLIP11D_1C
DeltaflumeLIP11D_2E
Deltaflume_T3
Lip11D_2E
.....
>60 cases

20. 20
Skillbed

21. 21
Overwash case: Santa Rosa Island
Hurricane Ivan: 16 September 2004
Start of
hurricane
Hurricane
landfall

22. 22
Black belt Matlab coding...
courtesy Dave Thompson, USGS

23. 23
Similar erosion and
deposition patterns:
• erosion of dune
face
• deposition on
island
Overwash case: Santa Rosa Island
McCall et al,
CE 2010

24. 24
18th month meeting – 23-27th Nov. 2009, Montpellier
MICORE.EU Field measurement sites in Europe

25. 25
Some MICORE results..
• Wide application, away
from developers, of
XBeach
• Good results for a
number of cases
• Acceptable but not very
accurate results for
intermediate beaches
• Gravel beaches require
more sophisticated
approach
Oostende
Praia de Faro

26. 26
Ap’s
baby...
• XBeach
just one
link in a
chain of
models
• From
offshore
hazards to
onshore
risk

27. 27
Meanwhile on the other side of the world...
13/01/2011 06:15
Hi,
my name is Jens Kruger, Oceanographer at the Secretariat of
the Pacific Communities (SPC), a regional organisation
serving 26 members in the Pacific region. I am currently
writing a research proposal in partnership with Susanne
Pohler, Lecturer in Marine Geology, University of the South
Pacific (USP), the primary tertiary institute for the Pacific
region; and also in association with Paul Kench, Associate
Professor, University of Auckland, NZ (hello again Paul!).
I am writing to invite you to go enter into a partnership with
us on this research proposal tentatively entitled “Changing
Waves and Coasts in the Pacific (WACOP Project)”, with
the objective to provide information on wave climate
variability and trends and consequent coastal vulnerability
for informed adaptation and energy security......

28. 28
Extreme sea levels at Pacific islands
Commonly cited cause:
tropical cyclones =
storm surge (+ tides)
Cyclone Mena - Rarotonga
Hydrodynamic modelling of wind- and pressure driven
storm surge from “synthetic” cyclones (Fiji)
1 in 100 year storm tide
McInnes, at al. (in review)
Global and Planetary
Change

29. 29
IG wave modelling at Ningaloo
 In collaboration with Univ. of
Western Australia, Ryan Lowe
 Recent field studies show
dominance of infragravity wave
motions
Research questions
 What is the skill of XBeach on a coral
reef?
 What effect do infragravity waves
have on particle motions and
currents?

30. 30
Instrument array and wave data (R. Lowe,
UWA)
swell energy decays rapidly, IG motions pick up and dominate
spectrum

31. 31
Model (red)-data (black) comparison
(swell)
rapid swell wave
height decay from
C1 to C3
swell variance on
reef well
predicted.
very high bed
friction dominates

32. 32
Model (red)-data (black) comparison (IG)
IG wave height
response well
predicted across the
reef.
IG wave heights are
larger than swell
wave heights.

33. 33
Example Bonriki

34. 34
Ongoing work (PhD Johan Reyns)
• Implementation/validation of XBeach (and Delft3D) sediment
transport and morphology algorithms in Delft3D-FM
• Application to complex coral reef environments
• Perspective: most of XBeach funtionality available in flexible-
mesh environment soon

35. 35
Going into less and more detail...

36. 36
Gravel beaches
• Wave-by-wave version of
Xbeach
• with groundwater
in/exfiltration
• coarse sediment
formulations
• PhD Robert McCall Univ.
Plymouth/Deltares

37. 37
Xbeach-G overtopping validation

38. 38
XBeach-G validation

39. 39
Runup and overtopping for steep profiles
• Use of nonhydrostatic XBeach validated against lots of lab data
Runup on complex profile Dike overtopping CLASH

40. 40
Bigger, better, faster....
• Improvement of MPI parallellization
• Two instead of one row/column overlap between domains ->
fewer variables to pass
• Much more accurate
• Reasonable scaling up to 128 processors
• Nonhydrostatic model and stationary wave model now also
work with MPI
• Pretty much the default option thanks to Willem Vermin

41. 41
2D XBeach tends (tended?) to
underestimate IG runup

42. 42
Multi-dir vs single-dir wave propagation
• Multi-dir (standard XBeach):
• Wave refraction resolved as propagation through theta-space
• Leads to excessive reduction of groupiness for directionally spread waves
• Single-dir:
• Solves energy balance along mean wave directions
y w f vx
c A D D DA c A c A
t x y

 
    
    
   
cos sing g w f vc A c A D D DA
t x y
 

   
   
  

43. 43
Propagation over uniform depth

44. 44
Single vs. multi-direction approach
+
Roughly 3 times faster
Wave group patterns less smoothed out
Likely higher IG waves in 2DH
May solve problems reported in Stockdon et al 2014
-
Will only work where stationary approach works, viz. not too
curved grids

45. 45
La Faute sur Mer

46. 46
Dune impact
Chris Ashley, 2017

47. 47
Going to longer timescales – coupling Delft3D
and XBeach
• Example development of Wilderness Breach after Hurricane
Sandy
• PhD Maarten van Ormondt (Deltares/IHE)
• Competing effects of tides wind, waves, infragravity waves
• Delft3D for moderate conditions, XBeach for storms

48. 48
New Inlet
Green: XBeach
• 3 year hindcast
• Important effect of
swash berm
schematization

49. 49
Building better berms
• Problem: XBeach does not like steep beaches as are often
present with relatively coarse sediments
• Example: Praia de Faro,
standard settings

50. 50
Approaches
• Improve non-hydrostatic, wave-resolving modeling of
processes in swash zone
• Breaker turbulence
• Sheet flow behaviour
• In/exfiltration
• Vertical velocities... Much fundamental work to do
• See presentation Chris Daley
• Nudge XBeach- surfbeat to given slope in swash zone
• Swash zone narrowly defined as H/h>1
• Add transport term depending on difference between actual and target
slope
• Works like a charm 

51. 51
• Without
bermslope
• Bermslope
0.2
Effect of ‘bermslope’

52. 52
Coupled modeling of beach & dune
evolutions (Roelvink & Costas)
• Processes
• Accretion due to cross-
shore transport and
longshore gradients
• Storm erosion
• Aeolian transport
and vegetation
growth
• Tsunami
• Models
• 1D XBeach stationary with
longshore gradient term
• 1D XBeach instationary
• 1D Duna model inspired by
Kroy, Herrmann, Durán a.o.
• 1D XBeach nonhydro

53. 53
XBeach-D Praia de Faro
See poster
Costas,
Bon de
Sousa &
Roelvink!

54. 54
Coupling with aeolian transport
• WindSurf model
• Collaboration with UNC, OSU, Marum, TUD, Deltares
See Bas
Hoonhout for
BMI – Python
version

55. 55
The Future...

56. 56