Presentation by Julien Groenenboom (Deltares) at the Seminar Models and decision-making in the wake of climate uncertainties, during the Deltares Software Days South-East Asia 2023. Wednesday, 22 February 2023, Singapore.

1. Julien Groenenboom
22 February 2023
3D hydrodynamic modelling
of the Hong Kong waters
using Delft3D Flexible Mesh

2. Content of this presentation
2/21
• Background of the study
• Model setup
• Model calibration
• Model validation
• Recent model improvements
• Conclusions
Victoria Harbour
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters

3. Background of this study
3/21
• Area description
• Deltares is involved in Hong Kong
modelling studies for more than 25 years
• Aim: setup of a 3D hydrodynamic model
that is suitable for water quality modelling
• Study with/for ERM Hong Kong and EPD
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
Hong Kong SAR

4. Model setup – Model coverage and grid generation
4/21
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
• Compared to previous model (shown in red), we applied an
extended model domain with the aim to:
− Improve the modelled residual currents and their variability
− Improve the modelled surge
• Courant grid approach
− The resolution of the network increases with decreasing
water depths
Grid resolution in
Hong Kong waters ≤ 300 m

5. Model setup – Model coverage and grid generation
• Fine resolution where needed (physics and area of interest)
− Previous model (in red): structured grid
− New model (in blue): unstructured grid
• Compared to previous grid, the resolution has increased substantially!
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
5/21

6. Model setup – Scenario modelling
• Grids adjustments can easily be made using Delft3D Flexible Mesh
− Possible future reclamations
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
6/21

7. Model setup – Grid generation – Convergence tests
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
7/21
• Curvilinear vs. squared cells
• Convergence tests to determine
required resolution
• Resolutions: 150, 75, 37.5 m
• Conclusion:
Discharges through cross-sections
converged after applying a resolution of
75 m
• Final grid
− Courant-grid approach
− Curvilinear at selected locations
− Resolution:
Hong Kong waters ≤ 300 m
Hong Kong coastal zones ≤ 75 m

8. Model setup – Bathymetry and boundary conditions
Bathymetry
• GEBCO
• Hong Kong local data
Open boundaries
• Water level (astronomical components)
− FES2012
− Inverse Barometer Correction (IBC)
• Salinity and temperature
− WOA2013
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
8/21

9. Model setup – Rivers and meteorological forcing
Discharge-points
• River discharges
− eartH2Observe
Meteorological forcing
• Composite heat-flux model
• ECMWF’s ERA5 dataset
− Spatially- and time-varying: Hourly interval, on a 0.25˚ by 0.25˚
resolution grid
− Wind speed (u- and v-direction), atmospheric pressure and
Charnock coefficient
− Dew point temperature, cloud coverage and air temperature (in
addition to the wind speed) are used as input for the heat-flux
model.
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
9/21

10. Model setup – Model characteristics
Simulation period
• Four-year period (first year is considered spin-up)
Computational grid
• About 110.000 computational cells
• Resolution varies from approx. 5 km at open boundaries
to about 75 m
• Curvilinear grid in selected area and rivers
• Vertical grid: 20 equidistant sigma-layers
Runtime
• 20 partitions (5 nodes with 4 partitions per node)
− Intel quad-core e3-1276 v3 processor (4 cores per node with 3.6 GHz per core)
• 3.1 days per simulation-year (= 12.3 minutes per simulation-day)
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
10/21

11. Model calibration
• Spatially uniform bottom roughness → tidal amplitudes
• Horizontal/vertical viscosity and diffusivity → horizontal spreading and vertical mixing of salinity and
temperature
− Using the Smagorinsky sub-grid parameterisation
• Solar annual constituent SA and the solar semiannual constituent SSA → seasonal variation in the
water level
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
11/21

12. Model validation – Water levels
• The tidal part of the total water level was derived by
performing a harmonic analysis
• “Surge” = “total water level” minus “tide”
• Good agreement between observed and modelled
tide and water levels
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
12/21
Cheung Chau

13. Model validation – Transport patterns
• According to literature and common knowledge, there is a residual offshore current …
− … to the southwest during the winter monsoon (dry season)
− … to the northeast during the summer monsoon (wet season)
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
13/21

14. Model validation – Transport patterns – Dry season
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
14/21
Zoomed-out Zoomed-in
Surfaace
Bottom
Surfaace

15. Model validation – Transport patterns – Wet season
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
15/21
Zoomed-out Zoomed-in
Surfaace
Bottom

16. Model validation – Timeseries of salinity
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
16/21

17. Model validation – Timeseries of temperature
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
17/21

18. Coastal upwelling
Coastal upwelling
• In the summer (wet season), the SW monsoon
causes coastal upwelling. As a result, colder
oceanic water is transported towards the Hong
Kong waters.
Recent in-house model improvements
• Boundary conditions: WOA2013 → CMEMS
• Enhanced meteorological forcing
• Z-sigma-layering
• Adjusted grid
• Numerical settings
• In next slides referred to as “Run A → Run B”
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
18/21

19. Sea surface temperature
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
19/21
In the summer (wet season), the SW
monsoon causes coastal upwelling
Run B shows an improvement of
modelled temperatures near the bed
Run A Run B

20. Timeseries of temperature
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
20/21
Run A | Run B
Run B shows an improvement of
modelled temperatures near the bed
Observed
Run A
Run B

21. Conclusions
• A 3D hydrodynamic model of the Hong Kong waters was generated, validated and now applied in
hydrodynamic (and water quality) studies
• Using Delft3D FM’s unstructured grid, we were able to cover a large model domain
− Resolution in area of interest and where needed
− Capture more physics in enlarged model domain
• The model is forced using available data from open global products (FES, ERA5, WOA2013, CMEMS)
• The model validation shows satisfactory results
• Over the last couple of years, the model approach has been successfully applied in other regions as
well
• Recent model updates result in an improvements of:
− The temperature ‘dip’ in summer/coastal upwelling is now (better) reproduced by the model
− There is an improvement of the modelled total water levels due to an improvement of both the tidal-
and surge-component.
3D
hydrodynamic
modelling
of
the
Hong
Kong
waters
21/21

22. Julien Groenenboom
22 February 2023
3D hydrodynamic modelling
of the Hong Kong waters
using Delft3D Flexible Mesh