The document summarizes a storm surge model developed for Singapore Strait and South China Sea using Delft3D FM. The model was set up with a flexible mesh covering the region with elements ranging from 250m to 100km. Benchmark storm surge events were modeled and compared to tide gauge data. The model results showed good agreement with observations. Future work will focus on improving model accuracy and expanding it to 3D to better represent regional circulation.
1. Storm Surge Model for
Singapore Strait and South
China Sea using Delft3D FM
Haihua Xu, Jeng Hei Chow
Lucas Yiew, Pavel Tkalich
Centre of Excellence for Autonomous and Remotely Operated Vessels (CEAOPS)
Delft3D User Days, 14 November 2022
2. Outline
● Background and Introduction
● Storm surge model set-up, benchmarks and results
● Circulation model development
● Conclusion/Future work
3. Background and Introduction
● Singapore
○ Urbanized city, dense population of 5.69 million, 728 km2.
○ Low-laying island country.
○ High-density economic from coastline
● Potentially vulnerable to coastal hazards: coastal floods
○ Sea level rise + Storm surges +spring tides
Singapore
Sungei Buloh Wetland Reserve flood during 25 Jan 2015
Sea
Affect the marina reservoir crest gate operation during heavy rain
Reservoir
4. Storm Surge in Singapore
(Tkalich et al., 2013)
• North East Monsoon
• Sea level increase
• ~5-25cm
• South West Monsoon
• Sea level decrease
• ~5-25cm
North East Monsoon SLA
South West Monsoon SLA
5. Model step up (Delft3D Flexible Mesh)
Open BC
(India Ocean)
Open BC
(Pacific Ocean)
Open BC
(ArafuraSea)
Domain and Open Boundaries
Riemann invariant: 0
• Storm Surge Model Domain
o East part of Indian Ocean, Bay of Bengal, Andaman
Sea
o Malacca Strait, Singapore Strait
o South China Sea, West Pacific Ocean
o Java Sea, Timor Sea
• Open BC
o India Ocean, Pacific Ocean, ArafuraSea
• Coastlines
o Nautical chart
o Shoreline / Coastline Resources (for away from
Singapore)
(https://www.ngdc.noaa.gov/mgg/shorelines/)
o Google earth satellite image (for Singapore Region)
• Bathymetry
o Nautical chart
o GEBCO: Global ocean & land terrain models (15 arc
-second):
6. Model step up (Surface-water Modelling System (SMS))
● Mesh distributions (1)
○ Element size : 250m ~ 100km
○ Total Elements: 280.8K
#Elements 280.8k
5km
10km
8. Model step up
● External wind force: 10m Wind data: ERA5 Wind, hourly, 0.25 degree, netcdf file format
● Bottom Friction
○ Bottom roughness of manning coefficient : 0.023 in the sea
● Time Step
○ Implicit variable time step CFL ~ 0.5
○ Average time step size 30s
● 30days storm surge simulation take 5hours with desktop i7 CPU 3.2G Hz
Wind Velocity magnitude from ERA5
9. Wind Shear Stress
● Wind force on the sea surface
○ 𝜌𝑎: density of air
○ 𝑢10:wind velocity at 10 m above the free surface
○ 𝐶𝑑:wind shear stress coefficient
● Wind drag coefficient type
○ Smith and Banke (1975): piecewise linearly varying drag coefficient
○ Charnock (1955): formulation (no breakpoints),
𝑈10
𝑈∗
=
1
𝑘
ln
𝑧10
𝑧0
𝑈10
𝑈∗
=
1
𝑘
ln
𝑔𝑧10
𝑏𝑈∗
2
𝐶𝑑 =
𝑈∗
2
𝑈10
2
𝑈∗: wind friction velocity
10. Benchmarked Storm surge events
● Wind drag coefficient type
○ Smith and Banke (1975):
■ Delft3D default value
൝
𝐶𝑑
𝐴
= 0.00063, 𝑈10
𝐴
= 0𝑚/𝑠
𝐶𝑑
𝐵
= 0.00723, 𝑈10
𝐵
= 100𝑚/𝑠
■ Salehi et al., 2018 ൝
