DSD-INT 2017 Using XBeach to train a Bayesian network for coastal hazard predictions and Disaster Risk Reduction evaluation. A case (Ria Formosa) - Plomaritis
Presentation by Haris Plomaritis (University of Algarve-CIMA) at the XBeach X (10th Year Anniversary) Conference, during Delft Software Days - Edition 2017. Wednesday, 1 November 2017, Delft.
DSD-INT 2017 Delft3D FM hydrodynamic and morphological modelling, Waal River,...
Similaire à DSD-INT 2017 Using XBeach to train a Bayesian network for coastal hazard predictions and Disaster Risk Reduction evaluation. A case (Ria Formosa) - Plomaritis
Similaire à DSD-INT 2017 Using XBeach to train a Bayesian network for coastal hazard predictions and Disaster Risk Reduction evaluation. A case (Ria Formosa) - Plomaritis (20)
%in kaalfontein+277-882-255-28 abortion pills for sale in kaalfontein
DSD-INT 2017 Using XBeach to train a Bayesian network for coastal hazard predictions and Disaster Risk Reduction evaluation. A case (Ria Formosa) - Plomaritis
1. This project has received funding from the European Union’s Seventh Programme for Research, Technological
Development and Demostration under Grant Agreement No. 603458. This presentation reflects the views only of the
authors, and the European Union cannot be considered liable for any use that may be made of the information
contained therein.
Using XBeach to train a Bayesian network for coastal hazard
predictions and disaster risk reduction evaluation. A case
study (Ria Formosa).
XBEACHX Conference, Delft, November 2017
Theocharis Plomaritis, Oscar Ferreira and Susana Costas
Univeristy of Algarve, CIMA
2. Hotspot Tool
Main Objective: To surrogate the computationally expensive morphodynamic
simulations within an EWS with a BN for immediate evaluation of event impact.
Main steps and Advantages:
• Transform offshore physical parameters into onshore hazard (XBeach)
• Transform the hazard into impact for
selected receptors (Damage Curve)
• Incorporate into the tool DRR measures
so it can be used by coastal managers as a
DSS.
XBEACHX Conference, Delft November 2017
3. Study Site – Praia de Faro
Coastal lagoon system protected by a
barrier-island, with sandy shores and
dunes.
Storms Hs > 3m; up to 6-7m
Tidal range up to 3.5 m
Storm surge < 1 m
Reduced river input
XBEACHX Conference, Delft November 2017
Relevant Hazards
4. Car Park, Central Sector
(01/1996)
5. Car Park, Central Sector
(04/1990)
6. Car Park, Central Sector
(04/1990)
• Overwash
• Erosion
4. Bayesian Network (BN): Structure
Bound. Conditions Hazard
Receptors
Impact Measures
XBEACHX Conference, Delft November 2017
Relevant publications
Jäger, et al., Coastal Engineering (online)
Plomaritis et al., Coastal Engineering (online)
Sanuy et al., NHESS (under review)
Nodes Arcs
5. • Synthetic events
Hs vs. Tp
Hs vs. Surge
Hs vs. Duration
Poelhekke et al., 2016 Coastal Engineering
Number of
Events
Data Interval
Surge (m) Hs (m) Tp (s)
Historic 10 0.31 -0.58 4.5 - 7.1 9.6 - 13.5
Synthetic 30 0.15 – 0.66 3.0 - 7.9 8.1 - 17.4
BN training: Event Selection
XBEACHX Conference, Delft November 2017
6. Simulation for each tidal stage
3x30 =90 events
BN : Event Schematisation and Area Selection
XBEACHX Conference, Delft November 2017
Areas Selection: 4 areas based on exposure characteristics
Receptors Type
• Houses
• Infrastructure
7. BN: Model – Receptor - Impact
Praia de Faro:
• Overwash: Max water Flux
• Erosion: Vertical Erosion in 3 areas
XBEACHX Conference, Delft November 2017
Model – Receptors relation:
• Thiessen polygons
• Proxy selection
Hazard – Impact (Consequences) :
• Qualitative damage curve
8. DRR measures
Three DRR measures have been selected for the Praia de Faro case study
(Considering end-users input from interviews):
• Beach nourishment (T1) (hazard related, extra modelling)
• House removal (T2) (exposure related)
• Create and/or improve Communication Channels (T3) (vulnerability related)
XBEACHX Conference, Delft November 2017
9. Model Results
Observed Hazard threshold
Hs vs Tp: Good separation of hazard/no hazard event
XBEACHX Conference, Delft November 2017
10. BN Results
• For Hs> 7m and Tp> 14.5sec (Tr ≈ 50 years)
XBEACHX Conference, Delft November 2017
11. BN Results : Current Conditions
18
3
79
Potentially
Damaged
Damaged
Safe
• Tr ≈ 50 years
XBEACHX Conference, Delft November 2017
12. • Tr ≈ 50 years – Erosion
45
5
50
Potentially
Damaged
Damaged
Safe
BN Results : Current Conditions
XBEACHX Conference, Delft November 2017
15. BN Results: Stakeholder Contribution
XBEACHX Conference, Delft November 2017
Which is the importance of the beach state?
Extension of the BN to include post storm profile:
• Profile was schematised based on measured data
• Moderate storm was assumed:
Removal of berm and development of a low terrace
• 90 additional simulations
• Addition of an extra Boundary condition
node in the BN
16. BN Results: Stakeholder Contribution
• For Hs> 7m and Tp> 14.5sec (Tr ≈ 50 years) Area 3
XBEACHX Conference, Delft November 2017
17. BN Results : post-storm profile
• For Hs> 7m and Tp> 14.5sec (Tr ≈ 50 years) - Overwash
With Berm No Berm and Terrace
XBEACHX Conference, Delft November 2017
Hazard Impact Hazard Impact
18. Concluding remarks
• The use of a BN in EWS could result in significant reduction of the
operational computational effort.
• The inclusion of DRR measures makes the proposed tool useful also for long-
term decision making process.
• Nourishment is the most effective DRR in reducing the impact on the
houses.
• Incorporation of different initial topo-bathy could increase the BN
prediction capabilities
XBEACHX Conference, Delft November 2017