𝐶𝑑
𝐴
= 0.0012, 𝑈10
𝐴
= 0𝑚/𝑠
𝐶𝑑
𝐵
= 0.004, 𝑈10
𝐵
= 30𝑚/𝑠
■ Tkalich et al., 2014 ൝
𝐶𝑑
𝐴
= 0.008, 𝑈10
𝐴
= 0𝑚/𝑠
𝐶𝑑
𝐵
= 0.0073, 𝑈10
𝐵
= 100𝑚/𝑠
○ Charnock (1955):
■ Muis et al., 2016, b=0.041
Time of
event
SLA Tanjong Pagar station
Highest 12-hour average
25-hour
average
22/12/1999 0.75 0.60 0.55
26/12/2001 0.35 0.30 0.25
23/12/2006 0.50 0.45 0.40
14/01/2009 0.50 0.45 0.35
11. Comparisons of SLA at Tanjong Pagar tidal station
• Smith and Banke (1975):
• Delft3D default value
• Salehi et al., 2018
• Tkalich et al., 2014
• Charnock (1955):
• Muis et al., 2016
Tanjong Pagar
Tidal Station
12. Comparisons of water level with Satellite image
Satellite observations 23/Dec/2016
(no big difference for other days)
14. 14
Bottom roughness Calibration
Bottom Roughness Energy dissipation:
Internal Tide Friction :
:
Wang et al, 2021: https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2020JC016917
Bottom roughness
0.15
0.10
0.026
Other area
Manning formula
15. 15
Parallel speed up in HPC
Item Time of event Year
1 22/12/1999 1999-Jan to 2000-Jan
2 26/12/2001 2001-Jan to 2002-Jan
3 23/12/2006 2006-Jan to 2006-Jan
4 14/01/2009 2008-Jan to 2009-Feb
• HPC OS: Ubuntu
• Software: Delft3D FM 2022.02
Singularity Container
• TCOMS HPC AMD EPYC 7742 64-Core
Processor 2.3G
• Elements: 280.8k
• Internal Noes: 298.4k
• Internal Link: 442.8k
• Optimal elements/core : ~ 9k ;
• 1year event: 24hours (32 cores).
Simulation periods
Parallel Speed Up
0
10
20
30
40
50
60
70
0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64
Speed
up
Cores
Ideal
Delft3D FM
32 cores
16. 16
Tanjong Pagar SLA 01-Feb to 30 Dec-2011
Tanjong Pagar SLA 01-Feb to 30 Dec-2016
Tanjong Pagar Tidal Station
SLA Comparisons
17. 17
Time series comparison
Tanjong Pagar SLA 15-Nov to 30-Dec 1999 Tanjong Pagar SLA 15-Nov to 30-Dec 2001
Tanjong Pagar SLA 15-Nov to 30-Dec 2006 Tanjong Pagar SLA 01-Jan to 30-Jan 2009
Tanjong Pagar Tidal Station
20. Conclusion/Next Step
● Conclusion:
○ A 2D storm surge model is developed using Delft3D Flexible Mesh
○ The tide constitutions are considered to model the circulation in the region.
● Future work:
○ Further improve the accuracy of the model: bottom friction, Inverse Barometer Correction (IBV)
○ Develop the wave model to consider the wave-current interaction …..
○ Extended the model to 3D model the to consider the 3D circulation, temperature and salinity
changes.
21. - National Research and Development Centre setup by the Agency for
Science, Technology and Research (A*STAR) and the National University of
Singapore (NUS) to advance Singapore’s position in the M&O Industry.
- Integrating research and industry expertise to co-create and co-develop
innovative concepts and next-generation systems.
- A core feature of TCOMS is the next-generation Deepwater Ocean Basin
research facility which is equipped with advanced wave and current
generation systems to simulate the physical ocean environment and
complex scenarios that Marine & Offshore platforms and ships operate in.
- TCOMS is also supported by the petascale supercomputing capabilities of
the National Supercomputing Centre (NSCC) Singapore. Such capabilities
allow our researchers to better understand complex marine environments
and enhance the design and performance of their solutions.
About TCOMS